Calibration of the Land Uses submodule of the WILIAM-TERRA model

Calibration of the Land Uses submodule of the WILIAM-TERRA model Margarita Mediavilla, February 2025 Contents: 1. Land use change models 2. General description of the Land Uses submodule of WILIAM-TERRA 3. Data sources 4. Calibration of the Lad Uses submodule 1. Land use change models Most simulation models of land use changes are based on at least one of the following four core principles of land use changes (LUC) [1], not mutually exclusive: Continuation of historical development. Future land use can be predicted by land use historical changes. Suitability of land. Suitability covers different aspects, from e.g., maximum market profit (economic suitability) to soil suitability (biophysical suitability) for different uses. Neighbourhood interaction. The probability of transition from one use of land to another is dependent on the biophysical or socio-economic drivers conditioning the LU of its surrounding cells. Actor interaction. Land use change is the result of interaction between actors according with different socio-economic and political drivers. It is a common practise for modellers to describe the processes of LUC according to a particular mechanism that can be used to characterise these changes. Main mechanisms found in literature are: cellular automata, statistical analysis, markov chains, artificial neural networks, economic-based models and agent-based systems. This mechanism of LUC is then codified into algorithms, which lead to different computer simulation models. A huge diversity of model approaches can be found within literature. From the point of view of the spatial disaggregation, land use models can be classified as spatial versos non-spatial models: Spatial models. Spatial models (also called Geographical models) aim at spatially explicit representations of land-use change (LUC) at some level of spatial detail (pixels in a raster model, or other units –administrative, ecological, etc.- in a vector model). They are often associated with Geographical Information Systems (GIS). Non-spatial models. Non-spatial models focus on modelling land-use change without specific consideration for its spatial distribution, frequently economic landuse models. In [2, 3, 4] models are classified in four types: I. Geographical land-use models. These models allocate land area or land demand based on biophysical and socioeconomic properties, and the resulting suitability of land for a specific use. II. Economic land-use models. Models that use demand and supply functions as the main drivers of land-use change, giving total areas of specific land-use types within defined geographical regions. III. Integrated land-use models. These models combine natural and human subsystems. In most cases, these models consist of a combination of separate economic and environmental processes capable of spatially explicit modelling, typically at large (global) scales. IV. Other type of models. Here, the classification includes Urban growth, Machine learning and agent-based models. There is a myriad of LUC models, many of them currently used in IAM models, some of the most popular and relevant ones are: 1. The CLUE Framework. Set of models evolved from the original CLUE model [5]. It simulates LUC using empirically quantified relations between land use and several driving factors in combination with dynamic land competition allocated in a raster based system. The extrapolation of trends in land use change is a common technique to calculate land use requirements but, these trends can be corrected for changes in population growth and/or diminishing land resources. It is a spatially explicit model and can be freely downloaded from the model website. 2. The IMAGE modelling [6] is an ecological-environmental model framework that simulates the environmental consequences of human activities worldwide. It represents interactions between society, the biosphere and the climate system to assess sustainability issues such as climate change, biodiversity and human well-being. Is a spatially explicit model and uses regression-based suitability assessment to determine future land-use patterns. IMAGE has had several versions and has been integrated with other models such as the vegetation growth model LPJmL [7], the CLUMondo [8] for more precise LU representation. 3. The MAgPIE model is a global land-use allocation model [9] which is connected to the grid-based dynamic vegetation model LPJmL, with a spatial resolution of 0.5°x0.5°. It takes regional economic conditions such as demand for agricultural commodities, technological development and production costs as well as spatially explicit data on potential crop yields, land and water constraints (from LPJmL) into account. Based on these, the model derives specific land use patterns, yields and total costs of agricultural production for each grid cell. 4. GCAM [10,11] is a recursive -dynamics partial-equilibrium global IAM that represents the interactions between energy, water, agriculture and land use, economy, and climate. It is a dynamic-recursive model with technology representations of the economy, energy sector, land use and water linked to a climate model that can be used to explore climate change mitigation policies including carbon taxes, carbon trading, regulations and accelerated deployment of energy technology. It is not based on gridded data but the agriculture and land module uses more than 300 subregions and approximately a dozen types of land covers. 