Analysing hydrological services provided by forests to support spatial planning and land management

Analysing hydrological services provided by forests to support spatial planning and land management Cláudia Maria Carvalho dos Santos Tese de Doutoramento apresentada à Faculdade de Ciências da Universidade do Porto Biodiversidade, Genética e Evolução 2014 Analysing hydrological services provided by forests to support spatial planning and land management Cláudia Maria Carvalho dos Santos PhD FCUP 2014 3.º CICLO D D D ! Analysing hydrological services provided by forests to support spatial planning and land management Cláudia Maria Carvalho dos Santos Doctoral Program in Biodiversity, Genetics and Evolution (BIODIV) Department of Biology 2014 Supervisor João José Pradinho Honrado Professor Auxiliar, Faculty of Sciences, University of Porto Portugal Co-supervisor Lars Hein Associate Professor, Environmental Systems Analysis, Wagenigen University The Netherlands 3 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Água a correr na fonte, Uma quimera líquida que sai Das entranhas do monte A saber ao mistério que lá vai… Pura, Branca, inodora e fria, Cai numa pedra dura E desfaz o mistério em melodia… Miguel Torga, Diário II, 1948 ! 4 ! 5 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Foreword ! According to the number 7 of Article 6 from the Regulation of the Doctoral Program in Biodiversity, Genetics and Evolution (BIODIV), Faculdade de Ciências da Universidade do Porto (and in agreement with the Portuguese Law Decree nº 74/2006), this thesis integrates the articles listed below, written in collaboration with co-authors. The candidate hereby declares that she contributed to conceiving the ideas, compiling and producing the databases and analysing the data, and also declares that she led the writing of all Chapters. List of papers: Chapter 2 - Carvalho-Santos C., Hein L., Honrado J. (2014) Hydrological services and the role of forests: Conceptualization and indicator-based analysis with an illustration at a regional scale. Ecological Complexity. 20: 69-80. DOI: 10.1016/j.ecocom.2014.09.001 Chapter 3 - Carvalho-Santos C., Marcos B., Espinha Marques J., Alcaraz-Segura D., Hein L., Pradinho Honrado H. (2013) Evaluation of hydrological ecosystem services through remote sensing. In Alcaraz Segura D, Di Bella C M; Straschnoy J V (eds), Earth Observation of Ecosystem Services, CRC Press Taylors and Francis group, Boca Raton, pp. 219-249. ISBN 9781466505889. Chapter 4 - Carvalho-Santos C., Nunes J.P., Monteiro A.T., Hein L, Honrado J. (under review) Simulating the effects of land cover and future climate conditions on the provision of hydrological services in a medium-sized watershed of Portugal. Chapter 5 - Carvalho-Santos C., Silva A.R., Gonçalves J., Honrado J. (submitted) From hydrological services to a multifunctional watershed: trade-offs and synergies between biodiversity conservation and forest ecosystem services. V ! 6 ! 3 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS This thesis was supported by the Portuguese Science Foundation (FCT) through the PhD grant SFRH/BD/66260/2009 Este trabalho foi realizado com o apoio da Fundação para a Ciência e Tecnologia (FCT), através da bolsa de doutoramento com a referência: SFRH/BD/66260/2009 (QREN-POPH, tipologia 4.1) UNIÃO EUROPEIA Fundo Social Europeu GOVERNO DA REPÚBLICA PORTUGUESA VII ! 4 ! ! ! 11 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Resumo ! A água é um recurso vital para todos os organismos vivos da Terra. Os seres humanos dependem dos ecossistemas para a prestação de serviços hidrológicos, tais como o abastecimento de água e a mitigação de danos causados pela água, que são essenciais para o seu bem-estar. No entanto, as preocupações com os problemas relacionados com a água têm vindo a aumentar nas últimas décadas, com especial ênfase para a escassez de água, a sua qualidade e os desastres causados por ela, incluindo as questões relacionadas com as alterações climáticas. Isso tem suscitado o interesse na gestão sustentável dos ecossistemas, em particular das florestas, para a prestação de serviços hidrológicos, tendo em conta o reconhecimento dos processos que suportam essa provisão. Por conseguinte, os serviços hidrológicos devem ser adequadamente conceptualizados, quantificados, mapeados e monitorizados. O principal objetivo de investigação desta tese foi desenvolver quadros teóricos e analíticos para avaliar os serviços dos ecossistemas hidrológicos prestados pelas florestas, apoiando assim as opções para o planeamento e gestão do território. Embora o foco geral desta tese seja o desenvolvimento de abordagens metodológicas, os resultados dessas novas abordagens tem também um potencial para o apoio à tomada de decisão sobre a gestão de bacias hidrográficas. Este potencial foi ilustrado para a bacia hidrográfica do rio Vez, no Noroeste de Portugal, onde a precipitação é alta, embora desigualmente distribuída ao longo do ano, com potencial para a ocorrência de episódios extremos relacionados com a água. A pesquisa desenvolvida nesta tese é apresentada ao longo de seis etapas sucessivas. Primeiro foi apresentado o estado da arte sobre os diferentes níveis de análise dos serviços hidrológicos, capítulo 1. Em segundo lugar, no capítulo 2, foi elaborado um quadro conceptual para a prestação de serviços hidrológicos com base nas relações entre a água e as florestas, e no contexto da análise de sistemas sócio-ecológicos. Este quadro conceptual foi ilustrado para o abastecimento de água e os serviços de controle da erosão do solo e aplicado a uma escala regional, o Norte de Portugal. Os resultados mostram que o quadro conceptual proposto é uma ferramenta útil de apoio ao ordenamento do território e à gestão florestal, adequando a prestação de serviços hidrológicos com as condições biofísicas e sociais de cada região. A aplicação do quadro sugere que a combinação espacialmente explícita de indicadores relacionados com a propriedade do sistema, as funções, o serviço e o benefício pode ser uma forma eficaz de análise e gestão da oferta e da procura por serviços hidrológicos. Posteriormente, no capítulo 3 é apresentada uma visão geral dos produtos de satélite que podem ser usados para avaliar e monitorizar a prestação de serviços hidrológicos, com base nos diferentes compartimentos de água existentes na Terra (atmosfera, criosfera, água de superfície, solo, aquífero e vegetação). A avaliação dos serviços de abastecimento de água e de mitigação de danos causados por ela pode beneficiar fortemente do uso de produtos de satélites, contribuindo para uma melhor compreensão dos processos e funções que estão na base da sua provisão, com elevada cobertura espacial e resolução temporal. Como a ilustração dos serviços hidrológicos apresentados no capítulo 2 foi baseada em modelação estática, foi testada a aplicação de uma estrutura de modelação dinâmica para avaliar o papel das florestas na prestação de serviços hidrológicos. Assim, no capítulo 4, o modelo hidrológico SWAT (Ferramenta de Avaliação de Água e Solo) foi aplicado na bacia hidrográfica do Vez para analisar a prestação de serviços hidrológicos em cenários de alteração do uso/cobertura do solo e as condições climáticas futuras. Os resultados das simulações relativamente aos cenários de cobertura do solo revelaram que a opção por um determinado cenário não comprometeria a prestação global dos serviços hidrológicos. No entanto, cada cenário pode ser adoptado para ! ! 12 maximizar a prestação de um determinado serviço, por exemplo florestas de carvalhos naturais podem ser promovidas para melhorar a libertação gradual de água no rio. O estudo mostra ainda que a alteração climática pode afetar a prestação dos serviços hidrológicos de duas maneiras, reduzindo os caudais na estação seca e aumentando os riscos de picos de caudal passiveis de criar inundações durante os meses de chuva. Os efeitos combinados de clima e alteração da cobertura de solo podem compensar os picos de caudal durante o inverno, mas agravar os baixos caudais de verão na presença de florestas. A erosão do solo e a concentração de nitratos no rio aumentarão em condições climáticas futuras, podendo ser agravadas no cenário de uso do solo agrícola. Estes resultados reforçam a necessidade de se considerar não só o clima, como também os impactos da mudança de uso do solo em opções de gestão que visam uma melhor adaptação às novas condições de provisão dos serviços relacionados com a água à escala de bacias hidrográficas. Com base nos resultados das simulações do SWAT desenvolvidas no capítulo anterior, no capítulo 5 foi desenvolvida uma avaliação espacial dos serviços hidrológicos, da produção de biomassa e do armazenamento de carbono, juntamente com a valoração da conservação da biodiversidade para analisar os conflitos e sinergias entre eles. Os resultados mostraram que o desempenho para o fornecimento da água (quantidade e sazonalidade) é melhor nos cenários de arbustos e de carvalhos, na sub-bacia de alta montanha. Enquanto o cenário de eucalipto e pinheiro tem maior desempenho para a prestação da regulação de picos de cheia e controle de erosão, especialmente na baixa montanha. Este último service é também favorável no cenário arbustivo. O melhor cenário para a conservação da biodiversidade é o de carvalho. Finalmente no capítulo 6 foram discutidas as contribuições metodológicas, os principais resultados e sua relevância para a gestão de políticas gestão dos usos do solo. A análise de serviços hidrológicos prestados pelas florestas é um tema que combina os princípios e métodos da eco-hidrologia com conceitos e métodos da ciência dos serviços de ecossistemas. As dimensões de análise abrangem não só a aquisição e aplicação dos fundamentos conceptuais de ambas as ciências, mas também uma descrição de formas de avaliação e monitorização para a prestação de serviços hidrológicos. Para além disso, deve também abranger o uso de ferramentas de modelação para ser capaz de entender e prever os processos hidrológicos e suas ligações com as funções ecológicas dos ecossistemas. As principais contribuições desta investigação foram: (1) o desenvolvimento de um quadro conceptual para a análise dos serviços hidrológicos prestados pelas florestas; (2) uma revisão exaustiva das muitas vantagens que a detecção remota oferece para a avaliação, gestão e monitorização de serviços hidrológicos; e (3) a aplicação do modelo SWAT a uma bacia de média dimensão para avaliar as consequências de diferentes opções de uso/cobertura do solo, bem como condições climáticas futuras, na prestação de serviços hidrológicos. Em conclusão, para uma correta avaliação dos processos hidrológicos que levam à prestação de serviços hidrológicos, o uso de modelos hidrológicos é uma abordagem eficaz, especialmente em associação com produtos de satélite para prever e monitorizar a condição das funções hidrológicas dos ecossistemas e paisagens. Palavras-chave: Alterações Climáticas; Bacia Hidrográfica do Vez; Conservação da Biodiversidade; Florestas; Indicadores; Ordenamento do território; Produtos de satellite; Quadro conceptual; Serviços hidrológicos; SWAT . ! ! 13 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Abstract ! Water is a vital resource for all living organisms on Earth. Humans rely on ecosystems for the provision of hydrological services, namely water supply and water damage mitigation, essential for their well-being. However, concerns over the water problems have been increasing in the last decades, with special emphasis for water scarcity, quality and disasters, including issues related to climate change. This has been raising interest in the sustainable management of ecosystems, in particular of forests, for the provision of hydrological services, taking into account the recognition of the processes behind that provision. As a result, hydrological services should therefore be adequately conceptualized, quantified, mapped and monitored. The main objective of the research underlying this thesis was to develop theoretical and analytical frameworks to assess hydrological ecosystem services provided by forests, thereby supporting options for spatial planning and land management. Although the general focus of this thesis is on the development of methodological approaches, the outcomes from those novel approaches are expected to have a potential to support decision-making on the watershed management and governance. This potential was illustrated for the Vez watershed, in northwest Portugal, where precipitation is high, although unevenly distributed throughout the year, with possible occurrence of water extreme episodes. The research developed for this thesis was addressed throughout six successive steps. First the current state of knowledge on the different levels of analysis of hydrological services, chapter 1, was presented. Secondly, in chapter 2, a conceptual framework for the provision of hydrological services based on the relations between water and forests was elaborated, in the context of social-ecological system analysis. This conceptual framework was illustrated for the water supply and soil erosion control services, applied at a regional scale for northern Portugal. Results show that the proposed conceptual framework is a useful tool to support land planning and forest management, adapting the provision of hydrological services to the regional biophysical and social conditions. The application of the framework across a heterogeneous region suggested that a spatially explicit combination of system property, function, service and benefit indicators can be an effective way of analysing and managing the supply and demand of hydrological services. Subsequently, in chapter 3, an overview of the satellite products that can be used to evaluate and monitor the provision of hydrological services, based on the different water compartments on Earth (atmosphere, cryosphere, surface water, soil, ground and vegetation) was presented. The assessment of water supply and water damage mitigation services can strongly benefit from the use of satellite-based products, contributing to improve the understanding of the processes and functions behind their provision, on a spatially explicit and near-real time basis. As the illustration of the hydrological services presented in chapter 2 was based on a static modelling approach, the application of a dynamical modelling structure to evaluate the role of forests on hydrological services provision was then tested. Therefore, in chapter 4, the SWAT (Soil and Water Assessment Tool) hydrological model was applied in the Vez watershed to analyse the provision of hydrological services under scenarios of land use/cover change and future climate conditions. Results for the simulations under land cover scenarios revealed that the option for one particular scenario would not compromise the overall provision of hydrological services. However, each scenario may be adopted to improve the provision of a given service, for instance natural oak forests can be promoted to improve the gradual release of water into the river network. The study showed that climate change could affect the provision of hydrological services in two ways, by reducing dry season flows and by increasing flood risks during wet months. Combined effects of climate and land use change can offset peak flows during winter and reduce low flows in summer in the presence of forests. Soil erosion and ! ! 14 nitrates concentration in the river will increase under future climate conditions and could be aggravated in the agricultural land use scenario. These results emphasise the need to consider both climate and land use/cover impacts in management options aimed to improve adaptation to changed conditions in the water services at the watershed scale. Based on the outputs from SWAT simulations developed in the previous chapter, in chapter 5 a spatial assessment of the hydrological services, biomass production and carbon storage provision, together with the biodiversity conservation value was developed to analyse trade-offs and synergies to support land planning and management at the watershed scale. Results showed that the performance for the provision of water quantity and timing is better under the shrubland and oak scenarios, in the high mountain sub-basin. While the eucalyptus/pine forest scenario has higher performance for the provision of flood regulation and erosion control, especially in the low mountain. Erosion control is also favoured under the shurbland scenario. The scenario with less trade-offs with biodiversity conservation is the oak. The methodological contributions, the main findings and their relevance for land management and policies were finally discussed in chapter 6. Analysing hydrological services provided by forests is a topic that combines principles and methods of eco-hydrology with concepts and methods from the novel ecosystem services science. The dimensions of the analysis encompass the acquisition and application of the conceptual foundations from both sciences, as well as a description of different ways for evaluating and monitoring the hydrological services provision. Moreover, it should also encompass the use of modelling tools to be able to understand (and predict) hydrological processes and their links to the ecological functions of ecosystems. The main contributions of this research were: (1) the development of a conceptual framework for the analysis of hydrological services provided by forests; (2) an exhaustive review of the many advantages that Earth observation offers for the evaluation, management and monitoring of hydrological services; and (3) the application of the SWAT model to a mediumsized watershed for the evaluation of the consequences of both land use/cover options and future climate conditions on the hydrological services provision. In conclusion, for a correct evaluation of the hydrological processes that lead to hydrological services provision, the use of daily-runoff models is an effective tool, especially in association with the powerful satellite products, to predict and monitor the condition of water functions in dynamic ecosystems and landscapes. Keywords: Biodiversity conservation; Climate change; Conceptual framework; Forests; Hydrological services; Indicators; Land management; Satellite products; SWAT; Vez Watershed. ! ! 15 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS List of Figures Chapter 1 Figure 1. 1–The several types of ecosystem services and their relation with species diversity, from Elmqvist, 2010. _________________________________________________________________ 24 Figure 1. 2 - Forest ecosystem services organized by major classes, from Shvidenko et al., 2005. 25 Figure 1. 3 – a) Average annual precipitation in Portugal (1961-90 average, from the SNIRH), and forest distribution in the country by dominant species (National Inventory of Forests, 2010). ______ 26 Figure 1. 4 – The world distribution of water. __________________________________________ 27 Figure 1. 5 – The water cycle, from Diop et al., 2002. ___________________________________ 27 Figure 1. 6 - Hydrologic responses to changes in forest disturbance and management, from Jones et al. 2009. _______________________________________________________________________ 29 Figure 1. 7 -The relationship between the ecohydrological processes and the provision of hydrological services, connected by attributes such as of quantity, quality, location and timing, from Brauman et al. 2007. _____________________________________________________________ 31 Figure 1. 8 – The movement of water from precipitation through the vegetation and soil system into streams, and how it is affected by changes in vegetation, from Le Maître, et al., 2014. __________ 31 Figure 1. 9 – Simple schematic flow of the steps required in ArcSWAT. _____________________ 34 Figure 1. 10 – Spectral characteristics of the electromagnetic spectrum, from Tuner et al., 2008. _ 37 Figure 1. 11 - Typical spectral reflectance curves for water, soil and vegetation, from de Jong et al., 2004. _________________________________________________________________________ 37 ! Chapter 2 ! Figure 2. 1 -!Conceptual framework for hydrological services provision by forests, showing the relationship between the biophysical ecosystem (properties and functions) and the social system (services and benefits). Inside boxes are some examples of each step of the framework. Adapted from the ecosystem services framework by de Groot et al. (2010b) and Haines-Young & Potschin (2010).!_____________________________________________________________________________________________________________!47! Figure 2. 2 – The Framework for Ecosystem Service Provision (FESP) applied to hydrological services provided by forests. DPSIR (Drivers, Pressures, State, Impacts and Responses). ESp – ecosystem services providers. ESb – ecosystem services beneficiaries. Modified from Rounsevell et al. (2010).!_________________________________________________________________________________________________________!55! Figure 2. 3 – The study area. Indicators and sources: a) Location - Digital elevation model (SRTM – 90 m resolution); b)!Mean annual precipitation (mm/yr, 250 m2 pixel resolution) from Worldclim database, 1950-2000 (Worldclim, 2010); c) Mean annual evapotranspiration (mm/yr, 1km2 pixel resolution) from MODIS, 2000-2010 (MODIS, 2010); d) Forest distribution (evergreen, broadleaved and mixed) – from CORINE Land Cover 2006 (EEA, 2006); and e) Simplified map of potential native forests - modified from Costa et al. (1998).!___________________________________________________________________!57! Figure 2. 4 - Illustration of the conceptual framework of hydrological services provided by forests using two examples: water supply (quantity) at the municipality level, and water damage mitigation (soil erosion control) at the sub-basin level.!__________________________________________________________________!62! Chapter 3 ! Figure 3. 1 - Framework for the provision of hydrological services, in the context of social–ecological systems. Based on Rounsevell, M. D. A., et al., (2010) and Haines-Young, R., and M. Potschin (2010).!_____________________________________________________________________________________________________________!67! Figure 3. 2- The water cycle in the vicinity of the land surface. Conceptual model based on Fitts, C. R., (2002).!_________________________________________________________________________________________________________!69! ! ! ! 16 Chapter 4 ! Figure 4. 1 – Location map of Vez watershed in northwest Portugal with 10 sub-basins (white limits), and the Digital Elevation Model of the region (shadow); a) Land cover map, 2006 (SIGN II, 2008); b) Soil map (Agroconsultores and Geometral, 1995).!__________________________________________________________!90! Figure 4. 2 – Daily observed and simulated discharge for calibration and validation period, after parameter calibration (table 4.4 and 4.5). Performance statistics are presented in Tables 4.8 and 4.9. !_____________________________________________________________________________________________________________________!99! Figure 4. 3 – Comparison between monthly observed and simulated values for sediments (t/ha.yr) and monthly nitrates (ton. NO3), after parameter calibration.!_____________________________________________!100! Figure 4. 4 – a) The water balance simulated by SWAT in the Vez watershed (monthly averages for the period 2003-2008); and b) average monthly discharge under different land cover scenarios (m3/s). !____________________________________________________________________________________________________________________!101! Figure 4. 5 – Flow duration curves for: (a) land cover scenarios; (b) and future climate conditions (logarithm scale). Those curves express the probability of exceeding a given streamflow.!___________!103! Figure 4. 6 – a) Average monthly discharge (m3/s) under future climate conditions; and b) percentage of change in monthly average discharge under future climate conditions combined with land cover scenario of increase eucalyptus/pine forest or increase agriculture.!_____________________________________!107! ! Chapter 5 Figure 5. 1 – The study area, Vez watershed, with the location of observed data stations used to setup SWAT model, the range of altitude and a climate plot (yearly average values 1999-2008).!___!115! Figure 5. 2 – a) Land cover map of Vez watershed (year 2006); b) Biodiversity protection levels in the Vez watershed.!__________________________________________________________________________________________________!115! Figure 5. 3 – Ecosystem services provision by the major sub-basins (% of the service in the whole watershed). The values were weighted by the area of each HRU and by the area of the basin.!_____!120! Figure 5. 4 – Spatial distribution of biodiversity conservation value in the Vez watershed; a) current pattern; b) future pattern under eucalyptus/pine scenario; c) future pattern under oak scenario.!____!121! Figure 5. 5 – Maps of the ecosystem services simulated in SWAT at the HRU level, under three land cover scenarios in the Vez watershed.!_______________________________________________________________________!122! ! Appendices Figure B. 1 - Average evapotranspiration (mm/yr) by different land-covers in Vez basin from MOD16A2 product and simulated by SWAT.!________________________________________________________________!171! Figure B. 2 – Average evapotranspiration in Minho region, by pure 1km2 cells ( ≥ 75% occupied by a single land cover type).!_________________________________________________________________________________________!171! ! Figure C. 1 – Box plots representing the median value of the distribution in each scenario. (Shrub n = 500; Oak and EUP = 320). The outliers were not considered in the plots. The tables show the Wilcoxon test results, with respective hypothesis ans significance.!______________________________________!173! ! ! ! ! ! ! ! 17 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS List of tables Chapter 2 ! Table 2. 1 - Synthesis of general forests ecohydrology and hydrological services, compared forest to non-forest areas.!_________________________________________________________________________________________________!52! Table 2. 2 – Some examples of indicators for hydrological services evaluation, organized according to the conceptual framework!______________________________________________________________________________________!54! Table 2. 3 - Datasets and sources used to build the maps from Figure 4.4.!_____________________________!58! Table 2. 4 – Methods used in each step of the framework illustration!____________________________________!59! ! Chapter 3 ! Table 3. 2 – Examples of Sensors and Satellites to Measure the Elements of the Water Cycle!______!76! Chapter 4 ! Table 4. 1 - Land use classes used in SWAT (Figure 4.1a).!_______________________________________________!90! Table 4. 2 - Potential hydrological services provision related to the SWAT outputs indicators presented in this study.!__________________________________________________________________________________________!91! Table 4. 3 - SWAT data variables for model setup and calibration/validation.!___________________________!92! Table 4. 4 - Modified SWAT general parameters for the entire Vez watershed.!________________________!93! Table 4. 5 - Modified SWAT parameters by land cover for LAI/ET and erosion calibration in crop and management (.mgt) databases.!________________________________________________________________________________!95! Table 4. 6 -!Comparison between the average annual evapotranspiration, soil erosion and nitrogen rates by land cover (simulated by SWAT) with the values from the literature and MODIS ET.!________!96! Table 4. 7 - Land cover scenarios considered for SWAT simulation period 2003-2008. Urban areas (4%) are constant.!_______________________________________________________________________________________________!97! Table 4. 8 - Calibration and validation goodness-of-fit statistics for discharge in SWAT model.!______!99! Table 4. 9 - Calibration goodness-of-fit statistics for sediments and nutrients in SWAT model.!_____!100! Table 4. 10 - Land cover effects on hydrological services provision, simulation 2003-2008.!_________!102! Table 4. 11 - Changes in precipitation, maximum and minimum temperature, in northwest Portugal, under RCP 4.5 scenario (ensemble of 4 GCMs).!___________________________________________________________!105! Table 4. 12 - Future climate effects under the RCP4.5 scenario for 2021-40 and 2041-60 periods, combined with two different land cover scenarios. Baseline 1981-2000.!