5. GLOBIOM [12,13] is a land use model that works with the recursive-dynamic partial-equilibrium MESSAGE model designed to address various LUC related topics (bioenergy policy impacts, deforestation dynamics, climate change adaptation and mitigation from agriculture, long-term agricultural prospect). It is a partial equilibrium economic model that optimizes an objective function defined as the sum of producer and consumer surpluses under a certain number of constraints. LUC are based on a spatially explicit gid based framework. Most of these models are spatially explicit (all except GCAM). This spatial representation allows them to use very detailed information about the physical suitability of land use changes, but is computationally intensive. These models have a structure that is mainly based on linear flows of information, as described in Figure 1. In a linear flow, information about demand, suitability, and the socio-economic factors that drive land use change is provided in a priori scenarios generated by other models (or parts of the same model), and the model calculates land use changes based on optimization or recursive algorithms. LUCs are then used to provide information on crops or energy production, emissions, and other types of environmental or social impacts. Although some models include feedbacks, grid based data, recursive algorithms and optimizations are not the best tools for representing feedback-rich models with strong interactions, for which system dynamics simulations are the most appropriate tool. These models use economic drivers for land use changes such as agricultural prices, income, price elasticities or relative land profitability [15]. On the other hand, the most common policies applied are detailed decarbonization policies such as carbon taxes, subsidies, agricultural quotas or land protection policies. Figure 1: lineal information flows of LUC models 2. General description of the Land Uses submodule of WILIAM-TERRA WILIAM model is a System Dynamics feedback-rich model that addresses the biophysical limits of the energy transitions, and its spatial scale is global with a division in 9 large regions. Economic indicators such as prices or elasticities are hardly reliable at this level of aggregation, while the huge cultural and sociopolitical differences between world regions make it very difficult estimate the effect of detailed decarbonization policies. This is the reason why the approach of WILIAM-TERRA differs from that of other IAMS. The policies used in WILIAN-TERRA are not detailed political measures but physical outcomes that can be derived from all kinds of government measures or social changes (similar to those in the World 3 model [16]). Land use changes are driven by the continuation of observed trends and some basic demands plus the application of a wide range of policies. Thus, it is a policy evaluation model, not aimed at predicting the future, but at analysing the dynamic effects and interactions of a wide range of policies. WILIAM-TERRA can be classified as a model of continuation of historical development with limits to land expansion set by the land suitability and some features of actor interaction. It is a non-spatial model (since the very detailed grid-based models are hardly compatible with system dynamics) and an Integrated model that combines human and natural interactions. It is not an economic model, since it does not use demand and supply functions as the main drivers of land-use changes since their authors do not believe that reliable data for supply, demand and prices can be found to calibrate these exchanges for a global model at the level of aggregation used. The structure and submodules of WILIAM-TERRA are shown in Figure 2. It incorporates a wide range of policies (in pink in the diagram of Figure 2), including dietary changes, land use changes and land protection, livestock manure management, afforestation, urban density, transition to sustainable agriculture and to industrial agriculture, forest exploitation, crops allocation between regions and uses, and carbon capture in grassland soils. This document describes only the calibration and validation of the Land Uses submodule of this WILIAM-TERRA. Figure 2. WILIAM-TERRA module and its connection with the rest of WILIAM model modules. White-green boxes are submodules of WILIAM-TERRA, boxes in other colour belong to other modules of WILIAM. Variables in pink are exogenous policies chosen by the user. The WILIAM-TERRA Land Uses submodule is in charge of allocating the land among 12 uses. The demands of all uses are comprised in a vector named Vector of land use change demands, and it is generated by adding two components: Historical trends of land use changes, which are estimated using lineal approximations over the period from 2005 to 2019 (Source FAO). Land use changes driven by various demands: ourban expansion (driven by population growth) osolar energy (driven by the demand of solar electricity) ocropland loss due to sea level rise opolicies of land demand such as reforestations and land protection onew cropland (driven by the global physical shortage of crops) The competition between the demand of different uses takes place within a dynamic of “all against all” competition in which all uses have the same priority when it comes to demand from others and only land for solar energy and cropland have parameters that allow prioritizing their use over the rest. The expansion of land uses must be obtained from other land uses in order to ensure the physical coherence of the land allocation. This is specified in the matrix of land use change demands (Eq. 1), that describes the demand of changes from land use to another land use : (1) Where, , represents the vector of land use change demand by region and land type ; and represents the share of land use that is obtained from use . The are constant matrices. The land use changes demanded might not be fulfilled if policies of land use protection are activated. is transformed into a in which those land use changes that are not compatible with the physical boundaries or with the boundaries imposed by the user's policies are discarded. The is collapsed into a by adding the changes that are given to each use and subtracting the ones that demand from it: (2) The loss of agricultural land due to sea level rise is subtracted to this vector. And this loss is determined in our module by adapting the method reported Roson & Sartori [3] to WILIAM-TERRA regions and driven by the temperature change received from the WILIAM Climate module. Finally, the is calculated as the integral of although, the module only integrates some of the uses in the stock of and excludes wetlands, snow, ice and waterbodies and shrubland area. These land uses are not calculated via the because they are not driven directly by the policies of the rest of the module and, at present stage, are left constant. 