________________________________!106! ! Chapter 5 ! Table 5. 1 – SWAT indicators for ecosystem services provision analysis used in the Vez watershed !____________________________________________________________________________________________________________________!117! Table 5. 2 -!Land cover scenarios considered for SWAT simulation period 2003-2008. Urban, CORN and BSVG areas (4%) are constant.!_________________________________________________________________________!119! Table 5. 3 - Spearman correlation test between the services provision and biodiversity conservation value, in the three land covers.!________________________________________________________________________________!123! ! ! Chapter 6 ! Table 6. 1 - Performance of land cover change scenarios for provision of hydrological services in the Vez watershed (based on results from chapters 4 and 5).!________________________________________________!138! ! ! ! 18 Appendices ! Table A. 1 – RUSLE equation factors (K and C) applied in northern Portugal.!_________________________!169! Table A. 2 - Tree species and respective environmental characteristics.!_______________________________!169! ! Table C. 1 -!Biodiversity conservation value of each land cover type for five major taxonomic terrestrial groups. The values were weighted by the protection factor!___________________________________!172! Table C. 2 - Water damages in Vez watershed related in the newspaper “Notícias dos Arcos”, from a master thesis (Gonçalves, 2009, FLUP)!_____________________________________________________________________!174! ! ! ! ! ! 19 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS List of abbreviations ! ARIES - Artificial Intelligence for Ecosystem Services CBD – Convention on Biological Diversity CICES – Common International Classification of Ecosystem Services CMIP5 – Coupled Model Intercomparison Project Phase 5 (IPCC) DPSIR – Drivers Pressures States Impacts Responses ESA – European Space Agency ESb – Ecosystem Services beneficiaries ESp – Ecosystem Services providers ET – Evapotranspiration EU – European Union FAO – Agriculture Organization of the United Nations FESP – Framework for Ecosystem Service Provision GCM – General Circulation Models GEO – Geostationary Orbit GIS – Geographical Information System GNSS – Global Navigation Satellite Systems ha - hectare HRU – Hydrological response unit (SWAT) IAHS– International Association of Hydrological Sciences InVEST – Integrated Tool to Value Ecosystem Services IPBES – International science-policy Platform on Biodiversity and Ecosystem Services IPCC– Intergovernmental Panel on Climate Change IPMA – Portuguese Institute of the Sea and Atmosphere IR – Infrared Radiation LAI – Leaf Area Index LEO – Low Earth Orbits LiDAR – Light Detection and Ranging MA – Millennium Ecosystem Assessment MODIS – Moderate Resolution Imaging Spectroradiometer MR – Microwave radiation MUSLE – Modified Universal Soil Loss Equation (SWAT) NASA – National Aeronautics and Space Administration ! ! 20 CHAPTER 1. INTRODUCTION NGO– Non-governmental organizations NIR – Near-Infrared Radiances NOAA – National Oceanic and Atmospheric Administration NPP – Net Primary Productivity NSE – Nash-Sutcliffe efficiency PBIAS – Per cent bias PET – Potential Evapotranspiration POES – Polar Orbiting Platforms PR – Precipitation Radar R2 – Coefficient of determination RADAR – Radio Detection and Ranging RCM – Regional Climate Models RCP – Representative Concentration Pathways RUSLE – Revised Universal Soil Loss Equation SNIRH – National System for Water Resources Information SWAT – Soil and Water Assessment Tool SWE – Snow Water Equivalent TEEB – The Economics of Ecosystems and Biodiversity TRMM – Tropical Rainfall Measuring Mission VIS – Visible Radiation yr – year The abbreviations related to the satellite products were not described here, as they are described when mentioned in chapter 3. ! ! ! 27 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 1.2. Hydrological services and forests 1.2.1. Water and forests ! Water plays an essential role in the functioning of ecosystems, underpinning biochemical cycles, supporting living organisms and their growth, and creating aquatic habitats on Earth (ChapinIII et al., 2002). More than 70% of the Earth’s surface is covered by water, and in the atmosphere there is a layer of water vapour about 90 km thick (Chang, 2009). However, the distribution of water in the planet according to the different Earth systems is uneven (Figure 1.4). ! Figure 1. 4 – The world distribution of water. ! Freshwater, the source of water for human needs, corresponds to only a small portion of the world’s total water (Fetter & Fetter, 2001). Although not directly used by humans, oceans and glaciers, where the majority of water is located, play a rather important role in the water cycle (Figure 1.5). ! ! Figure 1. 5 – The water cycle, from Diop et al., (2002). ! ! 28 CHAPTER 1. INTRODUCTION The water cycle, powered by solar radiation, is an essential process to support the dynamics of distribution and purification of water. It is based on the three physical states of water, thereby including a liquid phase, solid phase and gaseous phase (Diop et al., 2002). Water evaporates from the oceans and from the land surface to be transported and lifted into the atmosphere until it condenses and precipitates over land and oceans (Chow et al., 1988). The quantities of water vapour in the atmosphere are mainly obtained from the oceans, followed by evapotranspiration from land (living organisms and soil), and to a lesser extent from freshwater surfaces (Figure 1.5). Evapotranspiration from land returns about 65% of water back to the atmosphere (recycling moisture), which will then feed precipitation somewhere else (van der Ent et al., 2010). Precipitated water over land may be intercepted by vegetation, or when reaching the land surface, water can become either overland flow, or infiltrate into the ground flowing through the soil as surface flow or base flow to be discharged into the streams as runoff (Chow et al., 1988). From the precipitation that falls over land, about 35% returns to the oceans as water runoff and a small portion (2.5%) ends in land water surfaces (Gleick, 1993). Plants have an active role in the water cycle, namely in water fluxes, such as evapotranspiration (transpiration and interception), throughfall and sediment transport (Kundzewicz, 2002). In particular, forests are considered the most prominent ecosystems interacting with water, due to their height, dense crown canopy, spread of the root system, wide horizontal distribution and vertical coverage (Calder, 2002; Chang, 2009). With forests, the occurrence, distribution and circulation of water are modified, the quality of water is enhanced and the timing of flow is altered (Chang, 2009). Forests and water are connected by physical and biological processes, so the management of forests affects the quantity, quality and timing of water (Jones et al., 2009) (Figure 1.6). ! ! 29 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS ! Figure 1. 6 - Hydrologic responses to changes in forest disturbance and management, from Jones et al. (2009). Ecohydrology is the discipline that links both Hydrology (the study of the water processes) and Ecology (the study of the interactions between the living organisms and their environment), which has evolved since the 1990s following the growing demand for protecting rivers and streams (David M Harper, 2008). It proposes an integrative way of thinking, combining different disciplines seeking solutions for environmental problems related with water (e.g. water scarcity and allocation, floods and their consequences, water quality, eutrophication, soil erosion) (David M Harper, 2008; Vose et al., 2011). The perspectives created by this interdisciplinary view are improving the understanding of vegetation-water-nutrient connections to address the water resource issues (Newman et al., 2006; Vose et al., 2011). Parallel to Ecohydrology, Forest Hydrology is a discipline of Hydrology that studies specifically the relation of forest processes with all phases of the water cycle. However, it is restricted to forest ecosystems in an one-way influence, not considering the influence of water on vegetation growth (Chang, 2009). Studies reporting the interactions between water, biota and the physical conditions of the watershed have been done since several decades ago (Bosch & Hewlett, 1982; Kundzewicz, 2002). Those were traditionally supported by paired watershed studies with experiments in forested/nonforested watersheds, monitoring the effects of forest structure and composition on the storage and flow path of water (Jones et al., 2009). Figure 1.6 shows how forest modifiers, for instance a change in tree species, can influence the hydrologic responses of a watershed. ! ! 30 CHAPTER 1. INTRODUCTION The broad scientific context of the research developed for this thesis is provided by Ecohydrology, complemented with Forest Hydrology regarding the influence of forest management options on water resources. This thesis applies an interdisciplinary approach to predict hydrological responses as a function of land cover and climate change, socioeconomic and ecological factors (Vose et al., 2011; Le Maitre et al., 2014). This arises from the growing interest in understanding forest ecohydrological processes, in order to address water resource issues under the worldwide water crises and the on-going climate change debate (David M Harper, 2008; Vose et al., 2011). ! 1.2.2. From hydrological processes to hydrological services ! As referred above, water regulation, carbon sequestration or pollination play an important role in the regulation of essential ecological processes and life support systems through biogeochemical cycles and other processes of the biosphere (de Groot et al., 2002; MA, 2003). Focusing on water and on the processes depending on water, a broad category of ecosystem services came up crossing the MA categories – the hydrological services, i.e. the water related benefits (Brauman et al., 2007). Examples of these benefits are the water supply for household use, mitigation of flood damages, cultural services, and water-related supporting services such as plant growth (Figure 1.7). One of the most important driving forces acting on hydrological services is forest cover (Calder, 2002). There is a strong link between forests, soil and water, with forests often being compared to sponges due to their soil capacity of storing water (Calder, 2002; Lele, 2009). On the other hand, forests reduce the total annual water yield through their high rates of evapotranspiration to the atmosphere (Bosch & Hewlett, 1982). Positively, there is an increase of water infiltration into the ground, which depends on the canopy intersection (by precipitation and fog), soil characteristics and litter absorption processes (Figure 1.8). In addition, forests maintain water quality by enhancing soil stability and nutrient usage (Pattanayak, 2004; EASAC, 2009). Thus, promoting forestation actions in a watershed may ensure water supply for people and agricultural uses, as well as for hydroelectric power production (Figure 1.7). The benefits of water flow regulation however, may not be provided locally but transferred downstream (Guo et al., 2000). ! ! 31 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS ! Figure 1. 7 -The relationship between the ecohydrological processes and the provision of hydrological services, connected by attributes such as of quantity, quality, location and timing, from Brauman et al. (2007). ! Figure 1. 8 – The movement of water from precipitation through the vegetation and soil system into streams, and how it is affected by changes in vegetation, from Le Maître, et al., (2014). ! Overall, climatic, topographic and ecological conditions of the regions influence the water balance, with a significant role played by land cover, which can be targeted for ecosystem services management (Calder, 2002). Chapter 2 of this thesis addresses in more detail hydrological services provision and the relationships between forests and water. ! ! 32 CHAPTER 1. INTRODUCTION 1.3. Tools to analyse hydrological services 1.3.1. Approaches to model hydrological services ! Mapping and modelling ecosystem condition is useful to deliver additional information about the quantity and quality of services that each ecosystem can provide, taking into account the site specific conditions determined by climate, geology and other natural factors, as well as the drivers and pressures affecting ecosystem services provision (Maes, 2014).The science and practice of hydrology includes evaluating, monitoring, assessing and forecasting the quantity, timing and quality of the water resources for a better water resource management and forecasting (Su et al., 2012). Thus, it is essential to understand the water cycle with its respective climate and ecosystem interactions (David M Harper, 2008; Su et al., 2012). Two different approaches for modelling hydrological services exist: (i) traditional hydrological tools, such as SWAT (Soil and Water Assessment Tool) (Arnold et al., 1998); and (ii) new ecosystem services oriented tools, such as InVEST (Integrated Tool to Value Ecosystem Services) (Tallis & Polasky, 2009) or ARIES (Artificial Intelligence for Ecosystem Services) (Villa et al., 2009). Traditional hydrological tools are based on robust daily-runoff hydrological models at the watershed scale, requiring detailed datasets and expert knowledge. Conversely, new ecosystem services tools are based on ecosystem services indicators and models across large scales, they are more accessible for non-experts users, and they provide a more integrative picture because it considers the demands for the service (Vigerstol & Aukema, 2011). A review of several ecosystem services tools with their respective advantages and disadvantages can be found in (Bagstad et al., 2013). Despite being rare on ecosystem services literature until now, since they analyse mainly ecosystem functions rather than ecosystem services, daily rainfall-runoff models are more robust to model and mapping the water flow and quality (Crossman et al., 2013). However, additional studies to analyse the demands of the services are needed for a complete ecosystem services evaluation (Schröter et al., 2014). In chapter 2 of this thesis, an ecosystem service oriented tool was applied to illustrate a novel framework to analyse hydrological services provided by forests. In Chapter 4, a hydrological modelling tool was used to evaluate the effects of land cover/use and climate change on hydrological services provision. ! 1.3.2. The SWAT hydrological model ! Hydrological models are powerful tools for simulating the effects of management and climate on water resources (Jayakrishnan et al., 2005). However, they are approximations of reality that should be improved and evaluated against uncertainty (Chow et al., 1988; ! ! 33 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Moriasi et al., 2007). The modelling of hydrological processes has evolved following two basic forms: (a) physical models, in which the processes are modelled according to fundamental physical laws; and (b) empirical models, in which the relationships are derived from measured data (Trimble, 2009). Hydrological models can also be classified according to the process approach: (a) deterministic models, in which a given input always produces the same output; or (b) stochastic models that consider the randomness of the variables using probabilistic distributions of each variable to generate random values (Chow et al., 1988). They can also be divided in: (a) lumped models, not considering the spatial variability, with the results being averaged according to the dominant characteristics of the watershed; or (b) distributed, in which models account for the spatial variability of conditions that occur in the watershed (Trimble, 2009). SWAT is a physically based, deterministic and semi-distributed hydrological model, developed by the United States Department of Agriculture to assess the impact of management on water supplies and non-point source water pollution in a watershed or large river basins (Arnold et al., 1998). Its progress was due to the development of databases (e.g. on soil properties) and the creation of several GIS (Geographic Information System) interface tools to support the input and output of data for detailed spatially distributed analysis of hydrologic and water resources systems (Arnold & Fohrer, 2005; Jayakrishnan et al., 2005). One of the most popular GIS interfaces is ArcSWAT, which runs in the ArcGIS software (ESRI), and uses to a geodatabase model to store SWAT input and output data (geographic, numeric, text) in an organized way (Olivera et al., 2006). Major model components include weather, hydrology, soil temperature and properties, plant growth, land management, nutrients, pesticides, bacteria and pathogens, being capable of continuous simulation over long time periods (Gassman et al., 2007). The major unit of simulation is the river basin and SWAT is prepared to run in either medium watersheds (above 100km2) until very large basins. River basins have a hierarchical structure and natural boundaries, turning them into a suitable scale for integrated ecohydrological studies and modelling (Krysanova & Arnold, 2008). In SWAT, the hydrological cycle is based on the following water balance equation: where t is the simulation period, SWt is the soil water content after the simulation period, SW0is the soil water content at day i (beginning of the simulation period), Rday is the amount of precipitation, Qsurf is the amount of surface runoff, Ea is the amount of evapotranspiration, Wseep is the amount of water entering the vadose zone, and finally Qgw is the amount of return flow, all at day i (Neitsch et al., 2011). ! ! 34 CHAPTER 1. INTRODUCTION The overall hydrologic balance is simulated for each Hydrological Response Unit (HRU), including canopy interception of precipitation, partitioning of precipitation, snowmelt water, and irrigation water between surface runoff and infiltration, redistribution of water within the soil profile, evapotranspiration, lateral subsurface flow from the soil profile, and return flow from shallow aquifers (Gassman et al., 2007). Surface runoff is calculated using an empirical model, the SCS curve number method, based on 20 years of studies in the United States involving rainfall-runoff relationships in several land use and soil types (Neitsch et al., 2011). Soil erosion is calculated using a Modified version of the USLE (Universal Soil Loss Equation). Figure 1.9 represents schematically the steps that are needed to run the model in the ArcSWAT interface. First, the watershed is delineated and divided into sub-basins using a digital elevation model (DEM). Then, each sub-basin is divided into HRUs, based on unique combinations of land cover, soil and slope attributes, improving the accuracy of predicted loadings from the sub-basin (Arnold et al., 2011). Although HRUs are geographical units, their spatial position in each sub-basin merely represents the sum of the total contribution of each unique combination for the model outputs: they correspond to scattered land pieces in a sub-basin, without spatial interactions between them (Neitsch et al., 2011). This is important, for instance, for sediment or nutrient loadings, in which the loadings of a given HRU do not pass to the HRU immediately adjacent, but are added to the sum of the loadings of all the HRUs within a given sub-basin (Arnold et al., 2011). ! ! Figure 1. 9 – Simple schematic flow of the steps required in ArcSWAT." Subsequently, the model is forced with daily climatic data (precipitation, maximum and minimum temperature, solar radiation, relative humidity and wind speed) (Figure 1.9). These daily climatic inputs can be obtained either from historical records, e.g. in Portugal from ! ! 35 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS SNIRH (National System of Water Resources Information) or IPMA (Portuguese Institute of the Sea and Atmosphere), and/or generated internally in the model using a weather generator tool, with the first option being preferred for optimal model performance (Neitsch et al., 2011). Here, the important condition is to have a good climatic dataset without gaps, mainly for precipitation representativeness and model performance (Stisen et al., 2012). For this, data gap filling is an important step, using for instance, regression analysis by correlations with the nearest weather stations with complete records. The calibration process (Figure 1.9) consists in changing sensitive parameter values to minimize the deviation between observed and simulated values of discharge (mandatory), sediments, nitrates, and total phosphorus, just to mention the most common (Moriasi et al., 2007). Calibration can be made manually in a trial error exercise, which is time consuming, or using sensitivity analysis software, such as SWATCUP (Abbaspour, 2013) or SWAT’s auto-calibration tool (van Griensven, 2005). The main advantage of the first is that users can control better the processes inside the model, whereas with sensitivity analysis tools users should be well informed about the processes in the model to understand the changing in the parameters. After calibration, a validation procedure should be done to assess the performance of the calibrated parameters for an independent set of data in a different time period, with no further adjustment of parameters (Arnold et al., 2011). The performance of the model can be evaluated by graphical techniques, and it is recommended using at least three quantitative statistics: (i) Coefficient of determination that describes the degree of collinearity between simulated and measured data (R2), ranging from -1 to 1; (ii) Nash-Sutcliffe efficiency (NSE) that measures the variability of the model residuals with respect to the variability of the observations, comparing the performance of the hydrological model to that of a hypothetical model that yields as predictions the mean value of the observations (value 1 indicates equal model performance); and (iii) the per cent bias (PBIAS) that measures the deviation (%) with simulated data in respect to the observed data (0.0 indicates no deviation) (Moriasi et al., 2007; Morán-Tejeda et al., 2013). The respective equations are described in Appendix B, and the performance rates can be found in Moriasi et al. (2007). SWAT has gained international acceptance as a robust interdisciplinary watershedmodeling tool since the late 1990s, with an annual international conference dedicated specifically to SWAT topics organized since 2003 (Gassman et al., 2007). SWAT has been used for many different applications, including (not exclusively): land use change impacts (Baker & Miller, 2013; Wagner et al., 2013; Yan et al., 2013); climate change (Nunes et al., 2008; Raposo et al., 2013; Zabaleta et al., 2014), combined climate with land use change (Mango et al., 2011; Kim et al., 2013; Shi et al., 2013; Khoi & Suetsugi, 2014); water quality (Moriasi et al., 2013; Narula & Gosain, 2013; Chen et al., 2014); soil erosion (Ullrich & Volk, ! ! 36 CHAPTER 1. INTRODUCTION 2009; Ouyang et al., 2010; Zhang et al., 2014); and ecosystem services (Quintero et al., 2009; Notter et al., 2012; Liu et al., 2013; Logsdon & Chaubey, 2013). In the last 5 years, the number of publications using SWAT has increased exponentially, highlighting the year of 2013 with the highest number of published articles (292), and 2014 with already 137 published articles, according to the SWAT literature database (SWAT, 2014). The same database includes 11 peer reviewed published articles in Portugal using SWAT, starting in 2008 with paper on hydrological processes under climate change (Nunes et al., 2008), but recently mainly about water quality issues (Demirel et al., 2009; Yevenes & Mannaerts, 2011; Ferreira et al., 2014; Mateus et al., 2014). Therefore, the application of SWAT to study the effects of land cover/use and climate changes on hydrological services, as well as its application to a watershed in northwest Portugal, are novel and an important contribution to the literature. Chapters 4 and 5 of this thesis describe and discuss the main results from such studies. 1.3.3. Remote sensing for eco-hydrological assessment ! Remote sensing data and products have been successfully used to map and model the structural and functional attributes of ecosystems, allowing to overcome limitations related to spatial and temporal scales of resolution (de Jong et al., 2004). Broadly defined, Remote Sensing is the process of getting information about an object, area, region or phenomenon without physical contact with it (Lillesand et al., 2004). This broad definition would encompass, among others, space probes, most of medical imaging, non-destructive testing, sonar, as well as instruments observing the earth from a distance (Schott, 2007). The latter, commonly known as Earth Observation, refers to the detection and interpretation of the electromagnetic energy of different objects from aircrafts or satellites (Turner et al., 2003; Chang, 2009). They are used for global change studies and environmental monitoring, and satellites with different instruments able to sense different wavelengths, which can be useful for meteorological, atmospheric, land, water, biodiversity, gravimetric, topographic and soil purposes (Lillesand et al., 2004; Schott, 2007). The different features of the Earth surface and atmosphere will absorb, emit or reflect the electromagnetic radiation from the sun, which can be sensed by different instruments on board of satellites and aircrafts (de Jong et al., 2004). The electromagnetic spectrum can be divided into optical wavelengths (between 0.4 and 14 µm) and longer wavelengths known as microwave (between 1mm and 1m) in one or more bands (Figure 1.10). ! ! 43 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Chapter 2 Hydrological services and the role of forests: Conceptualization and indicator-based analysis with an illustration at a regional scale ! ! ! Abstract Forests are among the most important ecosystems for the provision of hydrological services. These include water supply and water damage mitigation, in the dimensions of quantity, timing and quality. Although the hydrological role of forests is well documented in literature, a conceptual framework integrating these three dimensions is still missing. In this study, a comprehensive conceptual framework to improve the assessment of hydrological services provided by forests was developed. In addition, the framework was tested by an illustration for northern Portugal, a region with both Mediterranean and Atlantic climatic influences. The TEEB (The Economics of Ecosystems and Biodiversity) framework of ecosystem services was adapted to the relation between forests and water. Then, this new framework was complemented with a set of spatially explicit indicators that quantify the supply and demand of hydrological services. In addition, the implications of the framework were discussed in the context of the social-ecological systems, using the DPSIR (Drivers, Pressures, State, Impacts and Responses) model. Finally, the framework and the indicators were illustrated for northern Portugal using the water supply (quantity) and soil erosion control as examples. Results show that the proposed conceptual framework is a useful tool to support land planning and forest management, adapting the provision of hydrological services to the regional biophysical and social conditions. The test of the framework across a heterogeneous region suggests that a spatially explicit combination of system property, function, service and benefit indicators can be an effective way of analysing and managing the supply and demand of the hydrological services. Keywords: Hydrological services; Conceptual framework; Forests; Indicators; Northern Portugal; Social-ecological systems. ! ! ! ! ! Carvalho-Santos C., Hein L., Honrado J. (2014) Hydrological services and the role of forests: Conceptualization and indicator-based analysis with an illustration at a regional scale. Ecological Complexity. 20: 69-80. ! ! ! ! ! 44 CHAPTER 2. HYDROLOGICAL SERVICES AND THE ROLE OF FORESTS ! ! 45 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 2.1. Introduction ! ! Hydrological services, or the water-related services provided by ecosystems, are considered crucial for human well-being (MA, 2003). As defined by Brauman et al. (2007), hydrological services encompass the benefits to people derived from the regulation of water flows by ecosystems. They include a large group of services: water supply (diverted and in situ supply), water damage mitigation, water-related cultural services and water-associated supporting services. The provision of hydrological services is usually analysed according to three dimensions: (i) quantity (i.e. total water yield), (ii) timing (i.e. seasonal distribution of the flow) and (iii) quality (i.e. removal and breakdown of pollutants and trapping of sediments) (Brauman et al., 2007; Elmqvist et al., 2009). Any ecosystem is potentially important to the provision of hydrological services, but forests are considered the main contributors (Calder, 2002). Forests (soil and vegetation) promote infiltration, increasing soil moisture content, groundwater recharge and the gradual release of water (Bruijnzeel, 2004). Some benefits are associated to this water storage and gradual release, such as hydropower generation and water supply for households (Guo et al., 2000; Brown et al., 2005; Farley et al., 2005; Brauman et al., 2007). In addition, due to the intervention of tree canopies and root system, surface runoff is low, maintaining soil stability and improving water quality in terms of sediments (Ilstedt et al., 2007; Lele, 2009). Furthermore, there is evidence that the existence of forests contributes to moderate waterrelated hazards, such as floods and landslides (Calder & Aylward, 2006; Bredemeier, 2011). However, forests may reduce the annual water yield through increased loss by evapotranspiration, and consequently limit the amount of water available in the system (Bosch & Hewlett, 1982; Sahin & Hall, 1996; Bruijnzeel, 2004; van Dijk & Keenan, 2007; Bredemeier, 2011). One important note is that the magnitude of forest influencing hydrological services described before is very site dependent (Calder, 2002). Despite recent advances in understanding the role of forests in the provision of hydrological services, there are still a number of conceptual questions remaining. These include the specific components of the hydrological service, how these components can fit in the ecosystem services framework proposed in the context of the TEEB [The Economics of Ecosystems and Biodiversity] project (Haines-Young & Potschin, 2010; de Groot et al., 2010a), and how can these components be quantified and mapped. In order to address these questions, the objective of this article is to develop a conceptual framework for understanding the hydrological services and describing the role of forests in providing hydrological services. The conceptual framework is illustrated at the regional scale using northern Portugal as a case-study region. ! ! 46 CHAPTER 2. HYDROLOGICAL SERVICES AND THE ROLE OF FORESTS Forests in northern Portugal have distinct composition, structure, functioning and spatial distribution according to the prevailing Mediterranean or Atlantic influences (Costa et al., 1998) for which hydrological effects can be compared. Special attention is given to Mediterranean forests because here the different aspects of climate, land use and soils are very related to water availability (García-Ruiz et al., 2011). Typically in Mediterranean areas water is scarce in summer, which often results in serious water stress (Nunes et al., 2008). Therefore, assessing the role of these forests in the water cycle is an interesting context to test the possible applications of the conceptual framework for hydrological services provided by forests. The paper is structured as follows. First a conceptual framework to analyse the provision of hydrological services by forests was developed, based on the TEEB ecosystem services framework and on further literature review. Also, a set of indicators to quantify hydrological services was derived from the framework. In addition, an extension of the framework was suggested in the context of social-ecological systems theory. Finally, the framework was illustrated using indicators to analyse the spatial distribution of water supply (quantity) and water damage mitigation (soil erosion control) at a regional scale for northern Portugal. 2.2. Conceptual framework for hydrological services ! 2.2.1. Conceptual framework ! The TEEB project combined insights on the loss of biodiversity, the degradation of ecosystems and changes in the supply of ecosystem services, and analysed the ecological, social and economic implications of ecosystem degradation (de Groot et al., 2010a). Accordingly, the TEEB conceptual framework, which is based on the cascade diagram of ecosystem services (Haines-Young & Potschin, 2010), classifies services on the basis of the link between ecosystems and the social-cultural dimensions of human well-being (de Groot et al., 2010a). This cascade considers a stepwise description of properties and functions, related to the biophysical system, which provides potential services to support the social systems through concrete benefits (Haines-Young & Potschin, 2010). In order to improve our understanding of the provision of hydrological services by forests, a specific framework was proposed based on the one presented in TEEB. The biophysical functions behind the relation between forests and water, as well as the resulting contributions to human wellbeing were highlighted (Figure 2.1). ! ! 47 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS ! Figure 2. 1 -"Conceptual framework for hydrological services provision by forests, showing the relationship between the biophysical ecosystem (properties and functions) and the social system (services and benefits). Inside boxes are some examples of each step of the framework. Adapted from the ecosystem services framework by de Groot et al. (2010b) and Haines-Young & Potschin (2010). ! Forests regulate the water cycle. This major function, however, only becomes a service when people use or experience the water benefits resulting from this regulation, such as water supply for household or hydropower generation (Figure 2.1). Fundamental attributes of ecosystems (properties) provide the mechanisms responsible for generating potential services (de Groot et al., 2010b). Underlying ecosystem properties, biophysical structures are the platforms where processes occur (Fisher et al., 2009). For instance, the roots of trees and shrubs promote porosity in forest soils thereby improving the infiltration process (Figure 2.1). Functions are the result of such properties, and determine the capacity of the ecosystems to provide services and ultimately benefits. This intrinsic capacity exists independently of whether people use or feel the services, which is only materialized when a beneficiary is identified (Haines-Young & Potschin, 2010). For example, the function of water flow regulation will improve human well-being by reducing the number and severity of floods. This becomes a service only if there are people benefitting from the reduced flood risk (Figure 2.1). Note that the relation between ecosystem functions and services may be complex, with functions potentially contributing to multiple services (as in the case of regulation of water flows allowing both seasonal water supply and water damage mitigation) (Ansink et al., 2008). Services are the pure aspects of ecosystems utilized by humans contributing to their welfare (e.g. water supply), while the benefits are the actual and sometimes transformed advantages to people´s well-being (e.g. drinking water for household) (Boyd & Banzhaf, 2007; Fisher et al., 2009). From one service multiple benefits may be derived, which are the fundamental units to establish valuation. Therefore, benefits translate a welfare gain that can, in principle, be valued on ecological, social and economic grounds (Lele, 2009). In our adapted framework attention was not given to the values of ! ! 48 CHAPTER 2. HYDROLOGICAL SERVICES AND THE ROLE OF FORESTS hydrological services, being the purpose of this paper to examine the biophysical relations between forests and water. From the typology of hydrological services proposed by Brauman et al. (2007), the ones that are directly linked to the flow of water through forests were selected. Therefore, the hydrological services provided by forests that are accounted here are (i) water supply and (ii) water damage mitigation, both considered according to the three dimensions of hydrological services: quantity, timing and quality. Water supply, a provisioning service in the Millennium Ecosystem Assessment classification (MA, 2003), includes private and public water-use (irrigation, industry, urban), hydropower generation and flow of water for transportation, whereas water damage mitigation includes the regulation of floods and the reduction of soil erosion by water (Brauman et al., 2007). The complex interactions between forests and water will be developed in the next section according to the three dimensions of hydrological services. The focus will be on the provision of services by different vegetation/land cover types, comparing forests with non-forested areas (see table 2.1 for synthesis). ! 2.2.2. Dimensions of hydrological services and their relation to forests 2.2.2.1. Water quantity ! An increase in plant leaf area, associated with forest expansion, generally means higher rates of evapotranspiration and consequently a reduction of water available for recharge (Zhang et al., 2001). This reduction may negatively affect the water supply service, but may also be beneficial for water damage mitigation (Brauman et al., 2007). The decrease of water surplus in rainy tropical regions is a benefit to control water hazards, so forest should be composed by high water demanding vegetation (van Dijk & Keenan, 2007). On the other hand, in regions where water is a limiting factor, as in most of the Mediterranean, the impacts of increased evaporation of intercepted water should be considered before forest plantation, and the use of low water demanding vegetation should be encouraged to avoid water scarcity (Gallart & Llorens, 2004). In the hydrology literature, the relation between forests and water quantity at a watershed scale has been analysed using paired-catchments studies. These studies have shown that there is often an increase in water yield when forests are replaced by other type of vegetation, e.g. grasslands (Bosch & Hewlett, 1982; Brown et al., 2005). Studies conducted all over the world (compilation of data from 145 experiments) revealed that, with a reduction of 10% of conifer or deciduous hardwood cover, the water yield increased on average by 20- 25mm and 17-19mm, respectively, even though there are significant local variations (Sahin & Hall, 1996). Considering the reverse situation, when forests replace other types of vegetation cover, the common measure to detect changes in water quantity is either annual ! ! 49 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS runoff reduction or water yield reduction. In a review of 26 global catchments, annual runoff decreased about 44% where grasslands were forested, and 31% where scrublands were forested (Farley et al., 2005). A compilation of French literature for the Mediterranean has shown that, after clear-cutting or fire, annual runoff typically increased with 10%, and that after afforestation annual runoff decreased again (Cosandey et al., 2005). From a broader perspective, forests may contribute to increase the amount of water available, either by improving infiltration, contributing to atmospheric moisture by evapotranspiration or by fog interception (van Dijk & Keenan, 2007). There is a trade-off between what is a decrease in the total water yield and the improved infiltration under forest cover that contributes to gain water (Malmer et al., 2010). Some authors even discuss that the benefits of increased infiltration may compensate the costs of losing water in some specific locations, such in the tropical regions (Bonell & Bruijnzeel, 2005). Other authors argue that a decrease in forest area will have implications for the local and regional evapotranspiration, and consequently to the overall water regime (Ellison et al., 2012). Forests play an important role in supplying atmospheric moisture that may eventually become precipitation, and at global scales precipitation could decrease significantly if forest continue to be cut. In some areas, forests are associated with water gains by fog interception (Chang et al., 2006). Particularly in coastal and mountain areas, tall vegetation can intercept the equivalent between 5% and 20% of total rainfall (Bruijnzeel, 2004). In a recent experiment in Hawaii, groundwater recharge was 106% of rainfall above-canopy in dense forest cover, as a result of direct interception of cloud water (Brauman et al., 2012). Structural and functional traits, such as canopy structure and density (Figure 2.1), tree water use and rate of growth, vary among tree species contributing to different water balances in forests. In general, deciduous tree species intercept lower amounts of water from rainfall annually than evergreen species, about 18 % and 31 % of total rainfall, respectively (Barbier et al., 2009). In addition, rapid growing species, such as eucalyptus, are associated with higher water demands, increasing concerns of water availability when plantations are located in areas with prolonged dry periods (Garmendia et al., 2012). This was shown in an experiment in northern Spain, where seven years after the establishment of a eucalypt plantation, the decrease in the water table dropped from about 3 mm/day to 5 mm/day (Rodriguez-Suarez et al., 2011). Overall, there are positive and negative effects (see Table 2.1) of forests and afforestation on the several processes involved in water quantity (Brauman et al., 2007). The relative importance of these factors, as well as the final balance for water yield, seems to be highly dependent on the specific environmental and social-ecological context of each region (Calder, 2002). ! ! 50 CHAPTER 2. HYDROLOGICAL SERVICES AND THE ROLE OF FORESTS 2.2.2.2. Water timing ! Forests promote infiltration, consequently inducing groundwater recharge and influencing the hydrological functioning of the watershed (Bosch & Hewlett, 1982). Therefore, the general perception is that forests influence the regulation of seasonal flows, namely reducing peak flows and increasing water on dry flows. However, the way forests regulate seasonal flows is very site specific and dependent on the type of climate, biophysical conditions of the watershed, vegetation cover dynamics and species traits (Calder, 2002). To better understand this regulation process, three factors in forest water balance should be considered: (1) the input of rain; (2) throughfall and infiltration; and (3) evapotranspiration (Figure 2.1). The effects of rainfall pattern on water infiltration are determined by the amount and frequency of precipitation throughout the year, and by the intensity of the rainfall episodes (Brauman et al., 2007). In some regions of the tropics, rainfall intensity often exceeds 100 mm/hour (Cheng et al., 2002). These extreme episodes influence soil field capacity due to the amount of water that fills the soil empty spaces in a short time (ChapinIII et al., 2002). In addition, the function of water regulation is intrinsically linked to water retention capacity across the watershed, and soil depth determines the amount of water that can be stored in the system (Neary et al., 2009). After a rainfall episode in a forest, part of the water not evaporated back to the atmosphere is stored by leaves and drips to the ground (throughfall) or runs down stems to the ground (stem flow) (ChapinIII et al., 2002). Thus, trees, understory vegetation and litter dampen and slowly release rainfall to the ground, influencing water infiltration by avoiding surface runoff. In addition, soil hydraulic properties such as conductivity and porosity influence infiltration and storage capacity of each soil (Waring & Running, 2007). Tree root system and soil microorganisms are the major agents of soil porosity and structure (EASAC, 2009; Elmqvist et al., 2010). A meta-analysis of 14 experiments in the tropics showed that, with grass and crops, infiltration ranges between 3-47 mm/h, whereas after forestation it ranges between 6-172 mm/h (Ilstedt et al., 2007). Finally, the water balance is also influenced by evapotranspiration in forest ecosystems. Canopy evaporation or interception is high in forests and will vary according to the type of tree canopies, radiation amount, temperature and wind conditions (Waring & Running, 2007). Under temperate climate, forest interception ranges from 15-50% of rainfall (Gerrits et al., 2010). On the other hand, water use by trees is a key determinant of water availability, according to canopy leaf area and climate characteristics (Aranda et al., 2012). Evapotranspiration from a forest catchment with eucalyptus could be 40-250 mm higher than from a comparable catchment with grasslands in Australia (Farley et al., 2005). Specifically ! ! 51 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS during the dry period, increasing demands of transpiration will require deep water uptake by plants, and hence a reduction of water flows (Calder, 2002). Forests can, however, compensate this by enhancing water infiltration all over the year, increasing soil water recharge and water storage, with evidences in the tropics (Bruijnzeel, 2004). The timing of water is also dependent on the seasonality of tree species traits. In temperate forests for instance, deciduous tree species are foliated during summer, when evapotranspiration is higher and precipitation is lower. An experiment with beech species in Luxembourg, revealed that in summer, canopy and forest floor interception is 36% of rainfall, whereas in winter is 26% (Gerrits et al., 2010). In turn, forest management actions can also influence the water balance. In the Mediterranean, an experiment revealed that after 33% reduction of stem basal area by thinning, interception in holm oak decreased from 31% to 21% of total rainfall (Limousin et al., 2008). It is difficult to generalize the impact of vegetation on seasonal flow regime without considering the type of climate (Calder, 2002). Empirical evidence in the literature reveals that catchments with uniform rainfall throughout the year also exhibit uniform reductions in water flow across seasons. In an experiment in New Zealand, there was approximately 30% reduction of water flow as a result of the establishment of a pine plantation (Brown et al., 2005). On the contrary, catchments with predominant winter rainfall, like semi-arid or Mediterranean regions, show larger reductions in summer flows, since rainfall and evapotranspiration are out of phase (Brown et al., 2005). This is particularly important in water scarce environments, in which the potentially beneficial effects on peak flow reduction by forests are balanced with the negative effects on dry flow reductions (Brauman et al., 2007). There is wide agreement that the mitigation of water-related hazards, such as floods, is related to the regulation of flows and improved by the presence of forests (Calder & Aylward, 2006). In tropical regions, floods are a well known phenomena and some authors suggest that deforestation is amplifying the risk of flood-related catastrophes (Bradshaw et al., 2007). However, other authors argue that the causes are much more complex and site specific, mainly related to climate change, higher population density and urban infrastructures (van Dijk et al., 2009; Tran et al., 2010b). The growing consensus is that forests may help to mitigate floods from small storms, as forests evaporate more and soils facilitate infiltration, but the effects on strong storms are likely to be insignificant (Calder & Aylward, 2006). Accordingly, studies in the Central Spanish Pyrenees have shown a reduction in the frequency of floods when there is forest cover, even if this cover does not alter significantly the negative effect of extreme rainfalls (Serrano-Muela et al., 2008). ! ! 52 CHAPTER 2. HYDROLOGICAL SERVICES AND THE ROLE OF FORESTS Table 2. 1 - Synthesis of general forests ecohydrology and hydrological services, compared forest to non-forest areas. ! Service Ecohydrological process Rationale forests-water Effect on service" Reference Water supply " Quantity Runoff (Water yield) Results in a reduction of water for discharge ê Farley et al., 2005 Evapotranspiration High evaporation of intercepted water. Forest transpiration (depending on tree species and region) ê Zhang et al., 2001 Fog and cloud interception Forest intercept water from fog (mainly in coastal and mountain areas) and clouds (tropical rain forests) promoting gradual water throughfall é Brauman et al., 2012 Atmospheric moisture Evapotranspiration from forests contributes to increase atmospheric moisture and to feed precipitation é Ellison et al., 2012 Infiltration Forests promote infiltration thereby increasing soil moisture and promoting aquifer recharge é Ilstedt et al., 2007 Timing Evapotranspiration Seasonality on the rates of evapotranspiration may reduce water losses in some regions é Brown et al., 2005 Throughfall Usually high in forest, promoting infiltration and thereby aquifer recharge é ChapinIII et al., 2002 Infiltration Forests promote infiltration and the gradual release of water throughout the year é Calder, 2002 Quality Surface runoff Lower under forest cover contributing to reduce surface sediment transport for watercourses é Neary et al., 2009 Nutrient uptake Forest organisms use large amounts of nutrients, filtering them and thereby contributing water quality é Pert et al., 2010 Water damage mitigation ! Quantity Runoff Reduction is positive for water hazard mitigation e.g. peakflow reduction in regions with very high precipitation rates and steep slopes (but only at the limit of soil retention capacity) é Tran et al., 2010 Surface runoff Low, promoting soil stability and reducing erosion by running water é Lin et al., 2008 Timing Infiltration High infiltration regulates water flows, minimizing the frequency and the effects of droughts and floods (the latter at the limit of soil retention capacity) é Calder and Aylward, 2006 épositive effect; ênegative effect ! 2.2.2.3. Water quality ! The role of forest in determining water quality can be expressed in two main outputs: water with less sediment, and water with fewer nutrients (mainly nitrogen) (Elmqvist et al., 2009). Water with less sediment is directly linked to the process of surface runoff (Figure 2.1). Tree canopies avoid direct rainfall splash on the soil. In addition, the tree root system fixes the soil, consolidating slopes and preventing river siltation (Neary et al., 2009). Therefore, if forests are present, particularly in the tropics, less surface sediment goes into the rivers, and activities such as hydropower production may be performed more efficiently (Hewawasam, 2010). Reductions in sediments can be up to 50 % after forestation, as was shown in an experiment in Taiwan, where sediments drop from about 16 000 t/year under farmland to 6 000 t/year under forests (Lin et al., 2008). However, consideration should be given to the interventions on plantation forests, such as trees establishment, harvesting or bushfires, which can mobilize large amounts of soil sediments (van Dijk & Keenan, 2007; Figueiredo et al., 2011). ! ! 59 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS magnitude is very site specific. Recent studies performed in northern Portuguese watersheds have found an apparent small impact of mass afforestation on river discharge, when compared to the magnitudes reported in the literature (Hawtree et al., 2014). This small effect of forests, probably results from a difference in water use potential of vegetation, given that precipitation and evapotranspiration in Mediterranean type of climates are out of phase. Definitive conclusions on the effects of forests on the annual water quantity are not possible at this stage. The framework however, provides a good picture of the spatial distribution of rainfall partitioning in the region as well as of the spatial relations between water, forests and the societal demands for hydrological services provision. To illustrate a more evident positive contribution of forests for hydrological services, soil erosion control was modelled using the Revised Universal Soil Loss Equation (RUSLE) (A = R * K * LS * C * P) (Renard et al., 1997). Table 2.3 and 2.4 further details the equation factors and model steps. Here, special attention is given for C factor (Figure 2.4f), because it describes the effect of land cover on soil erosion (see Appendix A – Table A.1). For example, evergreen forests, shrubs and herbaceous vegetation are known to have an important contribution for soil erosion control, receiving the lowest values for C factor (Pimenta, 1998). Table 2. 4 – Methods used in each step of the framework illustration To illustrate the framework at a regional scale for northern Portugal, we used municipalities for the water supply service and the sub-basins from the WFD (Water Framework Directive) for the erosion control, as mapping units. This is in accordance with the national water management units, in which municipalities are responsible for the distribution of water, and Property Function Service Benefit Water supply (quantity) Representation of MOD-16 evapotranspiration product (improved algorithm, based on the Pennan-Monteith equation, computed from MODIS land cover, MODIS-15 LAI/FPAR, and global surface meteorology from GMAO). Map units were converted from mm to m3. Data was aggregated at municipality level, using zonal statistics (average value). Previously, all the municipalities with less than 50% of their area (i.e. pixel number) holding information in the original MODIS dataset were excluded. R = P – ET Water surplus (R), the water that stays in the system after rainfall, is more or less equal to the mean annual precipitation (Figure 2.3b) from Wordclim database 1950-2000, less evapotranspiration (Figure 2.4a). Data was aggregated at the municipality level, using the same method described in evapotranspiration map. Water availability corresponds to the ratio water surplus (Figure 2.4b) and water abstracted by municipality (Figure 2.4d). Water abstracted is based on average annual values (2006 – 2008) to fill data gaps, at municipality level (INE, 2007). To approximate the water abstracted values, which corresponds mainly to urban use, to the estimates of the total water abstracted, we multiply these values by 20. We based this value on the premise that in Portugal about 87% of water is used for agriculture, 8% for industry, and only 5% for urban consumption (INSAAR, 2011). Water damage mitigation (soil erosion control) Forest cover from CORINE land cover map (broadleaved, evergreen and mixed forest). Aggregated at the WFD (Water Framework Directive) sub-basin scale using zonal statistics, sum of the area. C factor used to compute RUSLE equation, according to Pimenta, 1998 and Jones et al, 2011. A = R * K * LS * C Factors computed at 250m2 pixel and aggregated at sub-basin scale, average values. P factor for land management practices was not considered because data was not available. LS factor calculation, was based on Jones et al. 2011 and Helena Mitosova webpage. (http://skagit.meas.ncsu.edu). Length - √(l/22.1) 0.6 Steepness - [(sin(slope)]2 0.09)/ 0.9, 1.4 Represents the basins with dams and upstream influence according to the soil erosion risk map (Figure 2.4g). ! ! 60 CHAPTER 2. HYDROLOGICAL SERVICES AND THE ROLE OF FORESTS the National Water Institute is responsible for monitoring soil erosion and water quality. The spatial analysis was conducted on the ArcGIS 10 ESRI software. Information for maps 2.4a, b, f and g were available at finer spatial scales (1km2 resolution), but for reasons of consistency between the maps, all information was presented at the same spatial scale (municipalities or sub-basins) (Figure 2.4). 2.4. Discussion ! Figure 2.4 aims to illustrate the utility of the framework (described before in Figure 2.1) for the regional assessment of water-related ecosystem services, using a sequential cartography of different indicators. In the context of a real assessment of hydrological services to support spatial planning or land management, these indicators should be estimated using dynamic, robust, spatially and temporally explicit, modelling frameworks (e.g. SWAT (Arnold et al., 1998)), in order to capture the complexity of eco-hydrological processes involved as well as the interactions among them (Vigerstol & Aukema, 2011). Thus, Figure 2.4 is only intended to document the coarse spatial distribution of water quantity and soil erosion control in northern Portugal, as an illustration of how both the supply and demand sides of the provision of hydrological services can be considered in a common framework. Concerning the water supply service, features for evapotranspiration (Figure 2.3c) attempts higher values in the western part of northern Portugal, where climatic conditions are favourable for higher potential evapotranspiration. This determines lower water surplus values (Figure 2.4b), even if the precipitation rates are relatively high (Figure 2.3b). At the municipality level (Figure 2.4a), the pattern of evapotranspiration and water surplus is area dependent, meaning that larger municipalities will hold the highest levels of evapotranspiration and water surplus. The lowest values for both maps are mainly concentrated in the west part, where high evapotranspiration and small municipalities prevail. Water surplus (Figure 2.4b) attains high values in mountain areas, where the highest rainfall values coincide with the lowest regional evapotranspiration rates (Figure 2.3c). At high altitudes, grasslands, shrubs and rock outcrops mostly replace forests. Those areas may therefore contribute to retain water that could be transferred gradually for water supply downstream, compensating for the high rates of forest evapotranspiration in western lowlands. The service indicator for water availability illustrates how water supply provisioning areas are spatially related to those where higher water abstraction occurs. This is important for a regional analysis of supply and demand across municipalities. The water availability ratio (Figure 2.4c) is higher in municipalities where water surplus is high (Figure 2.4b) and water abstracted is low (Figure 2.4d). Lower ratios are located in the west part where the ! ! 61 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS demand for water is higher (more densely populated municipalities) and water supply (water surplus) is lower due to high evapotranspiration rates. In the east part, the ratios of water availability are higher due to lower demand for water as well as lower rainfall values. In all municipalities analysed the supply of water after rainfall partitioning between evapotranspiration and runoff (water surplus) sufficed the demand (water abstracted), translating into values of the water availability ratio higher than 1 (Figure 2.4c). This would suggest that precipitation inputs minus evapotranspiration losses are enough to satisfy people’s needs at the municipality level, considering current domestic, industrial and agricultural water consumption. Considering adequate data is available, this modelling exercise could be replicated with regular time intervals, in order to monitor both water supply and water demand. For water damage mitigation (soil erosion), the service indicator (Figure 2.4g) refers to the potential soil erosion, considering, among others, the C factor of the RUSLE (Figure 2.4f) that signals forest, herbaceous vegetation and shrubs as favourable covers (i.e., with low C factor). The sub-basins with the highest soil erosion risk are located in the west part of the region, mainly related to rainfall erosivity, length/steepness of slopes and prevailing high C factors. In the southwest part, the lower levels of soil erosion risk may be related to higher forest cover and lower rainfall erosivity. In this case, the rationale for water damage mitigation (soil erosion) service follows a slightly modified supply and demand exercise, when compared to water supply exercise. The reason is the lack of available data for the benefit indicator (e.g. amount of sediments collected from dams or avoided maintenance costs). As a proxy indicator, we chose to distinguish the sub-basins with water dams according to their soil erosion risk. Following an ecosystem service rationale, a reduction of soil loss could be achieved if some forestation efforts may be implemented, but detailed studies at watershed level are recommended. Besides, forestation efforts may indirectly raise soil erosion risk due to high forest fire susceptibility and intensive management operations in the new forest plantations. Spontaneous forest regeneration through succession could thus be a more favourable management option (Jones et al., 2011). ! ! 62 CHAPTER 2. HYDROLOGICAL SERVICES AND THE ROLE OF FORESTS Figure 2. 4 - Illustration of the conceptual framework of hydrological services provided by forests using two examples: water supply (quantity) at the municipality level, and water damage mitigation (soil erosion control) at the sub-basin level. Indicators, sources and methods (see Table 2.3 and 2.4). a) Evapotranspiration varies according to different vegetation cover as a function of precipitation, solar radiation and temperature, therefore is an indirect indicator of forest and water functioning; b) Water surplus (runoff) reflects the capacity of the system to potentially provide water; c) Water availability reflects the ration between potential available water and the estimation of water use; d) Water abstracted for urban uses and estimated for agricultural and industrial uses; e) Forest cover represents the total forest area in each sub-basin; f) Capacity to retain soil, according to C factor values for land cover related to soil erosion (Pimenta, 1998); g) Potential sediment budget, represents the RUSLE map for potential soil erosion; h) avoided maintenance costs, represents the sub-basins with dams where forests can improve the service of soil erosion control. ! ! 63 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Considering the forecasted changes in climatic conditions (e.g. Nunes et al., 2008), some general guidelines for the development of effective rural policies in northern Portugal can be pointed out, from a forest management perspective and considering hydrological services. However, when formulating those guidelines we must be aware of the limitations of the static modelling approach presented here, which limits detailed considerations on forests-water relations. An increase of forest area, especially in the drier east part of the region, may have a negative effect on water flows, especially if this expansion is achieved through more waterdemanding, fast-growing tree species (e.g. eucalypts). However, based on recent studies, the magnitudes of change are expected to be low (Hawtree et al., 2014). In all cases, tradeoffs between forest growth and water yield should be considered in any dry environment (e.g. the Mediterranean) where precipitation does not largely exceed evapotranspiration, in order to avoid undesired effects (Bredemeier, 2011). On the other hand, demand for water is higher in the humid west part, requiring more water supply than in the east part. Here, forests may play different roles in mountains and lowlands, as described above, therefore more detailed studies at the watershed level are recommended to support forestation plans and forest management practices that optimize water supply. Throughout the Mediterranean, marginal farmland abandonment is leading to increasing forest and shrub expansion, which may partially explain the observed decline in water resources and reductions in soil erosion (García-Ruiz & Lana-Renault, 2011). The increase of forest area might have positive impacts on soil erosion control in both parts of the region. The utility of the final outputs resulting from the application of this framework will always be dependent on the quality of the input data. It should also be noted that northern Portugal is presented as a specific case of high heterogeneity in the spatial distribution of climatic conditions and vegetation attributes. The outcomes of applying the framework to other regions will most likely be different, especially if environmental conditions and its spatial heterogeneity are distinct. We acknowledge that the approach for some indicators in the illustration may be somewhat simplistic, since they are constrained by data availability. Nonetheless, the illustration provides a general picture of how the framework can be applied at a regional scale and for various services. To support ecosystem management, any indicator of the framework can be adapted using the four steps to analyse the provision (properties and function) and the use/demand (service and benefit) in a sequential way. A special attention should be given to the service indicator, which materializes the connection between the provision and demand. The added value of this novel framework is thus that: (1) it provides a comprehensive and structured basis for modelling/mapping relevant ecosystem features in a sequential way; (2) it is flexible and allows using different indicators at various spatial and temporal scales; (3) linked to a dynamic model, it can be used to monitor hydrological patterns through time and ! ! 64 CHAPTER 2. HYDROLOGICAL SERVICES AND THE ROLE OF FORESTS across space; and 4) it facilitates gaining an in-depth understanding of both the physical and social features involved in hydrological service provision as part of complex social-ecological systems. Moreover, focusing on the types of ecosystem services that are particularly relevant for each type of ecosystems (in our case, hydrological services of forests) is an important step to better understand the particular ecological and social interactions in the provision of ecosystem services to improve land management options (van Oudenhoven et al., 2012; Villamagna et al., 2013). The advantages of doing so extend to the assessment of urban forest ecosystem services and indicators to inform urban planning (Dobbs et al., 2011). Our framework can be used in support of strategic land management, particularly regarding forest management plans, but also territorial spatial planning across scales. 2.5. Conclusions ! The conceptual framework developed and illustrated in this study supports the development of methods and the selection of indicators for analysing hydrological services at regional scales. It can be used for understanding the interactions between forests, water supply and water damage mitigation, in the absence or presence of forests. Additionally, the use of spatially explicit indicators is useful to make the new conceptual framework a more operational tool to support assessment, planning and technical decision. The water supply and water damage mitigation services were illustrated at a regional scale for northern Portugal, by mapping the spatial distribution of the service under Mediterranean and Atlantic climate conditions. With available data, it can likewise be applied in other regions to assess hydrological services as well as be replicated through time to trace patterns of change. Our findings suggested that a combination of indicators related to system properties, functions, services and benefits is an effective way of analysing the provision and demand of hydrological services as a whole. Our framework can be used as a basis for modelling and mapping the complex relations between land cover, ecosystem management and the different aspects of the hydrological service. This is important to improve land planning and management initiatives, such as forestation plans, adapting them to the biophysical conditions of each region and to the roles played by ecosystems in their several biophysical and socio-economic settings. Acknowledgments This study was financially supported by FCT (Portuguese Science Foundation) through Ph.D grant SFRH/BD/66260/2009 to C. Carvalho-Santos. J. Honrado received financial support from Programa Operacional Fatores de Competitividade (COMPETE/FEDER) and the Portuguese Government / Fundação para a Ciência e a Tecnologia (FCT), through Project Grant PTDC/AAG-MAA/4539/2012 (“IND_CHANGE”). The authors would like to thank Isabel Poças (University of Porto) for her precious help on remote sensing images. Also thank Rik Leemans (ESA, Wageningen University) for his comments to improve the manuscript. ! ! 65 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Chapter 3 Evaluation of hydrological ecosystem services through remote sensing ! ! ! ! Abstract Ecosystems provide hydrological services, namely water supply and water damage mitigation, essential for human well-being. Concerns over water problems and hydrological services provision have been increasing in the last decades, with special emphasis for water scarcity. For an effective water management and planning, hydrological services should be quantified, mapped and monitored. Remote sensing observations, especially satellite-based products, offer reliable, relative low-cost, spatially explicit and near-real time data. Therefore, the use of satellite observations is crucial to support assessments of hydrological services. In this context, it is necessary to be aware of the available satellite products, as well as of their characteristics in order to apply them adequately in water management. The aim of this article is to present an overview of satellite products that can be used to evaluate and monitor hydrological services provision. We start with a conceptualization of hydrological services followed by a description of the relevant elements of the hydrological cycle. Then, we examine the characteristics of the most relevant sensors and satellites used to evaluate and monitor those elements of the water cycle, and we address the use of remote sensing in hydrological modelling. We conclude with some perspectives on the use of remote sensing to support hydrological services evaluation. We based our survey on literature review and information from official websites. The assessment of water supply and water damage mitigation services can strongly benefit from the use of satellite-based products, contributing to improve the understanding of the processes and functions behind their provision, on a spatially explicit and near-real time basis. ! Keywords: Hydrological modelling; Hydrological services; Satellite observations; Water cycle. Based on: Carvalho-Santos C., Marcos B., Espinha Marques J., Alcaraz-Segura D., Hein L., Pradinho Honrado H. (2013) Evaluation of hydrological ecosystem services through remote sensing. In Alcaraz Segura D, Di Bella C M; Straschnoy J V (eds), Earth Observation of Ecosystem Services, CRC Press Taylors and Francis group, Boca Raton, pp. 219-249. ! ! 66 CHAPTER 3. EVALUATION OF HYDROLOGICAL SERVICES THROUGH REMOTE SENSING ! ! 67 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 3.1. Society and hydrological services ! Water plays an essential role in the functioning of ecosystems, underpinning biochemical cycles, supporting living organisms and their growth, and creating aquatic habitats on Earth (ChapinIII et al., 2002). In addition, humans and society rely on ecosystems to provide hydrological services and the resulting benefits (MA, 2003). Two major types of hydrological services (Figure 3.1) can be identified according to the benefits they generate: (i) water supply, which includes water for household, irrigation and industry, hydropower generation, freshwater products, transportation, recreational and spiritual benefits; and (ii) water damage mitigation, which includes the reduction in the number and severity of floods, decrease in soil erosion and sediment deposition, and mitigation of landslides (Brauman et al., 2007). Both types of services may be evaluated according to three dimensions: (i) quantity (i.e. total amount of water), (ii) timing (i.e. seasonal distribution of the water) and (iii) quality (related to removal and breakdown of pollutants and trapping of sediments) (Brauman et al., 2007; Elmqvist et al., 2009). ! Figure 3. 1 - Framework for the provision of hydrological services, in the context of social–ecological systems. Based on Rounsevell, M. D. A., et al., (2010) and Haines-Young, R., and M. Potschin (2010). The provision of hydrological services depends on the biophysical structures and processes involving water in ecosystems (Figure 3.1). The rate of ecosystem functioning determines the capacity to deliver a potential service for people (Haines-Young & Potschin, 2010). The intrinsic capacity to provide services exists on nature independently of human ! ! 68 CHAPTER 3. EVALUATION OF HYDROLOGICAL SERVICES THROUGH REMOTE SENSING options, in the form of ecosystem functions, and services are only materialized when people use or feel the benefits of those functions (Fisher et al., 2009). From one service, multiple benefits can be generated that translate a welfare gain, subject of economic, ecological and social valuation (Ansink et al., 2008; de Groot et al., 2010b). For instance, the amount of water infiltrated will recharge groundwater reservoirs, increasing water storage capacity (properties and functions; Figure 3.1). Once people use this water, the water supply service is translated into economic benefits, such as water being available for household consumption. Concerns over water problems have been increasing in the last decades, with special emphasis for water scarcity in arid and semi-arid regions (van Beek et al., 2011). Water availability is a function of the biophysical conditions of each region, and the implications of reduced loss of hydrological services tend to be exacerbated in a context of climate change (Brauman et al., 2007). External drivers of ecological change such as climatic, socioeconomic and political changes may affect the provision of hydrological services (Figure 11.1). These drivers may influence internal pressures that are directly influencing the state of ecosystems, such as land-use change or increase in water demand (Rounsevell et al., 2010). Those pressures will impact the provision of hydrological services and ultimately affect the corresponding benefits. However, not only supply, but also demand of water must be considered for an integrated water evaluation and planning (Roo et al., 2012). Therefore, responses from both governments and society are needed. For adaptation, policies to reduce water consumption are very important as a response to water overuse. For mitigation, monitoring the internal pressures, such as land use change or precipitation pattern, will help in the design of responses that maintain the integrity of hydrological services (Figure 3.1). In the face of increasing pressure on water resources and stress on ecosystems regulating water flows, sustainable water management is critical. This is illustrated by the report from UN-Water for Rio+20, which stated that the success of “green economy” depends on the sustainable management of water resources, provisioning of water supply and adequate sanitation services (UN, 2011). Su et al. (2012) recommended to promote the sharing of scientific knowledge and to provide capacity building and transfer of technology on the water cycle among countries, which includes the use of remote sensing and specially the use of satellite technologies, to improve water resource management. Hence, the use of remote sensing is crucial in supporting water management, in particular through allowing assessment and monitoring of the elements of the water balance and of the condition of ecosystems providing hydrological services, over space and time (Wagner et al., 2009). The aim of this chapter is to present an overview of how remote sensing can be used to support sustainable water management. In particular, we examine how remote sensing, ! ! 75 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Table 3. 1 –Examples of Sensors and Satellites to Measure the Elements of the Water Cycle (Continued) Snow and ice Water supply quantity (glaciers) Water damage mitigation (see level rise, flooding) Optical Snow product from AVHRR (Advanced Very High Resolution Radiometer) NOAA’s family of POES 25km Weekly free 1966- present NESDIS http://www.nsof.class.noaa.gov/ MODIS snow product - NDSI (Normalized difference snow index) Terra and Aqua 500m-1km Daily, weekly and monthly composites free 2000- present NSIDC+DAAC http://modis-snow-ice.gsfc.nasa.gov Passive microwave SWE (Snow Water Equivalent) from AMSR-e (Advanced Microwave Scanning Radiometer) Aqua 25km Daily free 2002-2011 VUA + NASA+NSIDC http://nsidc.org/ Ice map of Artic from CryoSat mission CryoSat-2 5-10km (250m) 30 days yes 2010- present ESA http://www.esa.int/ Active microwave RADARSAT ice monitoring RADARSAT-1 / 3 500km Daily yes 1995-2007- present CSA+MDA http://www.asccsa.gc.ca/eng/satellites/radarsat/ Multisatellite /sensor and algorithms GLIMS database (The Global Land Ice Measurements from Space) for glaciers ASTER (Advanced Spaceborne Thermal Emission and reflection Radiometer) combined in a database with satellite imagery and other GIS data for glaciers monitoring 150m 16 days yes 2005- present NSIDC+USGS http://www.glims.org GlobSnow Combined ground data with SMMR, SSM/I and AMSR-e 25km Daily, weekly, monthly yes 1979- present ESA http://www.globsnow.info IMS (Interactive Multisensor Snow and Ice Mapping System) NOAA/POES -AVHRR; MODIS; METEOSAT 4-24km Daily, weekly free 1997- present NIC-NOAA/NESDIS http://www.natice.noaa.gov/ims/ims.h tml ET Water supply and drought monitoring Multisatellite /sensor and algorithms MOD 16 ET product from MODIS land cover, LAI/FPAR and global surface meteorology (GMAO) Terra 250m - 1km 16days free 2001- present NASA http://modis.gsfc.nasa.gov/ SEBAL (Surface Energy Balance Algorithm for Land) uses different kinds of information. e.g. surface temperature, hemispherical reflectance and NDVI e.g. METEOSAT and NOAAAVHRR According to the study According to the study free 1998- present Bastiaanssen et al 1998 Surface water Water supply (quantity, timing and quality) and water damage Water damage mitigation Radar altimeter Ocean topography from Jason-2 Ocean Surface Topography Mission (OSTM) 2 km 10days yes 2008- present NASA, CNES, EUMETSAT, NOAA http://www.nasa.gov/ Optical Surface water extent using visible band sensors e.g. SPOT SPOT 5 – 25m 26 days yes 1986- present CNES http://www.cnes.fr/ Water quality bands from MERIS (Medium Resolution Imaging Spectrometer) ENVISAT 300 m 2-3 days yes 2005- present ESA https://earth.esa.int/ ! ! 76 CHAPTER 3. EVALUATION OF HYDROLOGICAL SERVICES THROUGH REMOTE SENSING Table 3. 2 – Examples of Sensors and Satellites to Measure the Elements of the Water Cycle (Continued) Water cycle Service Type Product/sensor Satellite Spatial resolution Temporal resolution Costs Period Institution/source Soil moisture Water supply quantity and timing Water damage mitigation (landslides and flooding) Optical ASCAT (Advanced Scatterometer) MetOp-a 25km Daily yes 2006- present EUMETSAT+ESA http://manati.star.nesdis.noaa.gov Active microwave ASAR (Advanced Synthetic Aperture radar) ENVISAT 1-5km Weekly yes 2005- present ESA https://earth.esa.int/ AMSR-e (Advanced Microwave Scanning Radiometer) Aqua 25km Sub-daily free 2002-2011 VUA + NASA+NSIDC http://nsidc.org/data/amsre/ Multisatellite /sensor and algorithms LPRM (Land Parameter Retrieval Model) AMSR-e,Nimbus SMMR, TRMM TMI, SSM/I 50km Sub-daily free 1978- present VUA http://www.falw.vu/~jeur/lprm/pubs.ht m Passive microwave Soil moisture from MIRAS (Microwave Imaging Radiometer with Aperture Synthesis) SMOS (Soil Moisture and Ocean Salinity) 50km 3 days free 2009- present ESA http://www.esa.int/ Groundwater Water supply quantity Optical Visible band (vegetation identification points) and NDVI Landsat/IKONOS and MODIS 250m 10-30days composites free (Ikonos yes) 1972- present 1999- present NASA NIR band (temperature of groundwater discharge) Landsat-5/7 15-60m 16 days free 1999- present NASA http://landsat.usgs.gov Gravimetry GRACE Earth microgravity model GRACE (Gravity Recovery and Climate Experiment) 400-500km 30 days yes 2002- present NASA+DLR http://www.csr.utexas.edu/grace/ GOCE Earth’s gravity field and geoid models GOCE (Gravity field and steadystate Ocean Circulation Explorer) 100km 10-30 days yes 2009- present ESA http://www.esa.int/ ! ! 77 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Development of remote sensing techniques for monitoring water quality began in the early 1970’s. Most of these studies evaluated empirical relationships between spectral properties and the water quality parameters (Ritchie et al., 2003). The factors that affect water quality can be grouped in: (i) those that change the energy spectra of reflected solar and/or emitting thermal radiation from surface waters, which can be measured using remote sensing techniques, such as suspended sediments/turbidity (e.g. (Potes et al., 2011)), algae (i.e. chlorophylls, carotenoids; e.g. (Carvalho et al., 2010; Song et al., 2012), dissolved organic matter (DOM; e.g. (Del Castillo & Miller, 2008; Jørgensen et al., 2011)), oils (e.g. (Jha et al., 2008), aquatic vascular plants (e.g. (Santos et al., 2009; Ward et al., 2012)), and thermal releases (e.g. (Alcantara et al., 2010); and (ii) those that are inferred indirectly from measurements of other water quality parameters (sensible to energy spectra), like most chemicals (i.e. nutrients, pesticides, metals; e.g. (Hadibarata et al., 2012)) and pathogens (Tran et al., 2010a). Multispectral image sensors, such as MERIS instrument on board of Envisat satellite, provide spectral bands with potential applications on suspended sediments, chlorophyll and other water quality parameters (Ayanu et al., 2012). The advantage of using satellite observations tools to identify and monitor surface water quality problems, is the spatial and temporal coverage of the parameters, comparing with the non readily available in situ measurements (Ritchie et al., 2003). 3.3.1.4 Soil and ground Water under the land surface is observed as soil moisture, which is located in the unsaturated rooting zone, and as groundwater, in the saturated zone (Figure 3.2). Optical sensors operating in VIS and IR bands can indirectly infer soil moisture, through partitioning of the water balance equation elements (Su et al., 2012). On the other hand, due to the dielectric constant propriety of dry soil, which changes in the presence of water, soil moisture is more accurately measured under the microwave bands, both by active and passive retrievals (Schmugge et al., 2002). Active microwave sensors (SAR at local to regional scale and scatterometers for global monitoring) emit an electromagnetic pulse, and capture the returning electromagnetic energy scattered back from Earth that is measured (Su et al., 2012). One example is the Global Soil Moisture product derived from ASCAT (Advanced Scatterometer), on board of the METOP-A satellite from EUMETSAT, which is available from 2006 until present, at a 25 km spatial resolution (Bartalis et al., 2007). In turn, passive microwave sensors measure the radiation emitted from the Earth surface and there are several algorithms to retrieve soil moisture information. One example is LPRM (Land Parameter Retrieval Model), a global soil moisture model that combines historical datasets ! ! 78 CHAPTER 3. EVALUATION OF HYDROLOGICAL SERVICES THROUGH REMOTE SENSING from1978 to present (Owe et al., 2008). However, passive sensors working at C-band and longer wavelengths are limited in areas with abundant vegetation cover (Tang et al., 2009). To overcome this limitation, the SMOS (Soil Moisture and Ocean Salinity) satellite mission, launched in 2009, is taking observations every three days at a 50 km spatial resolution (Albergel et al., 2011). The advantages of near-real-time observations of soil moisture are: (i) the better understanding of the water cycle, (ii) of how it impacts climate change, (iii) and the improved forecast of natural hazards, such as floods and droughts (Su et al., 2012). Groundwater was the last component of water cycle taking benefit of satellite technologies, but its monitoring is very important due to significant seasonal and interannual variability. Satellite measurements of vegetation distribution, topography, temperature, soil moisture and gravity have been used to collect information about groundwater presence (Becker, 2006). The gravity measurement, is based on the principle of the redistribution of water in different compartments of the Earth, which changes with the gravitational field (Su et al., 2012). Likewise, groundwater is measured by GRACE (Gravity Recovery and Climate Experiment), launched in 2002, a two-satellite mission to map the static and time varying components of the Earth’s gravity field (Ramillien et al., 2008). GRACE provides measurements of groundwater storage change with high accuracy, by separating the contributions of the other water compartments (soil moisture, ocean, evapotranspiration) (Rodell et al., 2006; Llovel et al., 2010). Finally, GOCE (Gravity field and steady-state Ocean Circulation Explorer), launched in 2009, is mapping the Earth’s gravity with unrivalled precision and is a complement to GRACE measurements. 3.3.1.5 Vegetation Estimation of vegetation water content (VWC), from local to global scales, is central to the understanding of water flows in the environment, and it is an important variable for drought and fire monitoring. Remote sensing technologies offer an instantaneous and nondestructive method for VWC assessment (reflectance in NIR and short-wave IR), considering that in situ measurements can be related to spectral reflectance of VWC in a reliable way (Wu et al., 2009). However, this type of methods needs further refinement to account for the observed effects of leaf structure, leaf dry matter, canopy structure and leaf area index (LAI) (Zarco-Tejada et al., 2003). Canopy water content has been estimated through various vegetation water indices (Yilmaz et al., 2008) composed of bands in these absorption peaks – e.g. normalized difference water index (NDWI), normalized difference infrared index (NDII), maximum difference water index (MDWI), water band index (WBI) – have been proved to be applicable ! ! 79 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS in the estimation of VWC (Chen et al., 2005). Recently, researchers have explored methodologies for VWC estimation through remote sensing techniques based on radiative transfer models, such as PROSPECT and SAILH (Jacquemoud et al., 2009; Suárez et al., 2009). 3.3.2. Water damage mitigation ! All the satellites and sensors described before can also be used to monitor water related hazards. Particularly, precipitation is the triggering factor for related water hazards. Therefore, it is important to estimate precise precipitation rates at global scale for accurate risk assessment (Tapiador et al., 2012). The observation of meteorological phenomena from satellite platforms provides a more accurate view when compared to surface observations, with advantages for climate assessment and forecasting of extreme events (Kidd et al., 2009). Floods and related damages can be detected for different periods using optical imagery, for instance from the Landsat series. The past of land surface dynamics can be reconstructed using microwave sensors (SRTM, MODIS, AMSR-e), contributing to predict hazards due to previous events of flooding, coastal inundation and land sliding, especially the more devastating ones (Tralli et al., 2005; Syvitski et al., 2012). The monitoring of ice sheets and melting process is important to follow see level rise and possible coastal inundations and erosion processes. Missions like Cryosat (2010-present) and ICESat (2003- 2010) provide a multi-year elevation data needed to determine ice sheet mass balance. Landslides can be identified using optical images (e.g. Quickbird), and can be forecasted using RADAR and InSar techniques collecting information on geomorphology and soil moisture (Tralli et al., 2005; van Westen et al., 2008). High temporal and spatial resolution satellite can be used to forewarning signal of increased susceptibility to land sliding and help to understand the processes leading to slope failures (Wasowski et al., 2010). Droughts are caused by water deficit due to increased evapotranspiration and temperature, lack of precipitation and reduced soil moisture. Several indices have used different elements of the water cycle to infer about the severity of drought (vegetation, soil moisture and evapotranspiration) (Su et al., 2012). This is of particular relevance in arid and semi-arid ecosystems, where droughts are a major factor controlling the inter and intraannual variation in the productivity of the ecosystem and, consequently, the benefits provided by the ecosystem (Hein et al., 2011). Efforts have been made to collect different satellite data to support risk assessment. An example is the Dartmouth Flood Observatory that provides a global water database and critical indices on floods and droughts and can be used for global and regional risk assessments (e.g. global flood risk (Jongman et al., 2012). ! ! 80 CHAPTER 3. EVALUATION OF HYDROLOGICAL SERVICES THROUGH REMOTE SENSING 3.4. Remote sensing of drivers and pressures of hydrological services ! Remote sensing is important to monitor the drivers and pressures that may affect the provision of hydrological services, from the security of freshwater resources to the mitigation of water hazards in the context of climate change. Climate change is a global driver of ecosystem functioning and services, associated to the change in precipitation pattern among other pressures (Figure 3.1). Climate monitoring is therefore crucial for water resource management, and long-term datasets of precipitation have been taken since the late 1970’s with high accuracy from Meteosat and NOAA series satellites (Kidd et al., 2009). Land use change, from urban expansion to farmland abandonment or forest management intensification, is another important pressure on hydrological services. Land use dynamics can nowadays be monitored with very high spatial and temporal accuracy by satellite images, such as Quickbird and IKONOS (Rogan & Chen, 2004). ! 3.5 Integrating remote sensing data with hydrological modelling ! The use of hydrological models enables managers to understand the response of a river catchment to atmospheric forcing conditions, which is important for more accurate water resource management and water hazards forecast and mitigation (Xie & Zhang, 2010). The increasing demand of spatial data for more complex, physically based and distributed hydrological models, together with the emergence of more sophisticated remote sensing products have increased hydrologists’ interest in the use of remote sensing applications (Kite & Pietroniro, 1996; Pietroniro & Prowse, 2002). In particular, new opportunities have emerged from remotely sensed data to improve hydrological modelling calibration and validation (Montanari et al., 2009). The list of remote sensing products potentially useful for hydrological modelling includes the provision of data on precipitation, land use, soil moisture, discharge and evapotranspiration (Pietroniro & Prowse, 2002). The advantages of using remotely sensed products extend to the availability of near-real time data, with a complete area coverage, which make possible to perform hydrological modelling even in regions with spatially and temporally scarce ground observations (Grimes, 2008). Focusing on near-real time characteristic, hydrological models can be enriched with time continuity and dynamic information using a family of techniques known as data assimilation (Walker & Houser, 2005). Those techniques merge models and observations accounting for uncertainties from different forcing conditions and parameterizations, improving model performance (Xie & Zhang, 2010). Although remote sensing provides ! ! 81 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS continuous and up-to-date measurements at varying spatial scales, it still relies on ground observations for algorithm development, calibration and validation (Tang et al., 2009). Remote sensing data can be used to estimate hydrologically important bio-physiographic variables (terrain, land cover/use, soil data) as well as hydrologic-state variables (e.g. precipitation) that influence the water processes in a basin or region (Pietroniro & Prowse, 2002). 3.5.1 Hydrologic bio-physiographic variables ! Hydrologic bio-physiographic variables are the spatial input data for physical and distributed hydrological models. Terrain data are used in the delineation and discretization steps of hydrological modelling (Arnold & Fohrer, 2005) .This type of data is collected with high accuracy resolution, mostly from RADAR, but also short wavelength sensors. The most popular and freely available sources are SRTM (Shuttle Radar Topography Mission) by NASA and GDEM (Global Digital Elevation Model) from ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer). In addition, land cover/use and soil data are especially important for distributed hydrological models, where the hydrological response units are influenced by the spatial variability of land cover and soil characteristics (Pietroniro & Prowse, 2002). There are freely available land cover datasets based on satellite imagery interpretation. For the entire world, Global Land Cover 2000 (1: 5 000 000) is available at the U.S. Geological Survey website. For European countries, CORINE land cover (1:100 000) used satellite imagery interpretation (Landsat, SPOT-4, SOPT-5, IRS-P6 LISS III) to produce a land cover product, which is available for years 1990, 2000 and 2006 with a common classification for all European countries (EEA, 1996). Finally, for soil information, a lowresolution map Digital Soil Map of the World, from the Food and Agriculture Organization of the United Nations (FAO), is also freely available. ! 3.5.2 Hydrologic-state variables ! Hydrologic-state variables derived from satellite observations have been introduced to complement or even replace in situ model input for hydrological modelling (Tang et al., 2009). Some examples are climatological data, evapotranspiration, soil moisture, water storage and discharge derived mainly from satellite sensors, used to calibrate and validate hydrological models (van Dijk & Renzullo, 2011). Evapotranspiration (ET) is the major link between global energy budgets and the hydrological cycle (Smith & Choudhury, 1990). Satellite remote sensing provides routine observations, such as vegetation, energy and land surface temperature, used to estimate ET (Courault et al., 2005). Two kinds of approaches ! ! 82 CHAPTER 3. EVALUATION OF HYDROLOGICAL SERVICES THROUGH REMOTE SENSING have been taken: (i) energy balance-based physical models, and (ii) empirical models that relate ET to vegetation index measurements across the growing season (Zhang et al., 2009). An example based on the first approach is the widely used model algorithm SEBAL (Surface Energy Balance Algorithm for Land) (Bastiaanssen et al., 1998). SEBAL was introduced in the calibration process of hydrological modelling, with better results when compared to the use of traditional ground based data algorithm for ET calculation (Immerzeel & Droogers, 2008). Satellite derived hydrologic-state variables are essential in poorly gauged catchments where availability of hydrological ground data is a challenge for the calibration process (Milzow et al., 2011). Some of these variables have been successfully used in several hydrological modelling efforts (Kite & Pietroniro, 1996; Fernández-Prieto et al., 2012). Precipitation is the most important input for hydrological modelling, and several satellite missions with their derived products have estimated it, such as TRMM. In some regions, such as in Amazon basin, the performance of satellite rainfall data from TRMM is comparable to data obtained by rain gauge observations (Collischonn et al., 2008). However, a recent study in China has shown that the use of TRMM data is good for monthly streamflow simulation, but unsuited for daily simulations, when compared to the use of rain gauge observations. Therefore, further developments in the algorithms of satellite based rainfall of daily estimation are needed (Li et al., 2012). Satellite radar altimetry over land allows data to be retrieved for small and narrow water bodies, which can be converted into discharge using rating curve method, to calibrate and validate hydrological modelling (Leon et al., 2006; Calmant et al., 2008). In combination with ground discharge data, it may improve discharge estimates on finer time and spatial scales (Michailovsky & McEnnis, 2012). ! 3.5.3 Remote sensing applied in SWAT ! A combination between ground based measurements and remotely sensed earth observations in a coupled model output is an interesting approach to address more accurately the water balance equation (Tang et al., 2009). A widely used hydrologic model that combines this approach is SWAT (Soil and Water Assessment Tool), developed in the early 1990s by the USDA Agriculture Research Service, USA (Arnold & Fohrer, 2005). SWAT was developed to predict the impact of land management on water resources, performing routines of simulated discharge on monthly or daily time steps. It is a semidistributed and physically based model. Improvements to the original SWAT introducing remote sensing data to calibrate and validate have been emerging (Gassman et al., 2007). Also introducing data assimilation ! ! 83 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS techniques to improve model reliability, for instance using soil moisture (Han et al., 2012). Tobin and Bennett (2012) compared the efficiency of different precipitation products (one from rain gauged observations and the others from satellite data) in six watersheds in the USA, generating streamflow under sub-monthly time steps. Comparable performance between TRMM and ground precipitation data was observed. Narasimhan (2005) applied normalized difference vegetation index (NDVI), derived from NOAA-AVHRR sensor, to verify soil moisture simulated by SWAT, as a complement to traditional streamflow calibration and validation. This study showed that simulated soil moisture could be a good indicator of crop stress in semi-arid conditions. In a study of a southern African river, some changes were introduced in the original SWAT code allowing the combination of radar altimetry, precipitation products, SAR surface soil moisture and GRACE total storage changes. Although the estimates of rainfall differed among precipitation products, surface soil moisture and total water storage allowed to identify likely errors in the periods when precipitation had higher discrepancies (Milzow et al., 2011). 3.6 Conclusions and perspectives ! This chapter provides an exhaustive review of the many advantages that earth observation offers for the evaluation, management and monitoring of hydrological services. First, because it strongly improves the spatial quality of distributed information when compared to ground based measurements. Likewise, it allows better access to remote regions with poor ground measurements. Secondly, it can offer information on a near-real time base, which is particularly useful to predict natural disasters and activate emergency plans, already highly developed for precipitation alerts (e.g. TRMM). Finally, the use of remote sensing allows the assessment of the three dimensions of hydrological services, i.e. quantity, timing and quality. Considering the consistent and frequent time coverage information of the water cycle, it is possible to improve the study of interannual variability and seasonal behaviour of the water cycle elements (Schulz et al., 2009). Landsat and other images for more detailed analysis of vegetation and land use patterns are available on a planetary scale since the early 1970s. However, there are still some limitations regarding the parameterization as well as the costs of the observations. The hydrological information gathered by satellite needs a robust parameterization and validation to improve the accuracy and consistency in hydrological studies, preferably using ground-based measurements (Tang et al., 2009; Hein et al., 2011). Although the cost of earth observations can be high, depending on their spatial and temporal resolution, the use of satellite observations seems to show higher cost effectiveness than conventional methods for hydrological parameters observation (Dreher et al., 2000; Pietroniro & Prowse, 2002). ! ! 84 CHAPTER 3. EVALUATION OF HYDROLOGICAL SERVICES THROUGH REMOTE SENSING In the future, the increasing number of more sophisticated satellite missions will create new opportunities for observing, analysing and monitoring the different components of the water cycle (Fernández-Prieto et al., 2012). In addition, the integration of remote sensing applications, particularly satellite-derived products, in hydrological modelling has progressed significantly, improving the spatial and temporal resolution of the outputs. A remaining challenge is the development/improvement of satellite algorithms, adapting them to the routines of hydrological models, and further data assimilation process. Moreover, these satellite products should be calibrated and validated according to the environmental characteristics of each region. Overall, the assessment of water supply and water damage mitigation services can strongly benefit from remote sensing techniques and data, especially from satellite-based products as described above. Satellite products contribute to improve the understanding of the processes and functions behind the provision of hydrological services, on a spatially explicit and near-real time basis. Furthermore, the drivers and pressures affecting the water cycle and the provision of hydrological services can also be assessed and monitored, thereby contributing to a more robust and efficient management of hydrological resources and services. ! ! ! ! ! ! ! ! ! Acknowledgments This study was financially supported by Fundação para a Ciência e a Tecnologia (Portugal) through Ph.D grant SFRH/BD/66260/2009 to Claudia Carvalho-Santos. João Honrado received support from Fundação para a Ciência e a Tecnologia (Portugal) through project grant “ECOSENSING” (PTDC/AGR-AAM/104819/2008). Bruno Marcos received support from Fundação para a Ciência e a Tecnologia (Portugal) through project grant “MoBiA” (PTDC/AAC–AMB/114522/ 2009). ! ! 91 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 4.2.2. Input data and SWAT setup ! SWAT is a physically based semi-distributed model with daily and monthly calculations of hydrological balance parameters in the watershed (Arnold et al., 1998; Neitsch et al., 2011). The watershed is divided into homogeneous simulations units, the Hydrologic Response Units (HRUs), each consisting of a unique combination of land use, soil type, and slope features (Gassman et al., 2007). SWAT uses the curve number (CN) equation method to predict surface runoff from rainfall for different types of soil and land cover (Arnold et al., 1998). SWAT includes also routines allowing the simulation of vegetation growth from seedling to mature stands, considering plant phenological development, leaf area, radiation interception, and biomass (Gassman et al., 2007). SWAT itself only gives information on the capacity of provision for hydrological services (potential supply of services; Table 4.2). Further analysis should be carried out to have the demand side evaluated for a complete hydrological services study. Table 4. 2 - Potential hydrological services provision related to the SWAT outputs indicators presented in this study. Hydrological service SWAT outputs/indicators Water supply Quantity Total annual water yield (mm) and monthly discharge (m3/s) Timing Flow duration curves (Q95 - m3/s); soil water content (mm) Quality Nitrate concentration (mgN/l); nitrogen export (kg/ha.yr) Water damage mitigation Erosion control Soil erosion (t ha.yr); % of the basin under soil erosion risk Flood regulation Flow duration curve (Q5 - m3/s); soil water content (mm) Input datasets and sources for SWAT calibration in the Vez watershed are presented in Table 4.3. Land cover classes of the original dataset were aggregated into seven major groups, representing the similarities in terms of hydrological traits (Table 4.1). Five major soil groups were used (Figure 4.1b). The generic parameterization of vegetation and soil was based on a previous SWAT study in Portugal (Nunes et al., 2008). Subsequently, parameterization was adapted according to the vegetation characteristics of the study region based on literature with further refinement using the MODIS ET product (see section 2.4 and Table 5). Climatic parameters (Table 4.3) were used from 5 rainfall stations inside the watershed for precipitation, and 1 climatic station outside (5 km) for the other climatic variables (temperature, solar radiation, humidity and wind speed). Simulation period was limited to five years because of data availability and even during this period linear regression analysis were performed to fill data gaps, in particular for days with less than 20 hours of records. For regression analysis, data correlation with 10 rainfall stations surrounding the ! ! 92 CHAPTER 4. SIMULATING THE EFFECTS OF LAND COVER AND FUTURE CLIMATE watershed was considered. Data gap filling is important for precipitation representativeness and model performance (Stisen et al., 2012). Model setup was done using the ArcSWAT 2009.93.7a interface for ArcGis (Winchell et al., 2010). A digital elevation model (DEM) was used for watershed discretization and delineation, resulting in 10 sub-basins. Subsequently, slope, land cover and soil data (Table 3) were applied for model parameterization, resulting in 185 HRUs (using a 50 ha minimum threshold to reduce the number of insignificant combinations). Slope was divided in 3 classes, defined in order to represent the heterogeneity of this variable in the watershed: 0- 10%; 10-25%; > 25%. The major land cover class MIGS was defined as 80% shrubland and 20% of degraded forest and tall shrubs (Table 4.1). Also the class CORN was divided into 80% corn and 20% pasture. In addition, two management operations were applied for corn and pasture based on knowledge and fieldwork on crop management practices in the watershed. For corn, first a tillage operation was applied by the end of April, followed by a model-managed fertilization, based on soil N deficiency, with 15-15-00 fertilizer (175 kg/ha maximum) starting in early May. Finally, harvest operations were applied in September. In October a tillage operation for winter pasture was applied, which is finally harvested in late March to start the corn cultivation. Some fields, mainly in the northern part of the watershed are permanent pastures. 2 grazing operations were implemented (April and September), along with model-managed fertilization with daily fresh manure. In addition, a management operation was created for one land-cover scenario, which must take vineyards into account. This includes a model-managed fertilization with 22-14-00 fertilizer (40 kg/ha maximum) starting in February, and a harvest operation in September. Table 4. 3 - SWAT data variables for model setup and calibration/validation. Variables Source Description DEM (IGeoE, 2012) 20m resolution grid derived from contour line map 1:25 000 Land cover (SIGN II, 2008) (IGP, 1990) Land cover map 2006, 1:25 000. Classes were aggregated into 7 main cover classes (see Table 4.1) Validation period - Land cover map 1990, 1:25 000 Soil (Agroconsultores and Geometral, 1995) Soil map 1:100 000, with 5 main soil types (Humic Regosols, Humic Leptosols, Dystric Fluvisols, Antrosols and Urban) Rainfall SNIRH - National Hydrological Resources Information System (SNIRH, 2012) Inside the basin, hourly rainfall (mm) data were collected for 5 stations and turned into daily values (see Figure 4.1); Calibration period: 1999-2008 (4 year initialization period) Validation: 1984-1989 Climate (other variables) Hourly values were converted into daily temperature (min., max. Cº), solar radiation, relative humidity and wind speed (m/s) from climate station Ponte da Barca (see Figure 4.1) River discharge Hourly observed river stage were converted into daily river discharge values (m3/s) and averaged for the basin using stage-discharge relationships (see Figure 4.1) Calibration period: 2003-2008 Validation: 1984-89 Total suspended solids (TSS) Sparsely daily observations were taken from two sediment stations (period 2003- 2008) Nitrates (NO3) Sparsely daily observations were taken from two quality stations (period 2003- 2008) ! ! 93 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Finally, the model was forced with daily climatic parameters (precipitation, maximum and minimum temperature, solar radiation, relative humidity and wind speed) for the period 2003- 2008. A four–year (1999–2002) warm-up period was used to minimize the impacts of uncertain initial conditions. To improve model performance, 10 elevation bands for precipitation were established with a degree of change of 1 300 mm/Km (northern subbasins) and 1 100 mm/km (southern sub-basins). These elevation bands were calculated using precipitation records of 20 years (non consecutive) from the stations inside the basin, in relation to the altitude. To improve temperature inputs, we used the lapse suggested in the literature for northwest Portugal, of a 5ºC decrease on average per 1km altitude (Ribeiro, 1945). Evapotranspiration was estimated using the Hargreaves equation, since preliminary runs with Pennan-Monteith equation produced unsatisfactory results, mainly due to the low quality of wind-speed records. 4.2.3. Calibration and validation against discharge ! To reduce uncertainty in hydrological model predictions, parameters must be properly calibrated against observed values, especially and foremost against discharge (Engel et al., 2007). The observed daily discharge data at the Pontilhão de Celeiros hydrometric station (Figure 4.1) was used for this exercise. Here, a split-sample approach was applied, using part of the dataset for calibration and an independent dataset for validation, without further adjustment. SWAT calibration was done manually adjusting some sensitive parameters against daily-observed discharge for the period 2003-2008 (Table 4.4). Validation was done for the period 1984-1989 against daily-observed discharge. Table 4. 4 - Modified SWAT general parameters for the entire Vez watershed. Parameters Description in SWAT Initial value Calibration Calibrate streamflow Alpha_BF Baseflow alpha factor (days) 0.048 0.2 RCHGR_DP Deep aquifer percolation fraction 0.05 0 GW_delay Groundwater delay time (days) 31 1 CN2 Curve number for moisture condition various -10 (added to all classes) DEPIMP_BSN Depth to impervious layer (mm) 0 2000 Calibrate sediments SLSUBBSN Average slope length (m) (automatically estimated) 9 USLE_P Support practice factor 1 0.05 (Corn/PAST)/ 0.5 (Vine) USLE_K Soil erodibility factor 0.26 0.02 * Only applied to CORN and PAST with slope above 10% ! ! 94 CHAPTER 4. SIMULATING THE EFFECTS OF LAND COVER AND FUTURE CLIMATE Model performance was evaluated with statistical measures commonly used in hydrological modelling, namely the coefficient of determination (r2), the percentage bias (PBIAS), and the Nash–Sutcliffe-Efficiency (NSE). The corresponding equations and the scale of goodness can be found in Moriasi et al. (2007). 4.2.4. Model verification for Leaf Area Index (LAI) and Evapotranspiration (ET) Traditionally, calibration and validation in SWAT studies are based only in streamflow parameters, namely discharge and water quality, which can mask whether the internal catchment processes are simulated correctly, particularly the ones related with vegetation growth and evapotranspiration (ET) (Strauch & Volk, 2013). This becomes important when the objective is to simulate the influence of land cover changes on hydrological processes. In the absence of locally measured vegetation parameters (which is the most accurate approach for vegetation growth simulation), model simulations may rely on the adjustment of the SWAT generic vegetation parameters database, according to the vegetation traits of each region. Here, key vegetation parameters were adjusted to optimize LAI and ET based on values expected from the literature (Table 4.5). In addition, simulated values were compared to annual evapotranspiration for different land cover classes taken from MOD16A2 during 2000-2008 (Table 4.6). MOD16A2 is a 1 km2 resolution product from the Moderate Resolution Imaging Spectroradiometer sensitive to transpiration by vegetation, evaporation from canopy and soil surfaces (MODIS, 2010). Pure land cover pixels (≥ 75% occupied by a single land cover type) from MOD16A2 were selected across northwest Portugal, since the Vez watershed is too small to contain a representative number of 1 km2 pure land cover pixels (see Appendix B, Figures B.1 and B.2). 4.2.5. Model verification for Total Suspended Solids (TSS) and Nitrates (NO3) ! Given the limited amount of observed TSS and nitrate data, a comparison was made between the SWAT simulated values against estimated in streams, calculated from the existing observations. Observed data consisted of one measurement per month of TSS and nitrates at two water quality stations (Figure 4.1), during the period 2003-2008. The existing data was not sufficient for a complete calibration and validation study, and therefore a simple model calibration and verification was performed. In both cases, the lack of observed values for 1984-1989 prevented a full split-sample approach, and the poor quality of observations avoided a more in-depth analysis than a simple verification. ! ! 95 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Existing TSS data were used to calculate a sediment rating curve with a power function, estimating daily sediment yield from daily discharge (Lane et al., 1997). Discharge simulations from the calibrated SWAT model were used for this purpose. The following sediment rating curves were derived: 0.0018*discharge 1.1873, with = r2 0.85, if flow < 40m3/s; 0.0001*discharge 1.9665, with = r2 0.93, if flow > 40 m3/s. These curves were used to estimate daily sediment yield values from SWAT simulated discharge, which were then compared with monthly sediment yield modelled by SWAT (Table 4.6). For nitrates, existing measurements were considered representative of at least the typical value and seasonal variability for this parameter, although not of the variability with discharge. Nonetheless, each measured value was assumed to be representative of the days with no records for the respective month, and used together with simulated discharge to calculate daily nitrate export. These estimates were compared with SWAT simulations for total nitrate exports at the monthly scale. Table 4. 5 - Modified SWAT parameters by land cover for LAI/ET and erosion calibration in crop and management (.mgt) databases. Parameter Description Calibrated (initial values) PINE EUCL OAK MIGS CORN BSVG T_BASE Minimum temperature for plant growth (ºC) 9 (0) 9 (8) 9 (10) 9 (12) 9 (-) 9 (12) T_OPT Optimum temperature for leaf development (ºC) 23 (30) 22 (-) 23 (30) 22 (25) 25 (-) 23 (25) BLAI Maximum potential leaf area index (m2 /m2) 4 (5) 3.7 (5) 6 (5) 2 (2.25) 5 (6) 1.5 (-) ALAI_MIN Minimum leaf area index for plant during dormant period (m2 /m2) 3.9 (0.75) 3.4 (4) 0.75 (-) 1.8 (0) 0 (-) 1.4 (0) MAT_YRS Number of years required for tree species to reach full development (years) 30 (50) 15 (-) 50 (-) n/r n/r n/r CHTMX Maximum canopy height (m) 25 (10) 25 (-) 20 (6) 2 (1) 2 (2.5) 1 (-) RDMX Maximum root depth (m) 12 (3.5) 12 (-) 10 (3.5) 1.5 (2) 1 (2) 1 (2) PHU_PLT (.mgt) Total number of heat units for plant maturity 3500 3500 3500 1800 1800 1800 USLE_C Minimum value of USLE C factor for water erosion (factor) 0.001 (-) 0.001 (-) 0.001 (-) 0.001 (0.003) 0.005 (0.2) 0.002 (-) OV_N (.hru) Curve number for moisture condition 0.8 (0.14) 0.4 (0.1) 0.8 (0.14) 0.8 (0.15) 0.17 (0.14) 0.4 (0.15) Bibliography (Sabaté et al., 2002; Breuer et al., 2003; López Gonzalez, 2006; Nunes et al., 2008; Barbier et al., 2009; Vitale et al., 2011; Clark et al., 2012) n/r – not required; (initial value). A final verification of the exports from sediments and nitrogen by land cover was made, since good model performance for streams does not imply an adequate simulation of differences between land covers. Model results per land cover were compared with typical values for Portugal and other Mediterranean regions presented in the literature for soil erosion (Hooke, 2006; Cerdan et al., 2010) and nitrogen exports (Rial-Rivas, 2007; Machado, 2012). The comparison focused mainly on the relative differences between land covers for both parameters (Table 4.6). ! ! 96 CHAPTER 4. SIMULATING THE EFFECTS OF LAND COVER AND FUTURE CLIMATE Model performance for calibration of sediments and nutrients in streams was evaluated with the same statistical measures of performance referred before (r2; PBIAS and NSE). For the land cover scale, the data only allowed for a simple comparison. Some sensitive parameters were adjusted in SWAT for sediments (Table 4.4 and 4.5). In particular, erosion by land cover type was adjusted through the USLE_C parameter (Table 4.5). For nitrates and nitrogen, no parameters were modified other than the maximum amount of mineral N allowed to be applied in the management operations. Table 4. 6 -!Comparison between the average annual evapotranspiration, soil erosion and nitrogen rates by land cover (simulated by SWAT) with the values from the literature and MODIS ET. ET (mm) Soil erosion rates (t/ha.yr) N export (kg/ha.yr) Land cover SWAT MODIS SWAT Cerdan et al. (2010) Hooke (2006) SWAT Literature Corn + pasture rotation Vineyard 588 737 0.017 0.17 0.84 8.62 3.6 13.5 9.37 1.51 16.8 (3.9 – 40.2) (Rial-Rivas, 2007) Eucalypt 647 740 0.002 0.18 1.1 0.34 0.5 1.4 (Machado, 2012) Pine 691 763 0.001 0.25 Oak 599 0.003 0.46 Mixed trees/shrubs 495 660 0.002 0.54 0.48 0.42 Shrub 0.002 0.24 Pasture 433 - 0.003 0.32 - 0.25 Sparsely vegetated 461 457 0.005 - 4.77 0.28 4.2.6. Scenarios 4.2.6.1. Land cover ! In the last 50 years, landscape changed significantly in river Vez region. For the period 1958-1995 a large decrease in agricultural areas and low shrublands, counteracting with an increase in tall shrublands and forests was documented, probably related with land abandonment and emigration (Moreira et al., 2001). Since the 1990’s, forest declined 15% while low shrublands expanded 14%, mainly related with increasing fire occurrence and intensity, highlighting the major fire that occurred in August 2006 (Proença et al., 2010a). Considering this, our land-cover scenarios were developed according to four possible trends expected to Vez watershed, taking into account the current simulation as a reference scenario (Table 4.7). Three of them imply some management in the territory: (i) Oak - conservation with an increase of autochthones and deciduous tree species, in particular common oak (Quercus robur) in areas now covered with shrubs; (ii) Eucalyptus/pine - intensification of forest area with fast-growing evergreen tree species, such as eucalyptus (Eucalyptus globulus) and maritime pine (Pinus pinaster Aiton) in the shrubland areas; and (iii) Agri/vine - agriculture with the spread of corn and vineyards in the shrublands close to agricultural areas. The last scenario, (iv) Low vegetation - considers that environmental ! ! 97 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS pressures, such as grazing or fires, will lead to degradation with low and sparse vegetation. Classes were manipulated using the same land-cover map (2006). All the scenarios keep the present urban, agricultural and sparsely vegetated areas as static across the landscape (Table 4.7). Table 4. 7 - Land cover scenarios considered for SWAT simulation period 2003-2008. Urban areas (4%) are constant. % Land cover Current Oak Conservation Eucalyptus/pine Intensification Agri/vine Corn/vineyard Low vegetation Degradation () CORN 18.7 18.7 18.7 18.7 18.7 PINE/EUCL 8 - 63.5 - - OAK 8.7 63.5 - - - BSVG 13.8 13.8 13.8 13.8 60.6 MIGS 47.2 - - 16.7 16.7 VINE - - - 46.8 - 4.2.6.2. Future climate ! To simulate the effect of future climate conditions, two scenario periods were considered, 2021–2040 and 2041-2060. SWAT was forced with an ensemble of climate model simulations for scenario RCP 4.5 (representative concentration pathways) from the IPCC AR5 assessment report (Meinshausen et al., 2011). RCP 4.5 is a medium stabilization scenario, which assumes that all nations of the world undertake emissions mitigation policies simultaneously and effectively (Thomson et al., 2011; Vuuren et al., 2011). The ensemble approach incorporated the spread of individual ensemble members to reduce uncertainty (Déqué et al., 2007). The ensemble was composed by four GCMs (General Circulation Models) members from the coupled model intercomparison project phase 5 (CMIP5): CNRM-CM5 (Centre National de Recherches Meteorologiques, France), CSIRO-MK3.6 (Commonwealth Scientific and Industrial Research Organisation, Australia), MRI-CGCM3 (Meteorological Research Institute, Japan) and MPI-ESM-LR (Max Planck Institute for Meteorology, Hamburg). In order to compare model fields for the ensemble, models were regridded to the coarsest resolution model (1.875º x 1.875º) using a bilinear interpolation algorithm. For each GCM, the variables precipitation, maximum and minimum temperatures were subject to a linear scaling approach for bias correction (Lenderink et al., 2007; Hurkmans et al., 2010). This approach applied month correction factors to daily-modelled time series of the given month. For temperature, the station Viana do Castelo (about 40 km from the watershed) was used, whereas for precipitation data were collected from two stations inside the watershed. Correction factors are determined as differences (temperature) or quotients (precipitation) between long-term mean of observed and GCM ! ! 98 CHAPTER 4. SIMULATING THE EFFECTS OF LAND COVER AND FUTURE CLIMATE control data. Three 20-years windows were used for all variables: 1981- 2000 (baseline), 2021-2040 and 2041–2060 (scenario periods). The baseline period for temperature variables was affected by a lack of data between September 1996 and June1997, due to a shutdown in measurements. Moreover, our analysis relied on a 20-years period instead of a climate normal (30-year period) also due to the lack of data before this period. Bias corrected daily scenario temperature and precipitation variables (Tmax scen d,m; Tminscen d,m; P scen d,m ) were estimated by adding or multiplying the long-term monthly correction factors to the modelled daily time series: Eq. (B.1) Tmax scend,m = Tmaxensemb scend,m + (Tmaxobsm - Tmax ensemb con m)! Eq. (B.2) Tmin scend,m = Tmin ensemb scend,m + (Tmin obs m - Tmin ensemb conm) Eq. (B.3) P scend,m = P ensemb scend,m · (P obs m / P ensemb conm) ! where Tmax ensemb scend,m, Tmin ensemb scend,m and P ensemb scend,m are the ensemble modelled daily minimum and maximum temperature, and precipitation for the scenario periods. Tmax obs m , Tmin obs m and P obsm are monthly long-term mean of observed maximum and minimum temperature, and precipitation. Tmax ensemb conm, Tmin ensemb conm and P ensemb conm are the monthly long-term mean of the control period for maximum and minimum temperatures and precipitation. Monthly correction factor approach was adopted to reduce considerable day- to-day variability in the correction factors, which may occur when using daily correction factors. 4.3. Results and discussion ! 4.3.1. SWAT model performance ! The comparison between daily observed and simulated discharge in the calibration and validation periods indicated that SWAT was able to capture and reproduce the average flows and seasonal variations in Vez watershed, except during major high flows or extreme conditions (Figure 4.2). Predictions on monthly discharge were highly accurate (NSE = 0.87 and PBIAS = -13%) (Table 4.8), while daily simulations performance can be considered good (NSE = 0.76 and PBIAS = -15%) (Moriasi et al., 2007). Validation performance statistics (1984-1989) can be considered good for monthly simulation (NSE = 0.73 and PBIAS = 6%), but only satisfactory for daily (NSE = 0.38 and PBIAS = 6%) (Table 4.8). This may be partly related to the difference in the accuracy of climatic datasets for the validation period, particularly for precipitation, in which existing data was reported as daily sums at ! ! 99 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 9:00hrs, while streamflow was reported at 24:00hrs, causing a mismatch between the input and output data of the model, more relevant at the daily scale. In any case, lower performance on validation in respect to calibration simulations, as well as lower accuracy on daily regarding to monthly predictions, are typical pattern also for other studies (Engel et al., 2007). ! Figure 4. 2 – Daily observed and simulated discharge for calibration and validation period, after parameter calibration (table 4.4 and 4.5). Performance statistics are presented in Tables 4.8 and 4.9. Table 4. 8 - Calibration and validation goodness-of-fit statistics for discharge in SWAT model. A comparison between MOD16A2 ET values per land cover (Table 4.6) indicates that forest related covers have the highest ET values in the region, followed by agriculture. In turn, sparsely vegetated has the lowest ET values, being the most similar when compared to the simulated by SWAT. There is a difference of 100-150mm between the ET values from MODIS and the simulated by SWAT, with the latter showing the lower values by class. As the LAI achieved by SWAT after parameterization represents the values from each land cover class presented in the literature, the use of MODIS ET was convenient to verify the accurate simulation of the relative differences among land cover classes, rather than approximate the values. The calibration of sediments was not adequate in terms of model performance statistics, but this could be expected from the low data availability and related estimation uncertainty (Table 4.9). Several studies have presented similar issues in sediments and/or nutrients calibration due to data limitations (De Girolamo & Porto, 2012; Molina-Navarro et al., 2014; Calibration (2003-08) Validation (1984-89) Daily Monthly Daily Monthly R2 0.76 0.87 0.40 0.73 PBIAS (%) -15 - 13 6 6 NSE 0.73 0.86 0.38 0.73 ! ! 100 CHAPTER 4. SIMULATING THE EFFECTS OF LAND COVER AND FUTURE CLIMATE Zabaleta et al., 2014). In this case, a difference of one order of magnitude between observed and simulated sediments was evident (PBIAS = 102.9% and 65.3%), but with a good correlation (monthly R2 = 0.80 and 0.79). The soil erosion rates simulated by SWAT for each land cover are low, when compared to the ones presented in the literature (e.g. forests: 0.001 - 0.003 t/ha.yr in SWAT simulation; 0.18 – 1.1 t/ha.yr in the literature) (Hooke, 2006; Cerdan et al., 2010). This can be attributed to several factors, including soil stoniness, which leads to low soil erosion, and the presence of agricultural terraces in the Vez watershed (Cerdan et al., 2010). In addition, soil erosion is scale-dependent, considering that values reported in the literature are usually for hillslopes, while SWAT considers sediment yield at larger scales, where rates are expected to be lower (de Vente & Poesen, 2005). Therefore, the low simulated erosion rates were considered consistent with what is expected for the study site, especially given the low observed TSS data. Regarding nitrates, the simulation can be considered satisfactory (NSE = 0.38; and PBIAS between -13% and 30.2%). The simulated nitrogen exports are similar to values observed for a forested watershed with eucalypts and pines in central Portugal (Machado, 2012) (Table 4.6). When comparing arable land, simulated values are inside the range reported for Galician watersheds (Rial-Rivas, 2007). ! Figure 4. 3 – Comparison between monthly observed and simulated values for sediments (t/ha.yr) and monthly nitrates (ton. NO3), after parameter calibration. ! Table 4. 9 - Calibration goodness-of-fit statistics for sediments and nutrients in SWAT model. ! ! ! ! ! ! ! Sediments Nitrates Outlet 7 Outlet 10 Outlet 7 Outlet 10 R2 0.80 0.79 0.39 0.38 PBIAS (%) 102 65.3 -13.3 30.2 NSE -1.98 0.13 0.38 0.38 ! ! 107 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Figure 4. 6 – a) Average monthly discharge (m3/s) under future climate conditions; and b) percentage of change in monthly average discharge under future climate conditions combined with land cover scenario of increase eucalyptus/pine forest or increase agriculture.! Overall, the provision of hydrological services, particularly the timing of water supply with incidence on drier low flows and flood mitigation, should be affected by climate change conditions under a medium emission scenario in the Vez watershed, more pronounced for 2041-60. With this approach, a picture of changes in the provision of hydrological services under a medium emission scenario was given. Other scenarios and models could have been used, but the objective of this study was to give a medium scenario in combination with land cover. Another limitation of this study is the use of a GCM with statistical downscaling instead of dynamical downscaling by the use of a RCM (Regional Climate Model). At the moment, RCMs for the new RCP scenarios are not available, limiting the available choices for downscaling. However, the bias-correction step approximates the scenario values to the climatic patterns of the region. Moreover, the use of 20-year window instead of 30 years was imposed by the lack of observed data. Other studies have used the same time period with success (Morán-Tejeda et al., 2013; Shi et al., 2013). 4.3.4. Combined effects of land cover and future climate conditions on hydrological services provision ! The combined effects between future climate conditions and land cover were assessed using climate projections and the two land cover scenarios with the most substantial results from section 4.3.2, namely the eucalyptus/pine and agri/vine scenarios. These scenarios can also be considered the most realistic for future land cover in the Vez watershed, since current policy and incentives favour eucalypts and farming expansion, due both to the fast growth of exotic tree species, and the economic benefits associated with plantation forestry ! ! 108 CHAPTER 4. SIMULATING THE EFFECTS OF LAND COVER AND FUTURE CLIMATE and farm expansion. The model results for the combined scenarios were then compared to the baseline climate (1981-2000) with current land cover (shrubland dominated). The combined effects of climate and land cover change can both enhance or degrade hydrological services provision, depending on the type of service and of land cover scenario (Table 4.12). Both land cover scenarios showed a reduction in the total annual water yield under future climate conditions, more pronounced with eucalyptus/pine scenario with a reduction of 9%, when compared to the baseline climate and land cover. However, if only climate change is considered, the reduction is between -1 and -2%, indicating that the effect of forest cover in the total annual water yield is more important than the climate effect. The reduction in the total water yield is less visible in the agri/vine scenario, and a small synergetic effect is shown in 2041-60. This tendency of reduction is contradictory with the findings from section 4.3.2, in which agriculture increases water quantity. An increase in temperature might improve growing conditions for vineyard in the winter-wet season, therefore increasing the water use (Nunes & Seixas, 2011). Low flows decrease with the eucalyptus/pine scenario for both periods, especially in the summer months of July and August with changes of about -50% when comparing to the baseline (Figure 4.5b). Again, the effect of land cover is more visible than the climate effect. The low flows in agri/vine scenario are similar to the current land cover (Table 4.12). On the other hand, high flows (Q5 Table 4.12) are slightly reduced by climate change combined with the eucalyptus/pine scenario, especially in winter and beginning of spring. This means that forestation could have a positive role in mitigating the peak flows, particularly in 2021-40 (Figure 4.5b). In contrast, high flows are slightly exacerbated with the agri/vine scenario (Table 4.12). Regarding water quality, an increase in NO3-N concentrations and a reduction in N exports are expected for future climate, as it was described in section 4.3.3. This tendency is aggravated under the agri/vine scenario, which showed the greatest NO3-N concentrations and N exports, especially in 2041-2060 (Table 4.12). In fact, nutrient loads are very sensitive to land cover change, since it introduces different quantities of fertilization and organic matter (Panagopoulos et al., 2011). The evaluations of the daily-simulated NO3-N concentrations in the river never exceeded the value of 5.6 mg/l, with exception of 2 days under the agri/vine scenario. Nevertheless, low flows may be aggravated if a different scenario of climate change is considered (e.g. RCP 8.5), which could surpass this water quality threshold value, especially when associated to agri/vine scenario. Soil erosion should be higher when climate is combined with the agri/vine scenario, particularly in 2021-40 when there is more precipitation during winter (Table 4.12). The area suffering erosion rates above a tolerable value grows considerably in agri/vine scenario ! ! 109 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS (2021-40: 34.7% and 2041-60: 33.4%), when compared to eucalyptus/pine and current land cover. Overall, the combined effects of climate and land-cover seem to aggravate the decline in annual runoff from climate change in the eucalyptus/pine scenario (- 9%), with further reductions in the low flows during summer months (-32%; -49%). However, this scenario shows a synergetic effect on peak flows reduction and soil erosion control. This is consistent with tendencies reported in recent studies across the Mediterranean basins using combined effects of climate and re-vegetation scenarios (López-Moreno et al., 2013; Morán-Tejeda et al., 2013). For the agri/vine scenario with future climate conditions, the area under erosion risk should increase, together with a degradation of the water quality (increase in NO3-N concentration). It should be noted that the negative implications of the combined effects on hydrological services provision must be put in perspective, especially by also taking into account the demand side of services. A trade-off analysis between a given service provision with land cover, climate effects and service demand should be evaluated for an integrated analysis, considering flexible strategies for water management at a watershed scale. 5. Conclusion ! SWAT was applied to a medium-sized watershed in northwest Portugal, calibrated and validated against daily discharge, with good agreements between model predictions and field observations. Given the limitation of observed data, calibration of sediments and nitrates can be considered valid, strengthen by a comparison with values from literature in similar regions. The hydrological consequences of four different land cover scenarios were compared to the current land cover scenario. Different land cover options would not compromise the overall provision of hydrological services, but some scenarios can be assigned specifically to improve a certain hydrological service provision. For instance, allowing natural vegetation to recover (oak forest) would lead to a more regular distribution of water flows throughout the year, or allowing forest of eucalyptus/pine would reduce peak flows and soil erosion during winter. A statistical downscaling of four GCMs models, bias-corrected with ground observations was done for the period 2021-40 and 2041-60, using the RCP4.5 medium emission scenario. Results show an increase in temperature ranging from 0.32ºC in spring (2021-40), and 2.0ºC in summer (2041-60). Combined with a decrease in precipitation (-3.9%), which is more pronounced in summer (-25%) for the period 2041-60. Although climate change has slightly less effect in reducing total annual water yield, when compared to land cover, the ! ! 110 CHAPTER 4. SIMULATING THE EFFECTS OF LAND COVER AND FUTURE CLIMATE effect of climate change is significantly higher in further reducing the low flows in summer and increasing soil erosion in winter. The combined effects of climate and land cover change can both enhance or degrade hydrological services provision. The decline in annual runoff from climate change will be aggravated in the eucalyptus/pine scenario, with special emphasis in further reducing the low flows in summer (-16% in 2021-60; and -40% in 2041- 60). In turn, peak flows and soil erosion will be attenuated. On the other hand, annual runoff will be offset with agri/vine scenario, but soil erosion will be highly aggravated. Results emphasize the importance of building adaptation strategies that consider both climate and land cover changes, with special attention for water supply timing during summer, peak flows and soil erosion during winter. Acknowledgments This study was financially supported by FCT (Portuguese Science Foundation), and the European Social Fund through Ph.D grant SFRH/BD/66260/2009 to C. Carvalho- Santos, and post-doctoral grant to J.P. Nunes (SFRH/BPD/87571/2012). Antonio T. Monteiro was supported by the project “Biodiversity, Ecology and Global Change” cofinanced by North Portugal Regional Operational Programme 2007/2013 (ON.2 – O Novo Norte), under the National Strategic Reference Framework (NSRF), through the European Regional Development Fund (ERDF). J. Honrado received financial support from Programa Operacional Fatores de Competitividade (COMPETE/FEDER) and the Portuguese Government / Fundação para a Ciência e a Tecnologia (FCT), through Project Grant PTDC/AAG-MAA/4539/2012 (“IND_CHANGE”). The authors would like to thank Joaquim Alonso, IPVC Ponte de Lima, for providing land use and soil maps; João Gonçalves, CIBIO University of Porto, for providing DEM; and Isabel Poças for providing MOD16A2 ET product from MODIS. ! ! 111 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Chapter 5 From hydrological services to a multifunctional watershed: trade-offs and synergies between biodiversity conservation and forest ecosystem services ! ! Abstract Ensuring forest protection and the delivery of forest ecosystem services is one of the aims from the European Union biodiversity strategy to 2020. Therefore, accurate modelling and mapping of both ecosystem services and biodiversity conservation value is an important effort to support spatial planning and land management options involving forest and ecosystems. In this context, the objectives of this study are: to analyse the provision and dynamics of the multiple ecosystem services under two forest cover change scenarios (oak and eucalyptus/pine) compared to the current scenario of shrubland dominated, at the watershed scale; and to evaluate their possible trade-offs with the biodiversity conservation value. The Vez watershed, in northwest Portugal, was used as case-study area, in which SWAT (Soil and Water Assessment Tool) was applied to simulate the provision of hydrological services (water supply quantity, timing and quality, soil erosion, and flood regulation), as well as biomass and carbon storage services. Biodiversity conservation value was based on nature protection regimes and on expert judgement applied to a land cover map. Results indicated that the current delivery of ecosystem services is higher in the high and low mountain sub-basins, and lower in the valley sub-basin. The overall performance for water quantity and timing is better under the shrubland and oak forest scenarios, when compared to the eucalyptus/pine forest scenario. In turn, the latter has better performance for the provision of flood regulation and erosion control services, especially in the low mountain sub-basin. The current shrubland dominated cover also shows good performance for the control of soil erosion. The oak scenario is the one with less trade-offs between forest services and biodiversity conservation. Even if water quantity is currently not a problem in the Vez watershed and that eucalyptus/pine is the scenario with better results for flood regulation, caution should be taken with this scenario due to its effects on biodiversity conservation and likely also on fire hazard. Results highlight SWAT as an effective tool for modelling and mapping ecosystem services generated at the watershed scale, therefore contributing to improve the options for land management and governance. ! ! Keywords: Biodiversity conservation; Ecosystem services; Forests; Land cover scenarios; SWAT; Vez watershed.! ! ! Carvalho-Santos C., Silva A.R., Gonçalves J., Honrado J. (submitted) From hydrological services to a multifunctional watershed: trade-offs and synergies between biodiversity conservation and forest ecosystem services.! ! ! 112 CHAPTER 5. FROM HYDROLOGICAL SERVICES TO A MULTIFUNCTIONAL WATERSHED ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 113 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 5.1. Introduction ! Watersheds are considered primary units for ecosystem services management and planning (Panagopoulos et al., 2011; Cook & Spray, 2012; Boithias et al., 2014). Forests and forestation are used as a management strategy to improve ecosystem services provision in many watershed around the world (Lele, 2009; Quintero et al., 2009; Onaindia et al., 2013). Financial mechanisms are supporting several forestation actions, either by forest subsidies under agri-environmental schemes, for instance in Europe, or by payments for ecosystem services, mainly applied in tropical countries (Hein et al., 2013). Furthermore, ensuring forest protection and the delivery of forest ecosystem services is one of the aims inscribed in the EU Biodiversity Strategy to 2020 (EU, 2011). However, land cover and land use changes pose major challenges to water resources management and the provision of hydrological services (Kepner et al., 2012; Geneletti, 2013). Therefore, understanding the processes behind forest ecosystem services provision as well as their trade-offs with biodiversity conservation is a useful asset to support spatial planning and land management. In addition, several studies have shown that mapping ecosystem services can be highly useful for informing land-use and management decisions on trade-offs and win-win situations between ecosystem services (Chan et al., 2006; Egoh et al., 2008; 2009; Nelson et al., 2009). Beyond the traditional source of timber and pulp, forests have long been known for providing other services for human well-being, such as carbon sequestration and hydrological services (water supply and water damage mitigation) (Shvidenko et al., 2005; Brauman et al., 2007; Carvalho- Santos et al., 2013). Although the overall balance of forests on services provision is generally positive, their influence on the hydrological processes is complex and often site specific, requiring daily rainfall-runoff modelling for an accurate evaluation (Calder, 2002; Malmer et al., 2010; Crossman et al., 2013). This is especially important regarding the magnitude of decreasing the total annual water yield in the watershed (Bosch & Hewlett, 1982; Brown et al., 2005). In fact, forest plantations are encouraged for carbon sequestration strategies, but the environmental consequences of forestation are often not fully considered, for instance, on the reduction in annual streamflow (Jackson, 2005). After acknowledging the role of forests on hydrological services provision in a given watershed, further analysis can be performed to evaluate the provision of other services (Logsdon & Chaubey, 2013). SWAT (Soil and Water Assessment Tool) is a hydrological modelling tool applied worldwide for the evaluation of the hydrological regime in a watershed (Arnold et al., 1998). Ultimately, the SWAT outputs can also be used to model the provision of other non-hydrological services, such as carbon sequestration and timber production (Logsdon & Chaubey, 2013). Also, the option for different land covers and landscape configurations may induce trade-offs between the provision of multiple ecosystem services and biodiversity conservation (Chisholm, 2010; Villamagna et al., 2013; Wang & Fu, 2013). In this context, modelling and mapping those trade-offs is an important step to ! ! 114 CHAPTER 5. FROM HYDROLOGICAL SERVICES TO A MULTIFUNCTIONAL WATERSHED create sustainable options for forest management (Ninan & Inoue, 2013; Onaindia et al., 2013; Wang & Fu, 2013). The purpose of this study is to map the provision of multiple ecosystem services at the watershed level, using SWAT as the primary modelling tool, and to evaluate their possible trade-offs with biodiversity conservation based on habitat value and protection regimes. These trade-offs were analysed dynamically by comparing forest services provision and the biodiversity conservation value under two distinct forestation scenarios (eucalyptus/pine vs. common oak) compared to the present condition dominated by shrubland. The watershed level of analysis is particularly interesting here, because it constitutes the base unit when all the processes operate, namely the flow of water (Savenije & Van der Zaag, 2008). The research was developed following a four-stage workflow. First, the modelling and mapping of the ecosystem services and biodiversity conservation was developed. Subsequently, the spatial distribution of the ecosystem services and biodiversity conservation value was analysed under the forestation scenarios described before. Then, a spatial correlation was performed to analyse the conflicts or synergies between the provision of ecosystem services and biodiversity conservation. Finally, the results were analysed and discussed in a governance framework, particularly to support land management options for improving the decisionmaking processes for ecosystem services provision and biodiversity protection. 5.2. Methodology 5.2.1. Study-area ! The study area was the watershed of river Vez, a medium-sized watershed (252 Km2) located in the Soajo and Peneda mountains, northwest Portugal (Figure 5.1). The Vez is one of the main tributaries of river Lima, a major river in northwest Iberian Peninsula. Annual precipitation in the watershed varies from 1000 mm/year in lowlands up to 3000 mm/year in highlands (wet years), mainly concentrated in the autumn and winter months (October to March). Topography is complex with elevation ranging from 30 m to 1400 m, and slopes above 25% shaping 58% of the watershed. Granites and locally schist characterize the regional geology, with five major soil types occurring in the watershed: Humic Regosols (67%) and Leptosols (9%) prevail in highlands: Dystric Antrosols (22%), Fluvisols (1%) and Urban (0.56%) in lowlands. Regarding land cover/use, open areas of bare rock and heath occupy the top of the mountains, shrurbland with scattered woodland areas occupy the highlands, and agricultural land associated with forest areas of European oak (Quercus robur), Maritime pine (Pinus pinaster), and Eucalypts (Eucalyptus globulus) are common in lowlands (Figure 5.2a). Approximately one third of the watershed (the mountainous areas and the river Vez itself) is included in the EU Natura 2000 network, and the upper part of the watershed is further included in the Peneda-Gerês National Park (Figure 5.2b). These areas are very important for biodiversity conservation, with the presence of some charismatic species such as the iberian wolf ! ! 115 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS (Canis lupus signatus), roe deer (Capreolus capreolus) and several birds of prey. Phytogeographically the region represents the southwest limit of the Eurosiberian (Atlantic) region, with sub-Mediterranean, deciduous oak forests as the potential vegetation type (Costa et al., 1998; Capelo et al., 2007). ! ! Figure 5. 1 – The study area, Vez watershed, with the location of observed data stations used to setup SWAT model, the range of altitude and a climate plot (yearly average values 1999-2008). ! ! Figure 5. 2 – a) Land cover map of Vez watershed (year 2006); b) Biodiversity protection levels in the Vez watershed. ! ! 116 CHAPTER 5. FROM HYDROLOGICAL SERVICES TO A MULTIFUNCTIONAL WATERSHED 5.2.2. Assessment of ecosystem services provision 5.2.2.1. Hydrological services ! Hydrological services encompass the benefits for people involving water in the ecosystems (Brauman et al., 2007). Following the conceptualization used in a previous study (Carvalho-Santos et al., 2013), the provision of water supply (quantity, timing and quality) and water damage mitigation (flood mitigation and erosion control) were considered for modelling and mapping in the Vez watershed. The accuracy of the hydrological services modelling improves when using dailyrunoff models, such as SWAT (Crossman et al., 2013). SWAT is a deterministic, physically-based and semi-distributed hydrological model used successfully around the world (Arnold & Fohrer, 2005). The watershed is first divided into sub-basins, and after each sub-basin is divided into HRUs (hydrologic response units), which are land areas of varying size, but with the same land cover, soil type and slope class (Gassman et al., 2007). These units are spatially explicit and provide different contributions to the provision of hydrological services. However, SWAT is not a completely distributed model, meaning that inside a given sub-basin, such unique combination of land cover, soil type and slope (HRU) can be found in different not adjacent places. The hydrological contribution of the several areas belonging to a given HRU is considered the same. The use of HRUs and the analysis of their contribution for services provision converges with the ecosystem services providing units (Esp) concept used for ecosystem services analysis and evaluations (Rounsevell et al., 2010). SWAT was applied in the Vez watershed and detailed description of the model setup and parameterization can be found in the previous chapter 4. Here, the watershed was divided into 10 sub-basins with a total of 500 HRUs (320 HRUs for oak and eucalyptus/pine scenario), based on land cover, soil types and slope classes. SWAT was forced with daily climatic data for the watershed (precipitation, maximum and minimum temperature, solar radiation, relative humidity and wind speed) (Table 4.3). Finally, a calibration exercise was performed, in which some parameters were adjusted in order to achieve the best approximation to the reality. A good agreement between model predictions and field observations related with discharge was obtained (Figure 4.2). A comparison between observed sediments and nitrates (NO3) exports as well as with the values presented in the literature was made against SWAT simulated values. Overall, this calibration can be considered adequate given the limitations of observed data (Figure 4.3). The model outputs were analysed at the HRU and sub-basin level, and adapted to each service provision (Table 5.1). Three major sub-basins were used here to facilitate the spatial description of the service provision inside the watershed: high mountain, low mountain and valley (Figure 5.1). As the outputs from water quality, soil erosion control and flood mitigation services do not have a direct reading (i.e. higher values correspond to lower contributions for service provision) there are no units in the respective maps in Figure 5.4. To homogenise the maps interpretation, 1 was divided by the ! ! 123 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 5.3.4. Conservation value under alternative land cover scenarios and trade-offs with ecosystem services The conservation value for biodiversity in the Vez watershed would decrease dramatically under the eucalyptus/pine scenario (Figure 5.4b), but it would increase in the oak scenario (Figure 5.4c) when compared to the current shrubland dominated landscape (Figure 5.4a). In the latter, conservation value has a negative or low spatial correlation with most ecosystem services, with water quantity as the only exception with a high correlation (Table 5.3). Oak scenario has the strongest positive correlations with biodiversity conservation value, except for water quantity and timing. Conversely, implementing the eucalypt/pine scenario would hold advantages for several services (such as erosion control or flood regulation), but at the expense of strong negative effects on biodiversity conservation (Figure 5.4 and Table 5.3). ! Table 5. 3 - Spearman correlation test between the services provision and biodiversity conservation value, in the three land covers. Services Shrub Eucalyptus/pine Oak W_quantity 0.18 0.58 - 0.06 (ns) W_timing - 0.26 (ns) - 0.05 (ns) - 0.22 W_quality 0.09 - 0.25 0.34 S_erosion_contol 0.14 - 0.38 0.50 Flood_reg 0.11 - 0.33 0.08 (ns) Biomass 0.17 0.07 (ns) 0.41 C_storage 0.17 0.07 (ns) 0.41 All the analysis were statistical significant with p-value < 0.0001, with exception of the ones identified with (ns$–$non$significant) 5.4. Discussion 5.4.1. Land cover and ecosystem services in the Vez watershed ! Mapping ecosystem services at the watershed scale, which includes identifying the areas where the provision of a certain service needs to be improved, is an important communication tool and essential for proper management (Hauck et al., 2013). In the Vez watershed, the high mountain sub-basin is the one where the water quantity service is currently provided at the highest levels (Figure 5.5). This pattern is due to the high values of rainfall in the mountain, associated to the water use efficiency of the dominant shrubland vegetation. The oak forestation scenario also exhibits good performance for water quantity, especially if considering the median of the HRU values distribution (Appendix C.1). This is related to the fact that oaks are in dormant state during winter months, decreasing rain interception and transpiration fluxes, allowing water to either infiltrate or run to the rivers as surface runoff (Rosenqvist et al., 2010). In turn, the eucalyptus/pine scenario is the most unfavourable for the water quantity service. This tendency is observed in several studies, for instance in several watersheds of northern Spain, in which exotic species are negatively correlated with water yield (Garmendia et al., 2012). The same pattern applies to the water timing service, although here the most important explaining factor is the soil depth and texture associated to high vegetation cover, which improves water storage capacity and infiltration rates (Li, 2004; ! ! 124 CHAPTER 5. FROM HYDROLOGICAL SERVICES TO A MULTIFUNCTIONAL WATERSHED Ilstedt et al., 2007). The best provision for the water timing takes place at the valley and low mountain sub-basins, due to the presence of deeper soils. In the Vez watershed, water scarcity has not been a problem, even in summer where normally rainfall is scarce and plants need to use water from soil moisture. Nonetheless, different land management strategies should be considered according to their effects on the water quantity and timing especially during the summer months. In fact, an acknowledged problem is the occurrence of floods in wet years, associated to intense rain episodes and the rapid flow of water that is typical from mountainous watersheds. In the Vez watershed, the flood regulation service would be best delivered under eucalyptus/pine forestation scenario. Eucalypt and pine species are associated to a reduction in peak flows, mainly due to their rapid growth rates and the ability of their canopy to intercept rainfall during the rainy season (Robinson et al., 2003). The potential for flood regulation in the low mountain sub-basin would be particularly increased under the eucalyptus/pine scenario (Figure 5.5), suggesting this scenario to be favoured if the options for management would focus on the regulation of peak flows. On the other hand, under the same scenario, the provision of water quantity in the low mountain would decrease, which should be considered in a context of climate change in the Vez watershed (Chapter 4). Soil erosion appears to be best provided under the eucalyptus/pine scenario and the shrubland scenarios (Figure 5.5). In general terms, erosion rates appear to be related to the role played by vegetation in protecting the soil surface (Nunes et al., 2011). However, eucalypt and pines stands should not be intensively managed, otherwise the effect on soil erosion control would be the opposite (Figueiredo et al., 2011). This service was predicted to be less important under oak forestation scenario, as deciduous trees do not protect the soil as effectively during the rainy, cold season (Figure 5.5). It should be noted, however, that SWAT does not consider the structure of understory vegetation, which is usually more complex in native woodlands than in managed planted stands (Nunes et al., 2011). In all cases, fires are an underlying cause of soil erosion and degradation in the Mediterranean region, especially in Portugal where erosion by fires has been a major concern in the last years (Shakesby, 2011; García-Ruiz et al., 2013). Biomass production and carbon storage were predicted as slightly higher under the current shurbland dominated cover (Figure 5.5). However, the type of biomass from forest stands has a higher economic value (e.g. pulp from eucalypt, timber and wood from pines and oaks), which should be regarded as a good asset to improve the landowners’ incomes. Usually, the carbon storage service is considered most important in areas with stable vegetation (Chan et al., 2006). Nevertheless, the long-term carbon sequestration potential of forests can be destroyed by fires, an important carbon emissions component, due to be also influenced by climate change (Sommers et al., 2014). In the Vez watershed, fires have been a constant degradation process, as revealed by the extent of low vegetation areas (scrub and heath), impacting not only the resilience of the biotic ! ! 125 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS communities, but also the carbon storage and the overall provision of ecosystem services (Proença et al., 2010a). A flow of ecosystem services provision can be observed in Vez watershed, especially in the case of hydrological services, since the high and low mountain are the sub-basins where these ecosystem services are best delivered, whereas most of the corresponding benefits are actually felt downstream in the valley. This is in line with the traditional ecosystem services provision knowledge, in which the provision of services can be considered directional, e.g. the regulation of water by forests in the upstream mountains in a watershed resulting in a benefit downstream (Fisher et al., 2009). ! 5.4.2. Trade-offs with biodiversity conservation and implications for land management ! Oak forestation would be the most favourable scenario for biodiversity conservation value, whereas eucalyptus/pine was the least favourable (Figure 5.4). Eucalypt is an exotic tree species widely planted in Portugal for pulp production, and usually associated to low levels of biodiversity in their stands when compared to native stands (Kardell et al., 1986). Plant and bird diversity in the Vez watershed region is highest in oak forests, followed by pine stands and far less in eucalypt plantations (Proença et al., 2010b). This is particularly important inside protected areas, where eucalypt stands, even if contributing to ecosystem services provision, have negative effects on biodiversity, as emphasised in a recent study in Spain (Onaindia et al., 2013). Forest certification requiring the management for biodiversity conservation objectives is also an important action to promote native forests (Thompson et al., 2011). In our hypothetical eucalyptus/pine forestation scenario for the whole watershed, a large proportion of eucalypts and pines would be located inside protected areas. Here, the level of naturalness should be maintained, investing in measures for natural regeneration of oak and avoiding ecosystem degradation related with fires and invasive species (Proença et al., 2010b). Based on the patterns, flows and dynamics of ecosystem services as well as their trade-offs with biodiversity conservation, some recommendations may be suggested regarding the effects of different land cover scenarios on the ecosystem services provision in the Vez watershed. Considering flood regulation was identified as a key challenge, which can be aggravated in the context of climate change, the eucalyptus/pine scenario would appear to be the best option for future management strategies. However, there are three major disadvantages that should be considered: (1) the trade-offs with other services, namely water timing (considered here as the maintenance of summer low flows), which is better provided under the oak and shrubland scenarios; (2) the fire proneness of eucalypt and pine forests (Fernandes, 2012), especially during summer, and the consequences for soil erosion and landscape degradation; and (3) the trade-offs with biodiversity conservation, mainly inside protected areas but also across the watershed. For the latter, conservation strategies considering incentives for the use of native species, especially inside ! ! 126 CHAPTER 5. FROM HYDROLOGICAL SERVICES TO A MULTIFUNCTIONAL WATERSHED protected areas, would be a good option to promote biodiversity and ecosystem services (Proença et al., 2010b; Onaindia et al., 2013). Fire smart-management with intervention on vegetation reducing the flammability and increasing resilience (e.g. using native species) in relation to the fire regime can be pointed as efficient strategies (Fernandes, 2012). For the trade-offs with provision of other ecosystem services, spatial optimization simulations could be developed in which eucalypts and pines can be assigned to areas where flood risk mitigation is strategically more effective, provided that forest management regimes are adequate to promote regulation services. Other types of forests could be assigned to other areas in order to improve other ecosystem services, for instance oaks in protected mountain areas for water timing (also positive for biodiversity conservation). The creation of mosaics of native trees well connected through corridors will improve the resilience and connectivity at a landscape scale (Brockerhoff et al., 2013). This would also contribute to maintain and promote landscape diversity across the watershed, which is known for sustain social-ecological resilience and adaptive capacity in a changing world (Chapin et al., 2010). ! 5.4.3. SWAT as an effective tool for mapping ecosystem services ! ! The inclusion of additional ecosystem services in our framework, using SWAT, is an added value for understanding the processes and services provision using a daily-runoff model. Similar efforts have been done before (Logsdon & Chaubey, 2013), but to our knowledge this is the first time a special emphasis is placed on ecosystem service dynamics under different forestation scenarios, while also considering the biodiversity conservation value. Based on our results and others from previous studies, we advocate that, when accurately calibrated, SWAT can be an important tool for modelling and mapping ecosystem services related to the biophysical processes at the watershed scale. Although it is known as a tool without a central focus on ecosystem services modelling due to a primary module on the hydrological processes and the absence of the demand side of ecosystem service modelling (Vigerstol & Aukema, 2011; Bagstad et al., 2013), it can be used effectively for the biophysical modelling of all hydrological services, and as a complementary tool in more comprehensive modelling approaches. Future land use/cover policies should consider a holistic perspective as the one presented here, considering not only the hydrological/ecological processes with their related implications for ecosystem services provision, but also considering the spatial conflicts with biodiversity conservation value. In this regard, a current limitation of SWAT is the single consideration of a bundle of ecosystem services, namely those related to watershed processes (hydrological services, biomass production, and carbon storage). Nonetheless, SWAT has proved to be an important tool for modelling and mapping the potential provision of ecosystem services at the watershed scale, ! ! 127 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS thereby contributing to analyse and diversify options for land management and watershed governance. 5.5. Conclusions ! SWAT was used to model and map hydrological services and biomass with derived carbon storage at the HRU level, under three land cover scenarios (shrub, oak and eucalyptus/pine). In addition, biodiversity conservation value was mapped to assess the spatial trade-offs between ecosystem services provision. In Vez watershed, the current delivery of ecosystem services is major in the high and low mountain, with less performance in the valley. The high mountain sub-basin is strong in providing water quantity, biomass production and consequent carbon storage. The low mountain is prone to deliver soil erosion control and flood mitigation. When considering the two different forestation scenarios, water quantity decreases considerably in eucalyptus/pine scenario, as well with for water quantity service. Remarkably flood mitigation and soil erosion is improved with eucalyptus/pine scenario in particular in low mountain region. The eucalyptus/pine scenario is the one with more spatial trade-offs with biodiversity value, especially inside protected areas. It seems that different strategies may be suggested for an effective land use planning in Vez watershed. Considering that water quantity is not a problem is Vez watershed and eucalyptus/pine is the scenario with better results for flood regulation, cautions should be taken regarding strategies of conservation (preferably the use of native oak), and fire risk increase. SWAT has been proved to be an important tool for modelling and mapping ecosystem services generated at the watershed scale, thereby contributing to improve the options for land management. Acknowledgments This study was financially supported by FCT (Portuguese Science Foundation) and the European Social Fund through PhD grants: SFRH/BD/66260/2009 (C. Carvalho-Santos) and SFRH/BD/90112/2012 (J. Gonçalves). R. Sousa-Silva was supported by a PhD grant in the framework of the FORBIO Climate project, financed by BRAIN.be, Belgian Research Action through Interdisciplinary research. J. Honrado received support from FCT through project grant PTDC/AAGMAA/4539/2012 (IND_CHANGE). ! ! 128 CHAPTER 5. FROM HYDROLOGICAL SERVICES TO A MULTIFUNCTIONAL WATERSHED ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 129 ANALYZING HYDROLOGICAL SERVICES PROVIDED BY FORESTS Chapter 6 General Discussion and Conclusions ! ! Vez river ! ! 130 CHAPTER 6. GENERAL DISCUSSION AND CONCLUSIONS ! ! 131 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 6.1. General Discussion ! This thesis presents and tests a framework to analyse hydrological services and the role of forests, using a combination of principles and methods of eco-hydrology with concepts and methods from the novel ecosystem services science. The framework encompasses the conceptual foundations from both fields and provides guidelines for evaluating and monitoring hydrological services provision. This chapter discusses the cross-cut topics of the thesis, namely: the newly proposed conceptualization of forest hydrological services; the use of remote sensing to improve the assessment of hydrological services; the assessment of hydrological services based on the application of eco-hydrological modelling frameworks; the development of model-based simulations of hydrological services provision under climate and land use change scenarios; and the adaptive management of hydrological services provision in small watersheds. ! ! 6.1.1. Forest hydrological services in a social-ecological framework ! In the ecosystem services (ES) literature, conceptualizations have been evolving towards organizing and consolidating the inherent complexity of ecosystem services provision and the downstream societal benefits. Examples of this are the proposals by de Groot (1992), Daily (1997), Elmqvist et al. (2009), MA (2003), Haines-Young and Potschin (2012), Wallace (2007) and de Groot et al. (2010a). Attempts to further organize specific levels inside the broad concept of ecosystem services have emerged, for instance, addressing the valuation and the decision making context of ecosystem services (Fisher et al., 2009), the economic valuation (Boyd & Banzhaf, 2007), discussing the spatial scales of analysis and stakeholders (Hein et al., 2006), addressing land management and the provision of ecosystem services (van Oudenhoven et al., 2012), including the more societal components of the the ES cascade model (Spangenberg et al., 2010), different dimensions of value (Jax et al., 2010), poverty alleviation (Fisher et al., 2013), or their importance in environmental impact assessment (EIA) (Honrado et al., 2013). Specifically for this thesis, there were three conceptualizations that served as a basis to develop the conceptual framework (presented in chapter 2): i) the conceptualization of the hydrological services by Brauman et al. (2007), ii) the ES cascade model presented in Haines-Young & Potschin (2010) and further disseminated in the TEEB study (de Groot et al., 2010b); and iii) the DPSIR (Drivers, Pressures, State, Impacts and Responses) model adapted to the ecosystem services concept by Rounsevell et al. (2010). The objective was to create an integrative framework that could be useful to understand the biophysical processes and the functions behind the relations between water and the components of ! ! 132 CHAPTER 6. GENERAL DISCUSSION AND CONCLUSIONS landscapes, with an emphasis for forests, in a concise way and considering it in a socialecological systems context. The motivation was the high importance of promoting forest management as a strategy for the improvement of ecosystem services delivery, associated to the fact that not always the complex relations between forests and water are adequately considered and acknowledged (Crossman et al., 2013). The ultimate goal is to create frameworks addressing specifically a given ecosystem and its related services provision, for instance urban forest services (Dobbs et al., 2011), and now for forest hydrological services (chapter 2). This emphasis on the conceptualization and the identification of reliable and cost-effective indicators also reflected in the subsequent chapters of this thesis, when the review of relevant satellite products was presented (chapter 3), and when the SWAT model was applied to the Vez watershed, considering not only the biophysical processes, but also the drivers and pressures, including scenarios of land use/cover change and future climate conditions (chapter 4). The ecosystem services concept has been proposed as a promising framework for natural resources management, with complementary benefits and/or consequences for the forest sector (Patterson & Coelho, 2009). These benefits are manifold, from timber and pulp production, carbon sequestration, reduction of soil erosion and flood mitigation, to cultural services (Shvidenko et al., 2005). But the consequences of forestation should also be recognized, such as the reduction in local/regional water yield (Wang et al., 2011). In addition, when plantation forests are implemented, an increase of soil erosion may occur (Figueiredo et al., 2011) as well as changes in fire regime (Fernandes, 2012). Furthermore, when the plantation is done with exotic tree species, a reduction of the local biodiversity may be associated, e.g. in eucalypt stands (Onaindia et al., 2013). However, in other cases plantation forests, mostly in the tropics, are acknowledged for their direct benefits for biodiversity conservation, because they reduce the negative impacts on natural forests by offsetting the need to extract resources (Pawson et al., 2013). In chapter 5, two scenarios of forest occupation were used for a spatiotemporal analysis of service provision. It was not explicitly considered whether those scenarios would be plantation or natural regeneration with their respective consequences, because the aim was only to compare the hydrological consequences of different land cover scenarios. However, eucalypt and pine stands are managed for fast growing, production objectives, oriented towards economic revenues, which is often associated to intensive management. In this case, the provision of hydrological services might be changed due to changes in the structure of stands as well as of the plant undergrowth (Bredemeier, 2011). ! ! 139 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS restrictions to eucalyptus water use in regions above 1400 mm of precipitation a year (Dye, 2013). On the other hand, eucalyptus is a fast-growing, exotic tree species, with special regulations hampering its use inside some protected areas (Kardell et al., 1986). Moreover, eucalypt stands are more prone to invasion by non-native plants (Lomba et al., 2011). These issues should be considered, as one third of the whole Vez watershed is under Natura 2000 protection (see Figure 5.2b). 6.1.4.3. Effects of climate and interactions with land cover change ! It is broadly recognized that the patterns of the global climate have been changing in the last decades, and they are projected to change further in the future (Kovats & Valentini, 2014). Between 1947 and 2006, an increase in temperature and a decrease in precipitation in spring and autumn was observed in Portugal, with extreme heavy precipitation being more pronounced in autumn (de Lima et al., 2013). For areas with Mediterranean type of climate, a generalized warming, a reduction in precipitation rates and the occurrence of more extreme rain episodes are expected (Feng et al., 2014). Likewise, projections for northern Spain (Galicia region) forecast an increase in temperatures with marked uneven distribution of annual precipitation, meaning more rain in the autumn but drier springs and summers (Álvarez et al., 2011). This is in line with the simulations presented in chapter 4, in which an increase in temperature and a decrease in precipitation were assumed. Although the projected magnitude of reduction in yearly precipitation is not severe in the Vez watershed (- 3%), the intra-annual changes draw concerns due to a decrease of precipitation in summer and an increase in the autumn/ winter months. The most important climatic parameter in SWAT is precipitation, which drives how much water is entering into the system (Neitsch et al., 2011). A change in precipitation pattern will have a response in terms of the hydrological outcomes. Future climate conditions in the Vez watershed, under a medium emission scenario, is expected to decrease the annual flow, with special incidence on drier low flows and on the intensification of major peak flows, associated to an increase in soil erosion and the concentration of nitrates in the river (see chapter 4). As the Vez watershed is located in a transition zone between the Mediterranean and the temperate, more humid climates, changes are not so severe as the ones expected for southern Portugal (Kilsby et al., 2007; Stigter et al., 2012; Nunes et al., 2013). Nonetheless, adaptive management strategies with the potential to address the social– ecological complexity of ecosystem services flows, supported by a holistic approach involving natural and social sciences with communication with stakeholders, may be required (Allan et al., 2013). For this, alternative water resources, such as reclaimed wastewater for irrigation and increase of the water use efficiency in agriculture, public supply ! ! 140 CHAPTER 6. GENERAL DISCUSSION AND CONCLUSIONS and domestic use can be pointed as adaptation strategies to cope with future changes (Stigter et al., 2012). Taking advantage of the synergies and recognising the trade-offs between future climate and land use/cover is also an effective approach (López-Moreno et al., 2013; Shi et al., 2013; Khoi & Suetsugi, 2014). The hydrological consequences from changing climate conditions may be ameliorated or aggravated under different land cover scenarios (chapter 4). As described before, future climate will reduce the low flows, which will be aggravated with the forestation scenario of eucalyptus/pine. This was also verified in a watershed in Spain, where future climate projections together with reforestation scenario will reduce the annual runoff in about 30%, with consequences for reservoir management (López-Moreno et al., 2013). This should also be considered for future plans on hydropower production in the region, which can be highly threatened in dry years, as was verified in a watershed in Spain (Terrado et al., 2014). The combined effects can offset peak flows during winter for flood mitigation in the presence of forests. This would be the positive asset from a reforested watershed. Future climate will likely increase soil erosion and nitrate concentration, which could be aggravated with a scenario of agricultural expansion, especially in the case of vineyards. The expansion of vineyards has been simulated in the Sordo watershed (Douro basin), with consequent decrease in groundwater quality (Valle Junior et al., 2014) and increase soil losses (Pacheco et al., 2014). The construction of terraces is indicated as a conservation practice to minimize soil erosion (Junior et al., 2014). Overall, considering the combined effects is thus a promising approach for watershed management. This approach is strengthened in the presence of a broader evaluation perspective using the DPSIR framework (see Figure 2.2), and considering stakeholder participation. An interesting ninestep approach for decision-making was presented in a recent study, which includes future climate conditions and the evaluation of various policies with regard to water resource planning and management (Kim & Chung, 2014). From a more technical perspective, it should be emphasised that the estimated impacts of climate projections can be very dependent on the approach used to downscale climate model data (dynamic or statistical). The important thing here is not to use the GCMs as direct input for simulations because it is inappropriate (Xu et al., 2007). In fact, the range in projected change between downscaling approaches can be as large as the range between different climate models (Kovats & Valentini, 2014). In chapter 4, only the new RCP 4.5 scenario was used, based on four GCMs with statistical downscaling, bias-corrected with ground measurements. Yet, this approach could be further strengthened if a dynamical downscaling would be used as it was done in a study in South Korea (Kim et al., 2013). ! ! 141 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 6.1.5. Adaptive management of hydrological services in small watersheds ! Watersheds are vulnerable to on-going and forecasted environmental and socio-ecological changes, and the provision of hydrological services is a key policy area in the Anthropocene (Allan et al., 2013). This implies a trans-disciplinary effort for an effective implementation of the Integrated Water Management and the Ecosystem Based approaches, connecting the environment (physical and biological) and the human well-being (Savenije & Van der Zaag, 2008; Cook & Spray, 2012). Concerns over water problems are particularly emphasised under the current IAHS (International Association of Hydrological Sciences) decade (2013- 2022) of socio-hydrology, identified as “Panta Rhei – Everything Flows”. This concept implies a focus on hydrological systems as a changing interface between environment and society, improving our capability to make predictions for water resources dynamics to support sustainable societal development in a changing environment (Montanari et al., 2013). Monitoring changes and resilience in ecosystems and in watersheds is an essential step for adaptive water management (Pahl-Wostl, 2006; Iglesias et al., 2011). Here, the usefulness of satellite-based products to monitor the components of hydrological services should be highlighted, especially for the climate related parameters, in order to trace the patterns of possible change (e.g. climate, land cover) (see chapter 3). In parallel, groundbased measurements linked to water resources should be maintained, with a special alert for the Portuguese water information system (SNIRH), which has been closed since 2010. With these available datasets, it is possible to characterise and map hydrological services supply and demand to improve the efficiency of ecosystem services flow inventory (Bangash et al., 2013). This is particularly important to further prioritise socially and economically sensitive policies, such as PES or other ecosystem services-oriented incentives (e.g. agri- and silvi-environmental subsidies). Mountains are considered as “water towers” and in Portugal the mountainous watersheds are very important for seasonal water supply (Viviroli et al., 2007). Creating governance mechanisms in the Vez watershed, bringing together ES providers, often located in mountains, and ES beneficiaries would be an exceptional way of testing PES in a small watershed context. This could include promoting ecosystem service provision through forestation actions, with direct payments, or through actions involving people in forest plantations or other initiatives. It is important to get society understanding the ecosystem processes and involved in the ecosystem services chain (García-Nieto et al., 2013). To this end, the results from ecosystem services studies should convey clear messages to resource managers and decision makers (Thompson et al., 2011). ! ! 142 CHAPTER 6. GENERAL DISCUSSION AND CONCLUSIONS 6.2 Conclusions 6.2.1. Overarching conclusions The main objective of the research presented in this thesis was to develop and test conceptual and analytical frameworks for assessing the hydrological ecosystem services provided by forests, to support options for spatial planning and land management. Four studies were presented, each one comprising a different perspective of analysis: (I) a conceptual framework for hydrological services provided by forests, with a regional assessment for northern Portugal; (II) a review of satellite products, with their respective characteristics, potentially useful for hydrological services assessments; (III) an ecohydrological modelling exercise applied to the Vez watershed, simulating the consequences of different land covers and future climate conditions on water resources; and (IV) a spatially explicit simulation of the consequences of different forest cover options for ecosystem services provision and biodiversity conservation. The particular novelty of this work is the integration of ecosystem services science with remote sensing and eco-hydrological tools, with a special focus on ecosystem functioning related to forests and water. The dialog among several fields of knowledge requires an integration of the different perspectives, with well-recognized advantages for the creation of novel outputs and syntheses useful for regional and local spatial planning. The main contributions from this thesis can be summarised as follows: § A conceptual framework for hydrological services provided by forests was developed, based on the ecosystem services cascade model, with a discussion of the relations between forests and the water flows and an illustration of the framework for northern Portugal (water supply and water damage mitigation); § An exhaustive review was provided of the most valuable satellite products as well as a discussion of the many advantages that Earth observation offers for the evaluation, management and monitoring of hydrological services; § An application of the SWAT model to Vez watershed was developed, with calibration of parameters against observed values as well as values presented in the literature and others from satellite products; § Several SWAT simulations were made, combining land cover change scenarios with future climate conditions under the new RCP 4.5 scenario, for the eco-hydrology of the Vez watershed (including all the hydrological ecosystem services presented in the conceptual framework); ! ! 143 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS § A spatially explicit assessment was built upon the outputs of SWAT and considering the trade-offs between ecosystem services and biodiversity value for conservation. Regarding the analytical component of the research developed for this thesis, the following main conclusions can be highlighted: • Earth observations, and especially satellite-based products, offer reliable, relative low-cost, spatially explicit and near-real time data for evaluating the water functions, with possible integration of routines for data assimilation processes in hydrological modelling; • Good agreements were obtained between observed values and data simulated by SWAT in the calibration process. Therefore, SWAT can be applied to watersheds with environmental conditions similar to the Vez watershed. • The scenario analysis with SWAT revealed that the effects of land cover change are in line of what is suggested in the literature, although the magnitude of change is lower. SWAT is generally more sensitive to changes in climate than in land use/cover. Scenario analysis for hydrological services provision is more effective if combined effects of land cover and future climate are considered. 6.2.2. Additional conclusions from the several studies ! The main specific conclusions from each chapter can be summarized as follows, adding to the overarching conclusions presented above: (I) Conceptual framework for hydrological services provided by forests (chapter 2): § A combination of indicators related to system properties, functions, services and benefits is an effective way of analysing the provision and demand of hydrological services as a whole; § The framework illustrated for northern Portugal shows that the delivery of hydrological services and forest distribution are shaped by the different climatic characteristics according to the Atlantic or Mediterranean influences; § The novel framework and indicators can be adapted to other regions for analysing the condition of forest hydrological services; (II) Review of satellite products available (chapter 3): § The assessment of water supply and water damage mitigation services can strongly benefit from satellite-based products, considering the different water reservoirs on Earth; ! ! 144 CHAPTER 6. GENERAL DISCUSSION AND CONCLUSIONS § Satellite products contribute to improve the understanding of the processes and functions behind the provision of hydrological services, on a spatially-explicit and near-real time basis; § Also, eco-hydrological modelling tools may benefit from satellite products, namely as input data, in data assimilation processes and/or in support of calibration parameters. (III) Eco-hydrological modelling exercise applied to the Vez watershed (chapter 4) § A reduction in the total annual water yield is expected with the eucalyptus/pine forestation scenario, in parallel with an attenuation in the peak flows (compared to the current scenario of shrubland); § The water timing service is favoured in the oak scenario; § Soil erosion control and nitrogen concentration are higher in the agriculture scenario, but also favoured in the current shrubland cover and in the eucalyptus/pine scenario; § Water supply timing, with special incidence on drier low flows and flood mitigation (major peak flows), will be affected by future climate conditions, more pronounced for the period 2041-60 (under the RCP 4.5 climate scenario). § Future climate will increase soil erosion in the period 2021-40, mainly during winter months; § Combined effects can offset peak flows during winter in the presence of forests under the eucalyptus/pine forestation scenario; § Low flows in summer will be reduced with future climate and further aggravated with forestation scenarios; § Soil erosion and nitrogen concentration in the river will increase under future climate and may be aggravated by a scenario of agricultural expansion. (IV) Spatially explicit simulation of ecosystem services and biodiversity in the Vez watershed (chapter 5): § The current provision of ecosystem services is more important in the high and low mountain areas than in the valley; § The performance for the provision of water quantity and timing is better under the shrubland and oak scenarios when compared to eucalyptus/pine scenario; § The eucalyptus/pine scenario has higher performance for the provision of flood regulation and erosion control, especially in the low mountain sub-basin, although erosion control is also well performed under the shrubland scenario; § The oak scenario is the one with fewer trade-offs between forest ecosystem services provision and biodiversity conservation. ! ! 145 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS 6.2.3. Recommendations for future research The new framework for analysing hydrological services in relation to forests presented in this thesis provides theoretical guidance on how to analyse hydrological services and, in the case of the Vez watershed, on how to interpret the effects of different land cover and future climate conditions in a watershed where water scarcity is not a problem, but where the water timing and the regulation of peak flows bring strong concerns. It would be interesting to apply and compare the same simulations for future land cover and climate conditions in watersheds with more pronounced Mediterranean climate. Also considering climate and taking into account that only the RCP 4.5 scenario of climate change was considered in chapter 4, it would be of interest to simulate the hydrological consequences of other climate change scenarios from the new CMIP5 report, in particular the more extreme RCP 8.5 scenario. Regarding changes in land use/cover, it would be interesting to measure in the field the biophysical characteristics of wattle (Acacia) species and/or other exotic invasive species in Portugal, to feed the parameters in SWAT. It is unknown what would be the hydrological responses in a watershed where wattles are gaining high expression, as is the case of Vez watershed, especially under climate change. Societal demand of hydrological services was only considered explicitly in chapter 2 and at a regional scale, as our research objective was more oriented towards the effects of land cover change on hydrological services provision (i.e. the more functional, provision side of the cascade). However, it would be interesting to compare our simulations against the local and regional demand for hydrological services provided in the Vez watershed, even if reliable data on water uses are often hard to get. Eco-hydrological modelling needs high spatial and temporal resolution ground-based measurements to have a good performance. It is therefore of extreme importance to foster the monitoring of water related services, especially under future climate conditions, and in particular at the local scales, for effective land planning and management. In Portugal, the hydrological information system (SNIRH) has recently stopped the measurements of both the instant level height of some rivers and the climatic parameters from the more local meteorological stations. This means that even calibrated models, such as the one applied to Vez in this thesis cannot be correctly fed for future simulations, because of the lack of precipitation and other climatic variables. Finally, regarding the importance of the satellite products for water resources monitoring, future research should encompass: (i) the use of satellite products to feed SWAT models at the local scale; and (ii) the testing of proxy indicators of the water functioning in a watershed, derived from satellite products, which could complement or replace the information that is not recorded by national hydrological measurement systems. ! ! 146 REFERENCES ! ! ! 147 ANALYSING HYDROLOGICAL SERVICES PROVIDED BY FORESTS References Abbaspour, K. C., 2013. SWAT-CUP 2012: SWAT Calibration and Uncertainity Programs - A user manual. 1– 103. Agroconsultores & Geometral, 1995. Carta dos Solos e Carta de Aptidão da Terra de Entre Douro e Minho (Soil map and capability of soil map). Direcção Regional de Agricultura Entre-Douro-e-Minho [in Portuguese] Albergel, Zakharova, Calvet, Zribi, Parde, Wigneron, Novello, Kerr, Mialon, Fritz, 2011. A first assessment of the SMOS data in southwestern France using in situ and airborne soil moisture estimates: The CAROLS airborne campaign. Remote Sensing of Environment 115: 11–11. Alcantara, E. H., J. L. Stech, J. A. Lorenzzetti, M.-P. Bonnet, X. Casamitjana, A. T. Assireu, & E. M. L. de Moraes Novo, 2010. Remote sensing of water surface temperature and heat flux over a tropical hydroelectric reservoir. Remote Sensing of Environment 114: 2651–2665. Allan, C., J. Xia, & C. Pahl-Wostl, 2013. Climate change and water security: challenges for adaptive water management. Current Opinion in Environmental Sustainability 5: 625–632. Alsdorf, D. E., & D. P. Lettenmaier, 2003. Geophysics. Tracking fresh water from space. Science 301: 1491– 1494. Ansink, E., L. Hein, & K. P. Hasund, 2008. To Value Functions or Services? An Analysis of Ecosystem Valuation Approaches. Environmental Values 17: 489–503. Aranda, I., A. Forner, B. Cuesta, & F. Valladares, 2012. Species-specific water use by forest tree species: From the tree to the stand. Agricultural Water Management 114: 67–77. Arnold, J. G., & N. Fohrer, 2005. SWAT2000: current capabilities and research opportunities in applied watershed modelling. Hydrological Processes 19: 563–572. Arnold, J. G., R. Srinivasan, R. S. Muttiah, & J. R. Williams, 1998. Large area hydrologic modeling and assessment part I: model development. Journal of the American Water Resources Association 34: 73–89. Arnold, J. G., R. Srinivasan, W. E. B. Haney, & S. L. Neitsch, 2011. SWAT Soil and Water Assessment Tool - input/output file documentation version 2009. 1–662. Ayanu, Y. Z., C. Conrad, T. Nauss, M. Wegmann, & T. Koellner, 2012. Quantifying and Mapping Ecosystem Services Supplies and Demands: A Review of Remote Sensing Applications. Environmental Science & Technology 46: 8529–8541. Álvarez, V., J. J. Taboada, & M. N. Lorenzo, 2011. Cambio climático en Galicia en el siglo XXI: Tendencias y variabilidad en temperaturas y precipitaciones (Climate Change in Galicia by the XXI century: tendencies and variability in temperatures and precipitation). Revista Avances en Ciencias de la Terra (ACT) Vigo 2: 65–85. Bagstad, K. J., D. J. Semmens, S. Waage, & R. Winthrop, 2013. A comparative assessment of decision-support tools for ecosystem services quantification and valuation. Ecosystem Services 5: 27–39. Baker, T. J., & S. N. Miller, 2013. Using the Soil and Water Assessment Tool (SWAT) to assess land use impact on water resources in an East African watershed. Journal of hydrology 486: 100–111. Bakker, M. M., G. Govers, C. Kosmas, V. Vanacker, K. V. Oost, & M. Rounsevell, 2005. Soil erosion as a driver of land-use change. Agriculture, Ecosystems and Environment 105: 467–481. Balvanera, P., A. B. Pfisterer, N. Buchmann, J.-S. He, T. Nakashizuka, D. Raffaelli, & B. Schmid, 2006. Quantifying the evidence for biodiversity effects on ecosystem functioning and services. Ecology Letters 9: 1146–1156. Bangash, R. F., A. Passuello, M. Sanchez-Canales, M. Terrado, A. López, F. J. Elorza, G. Ziv, V. Acuña, & M. ! ! 148 REFERENCES Schuhmacher, 2013. Ecosystem services in Mediterranean river basin: Climate change impact on water provisioning and erosion control. Science of the Total Environment 458-460: 246–255. Barbedo, J., M. Miguez, D. van der Horst, & M. Marins, 2014. Enhancing ecosystem services for flood mitigation: a conservation strategy for peri-urban landscapes? Ecology and Society 19: art54. Barbier, S., P. Balandier, & F. Gosselin, 2009. Influence of several tree traits on rainfall partitioning in temperate and boreal forests: a review. Annals of Forest Science 66: 602–602. Bartalis, Z., W. Wagner, V. Naeimi, S. Hasenauer, K. Scipal, H. Bonekamp, J. Figa, & C. Anderson, 2007. Initial soil moisture retrievals from the METOP-A Advanced Scatterometer (ASCAT). Geophys. Res. Lett. AGU 34: L20401. Bastiaanssen, W., H. Pelgrum, J. Wang, Y. Ma, J. F. Moreno, G. J. Roerink, & T. Van der Wal, 1998. A remote sensing surface energy balance algorithm for land (SEBAL).: Part 2: Validation. Journal of hydrology 212: 213–229. Becker, M. W., 2006. Potential for Satellite Remote Sensing of Ground Water. Ground Water 44: 306–318. Bjerklie, D. M., D. Moller, L. C. Smith, & S. L. Dingman, 2005. Estimating discharge in rivers using remotely sensed hydraulic information. Journal of hydrology 309: 191–209. Bjerklie, D. M., S. Lawrence Dingman, C. J. Vorosmarty, C. H. Bolster, & R. G. Congalton, 2003. Evaluating the potential for measuring river discharge from space. Journal of hydrology 278: 17–38. Boithias, L., V. Acuña, L. Vergoñós, G. Ziv, R. Marcé, & S. Sabater, 2014. Assessment of the water supply:demand ratios in a Mediterranean basin under different global change scenarios and mitigation alternatives. The Science of the total environment 470-471: 567–577. Bonell, M., & L. A. Bruijnzeel, 2005. Forests, water and people in the humid tropics: past, present and future hydrological research for integrated land and water management. Cambridge University Press. Bosch, J. M., & J. D. Hewlett, 1982. A review of catchment experiments to determine the effect of vegetation changes on water yield and evapotranspiration. Journal of hydrology 55: 3–23. Boyd, J., & S. Banzhaf, 2007. What are ecosystem services? The need for standardized environmental accounting units. Ecological Economics 63: 616–626. Bradshaw, C. J. A., N. S. Sodhi, K. S. H. PEH, & B. W. Brook, 2007. Global evidence that deforestation amplifies flood risk and severity in the developing world. Global Change Biology 13: 2379–2395. Brauman, K. A., D. L. Freyberg, & G. C. Daily, 2012. Land cover effects on groundwater recharge in the tropics: ecohydrologic mechanisms. Ecohydrology 5: 435–444. Brauman, K. A., G. C. Daily, T. K. Duarte, & H. A. Mooney, 2007. The nature and value of ecosystem services: an overview highlighting hydrologic services. Annu. Rev. Environ. Resour. Annual Reviews 32: 67–98. Bredemeier, M., 2011. Forest, climate and water issues in Europe. Ecohydrology 4: 159–167. Breuer, L., K. Eckhardt, & H.-G. Frede, 2003. Plant parameter values for models in temperate climates. Ecological Modelling 169: 237–293. Brockerhoff, E. G., H. Jactel, J. A. Parrotta, & S. F. Ferraz, 2013. Role of eucalypt and other planted forests in biodiversity conservation and the provision of biodiversity-related ecosystem services. Forest Ecology and Management 301: 43–50. Brouwer, R., A. Tesfaye, & p. Pauw, 2011. Meta-analysis of institutional-economic factors explaining the environmental performance of payments for watershed services. Environmental Conservation 38: 380–392.