3. Data sources The Land Uses submodule is mainly based on land use data from FAOSTAT and land cover data from the same source, trying to maintain the consistency of these sources although relevant discrepancies are found between them. Data sources are detailed in Table 1. Some of the WILIAM categories come from “land uses” categories and others from “land cover”. SHRUBLAND and OTHER LAND are calculated using a mix of land uses and land cover. The categories CROPLAND_RAINFED, CROPLAND IRRIGATED, FOREST MANAGED, FOREST PRIMARY, FOREST PLANTATIONS and GRASSLANDS are taken from FAO “land use”. URBAN, SNOW-ICE-WATERBODIES and WETLAND are obtained from “land cover” data. SHRUBLAND and OTHER LAND (basically bare areas) are adapted, since taking them from land cover creates incoherences (the sum of all categories is greater or smaller than the total area in some cases, for example). In order to avoid those incoherences, all the uses except SHRUBLAND and OTHER LAND are subtracted from total land and the resulting are is divided between SHRUBLAND and OTHER LAND on the bases of the share obtained with the data of land cover. Table 2 describes the land use FAO categories. Table 3 describes the FAO land uses and Table 4 the mix of both sources of information used for the categories of WILIAMTERRA model. The numbers beside the description correspond to FAO codes [18]. As pointed out by Tubiello et al. [19] there are big discrepancies between land use measures of different sources including satellite data, therefore FAO database has been used as the standard data despite these incoherences. All the FAO data has been revised to check those years when countries do not report and the data that appears in tables in zero. In those cases, the data has been interpolated. The historical values of land use area are shown in table 5 and the correspondence of WILIAM regions and countries is in table 6. Table 1: data sources of the Land Uses submodule Table 2. “Land use” FAO categories L. temporary crops 6630 L. temporary meadows and pastures 6633 L. temporary fallow 6640 L. permanent crops 6650 L. permanent meadows and pastures cultivated 6656 L. permanent meadows and pastures naturally growing 6659 Protective cover (buildings in agricultura land) 6649 primary forest 6714 naturally regenerated forest 6717 planted forest 6716 Inland waters 6680 Coastal waters 6773 arable land 6621 L. permanent meadows and pastures 6655 forest land 6646 water bodies total area Land area 6601 agriculture 6602 agricultural land 6610 cropland 6620 Land use by category thousand ha Food and Agriculture Organization of the United Nations (FAO), Statistics Division (ESS), Environment Statistics team http://www.fao.org/ faostat/en/#data/RL Land cover by land cover class thousand ha Food and Agriculture Organization of the United Nations (FAO), Statistics Division (ESS), Environment Statistics team http://www.fao.org/ faostat/en/#data/RL Table 6: correspondence of countries and WILIAM regions COUNTRY REGION COUNTRY REGION Austria EU27 Rest of Oceania LROW Belgium EU27 Mongolia LROW Bulgaria EU27 Rest of East Asia LROW Croatia EU27 Rest south East Asia LROW Cyprus EU27 Bangladesh LROW Czech Republic EU27 Pakistan LROW Denmark EU27 Shri Lanka LROW Estonia EU27 Rest South Asia LROW Finland EU27 Rest of North America LROW France EU27 Ecuador LROW Germany EU27 Paraguay LROW Greece EU27 Uruguay LROW Hungary EU27 Venezuela LROW Ireland EU27 Rest South America LROW Italy EU27 Guatemala LROW Latvia EU27 Honduras LROW Lithuania EU27 Nicaragua LROW Luxembourg EU27 El salvador LROW Malta EU27 Panama LROW Netherlands EU27 Rest central America LROW Poland EU27 Republica dominicaDa LROW Portugal EU27 Jamaica LROW Romania EU27 Puerto Rico LROW Slovakia EU27 Trinidad y Tobago LROW Slovenia EU27 rest Caribe LROW Spain EU27 Norway LROW Sweden EU27 Rest of EFTA LROW United Kingdom EU27 Albania LROW Canada USMCA Ucrania LROW Mexico USMCA Rrest of eastern europe LROW United States USMCA Georgia LROW Argentina LATAM Iran LROW Brazil LATAM Israel LROW Chile LATAM Jordania LROW Colombia LATAM Kuwait LROW Costa Rica LATAM Oman LROW Peru LATAM Qatar LROW China (People's Republic of) China Saudi Arabia LROW Taiwan Turkey LROW Hong Kong SAR China United Arab Emirates LROW India India Rest wester Asia LROW Russian Federation Russia Egipt LROW Australia EASOC Marocco LROW Brunei Darussalam EASOC Tunisia LROW Cambodia EASOC Rest north Africa LROW Chinese Taipei EASOC Benin LROW Indonesia EASOC Burkina faso LROW Japan EASOC Camerun LROW Korea EASOC C'ote D'Ivoire LROW Malaysia EASOC Ghana LROW New Zealand EASOC Ginea LROW Philippines EASOC Nigeria LROW Singapore EASOC Senegal LROW Thailand EASOC Togo LROW Viet Nam EASOC Rest west Africa LROW Rest central Africa LROW Rest south central Africa LROW Etiopia LROW Kenya LROW Madagascar LROW Malawi LROW Mauritius LROW Mozambique LROW Rwanda LROW Tanzania LROW Uganda LROW Zambia LROW Zimbawe LROW aafricaRest East Africa LROW Botwana LROW Namibia LROW South Africa LROW Rest of south Africa cu LROW Rest LROW 4. Calibration of the Lad Uses submodule This section describes the obtention of the trends of the vector of trends of and the matrices of shares of land uses from other, , described in previous section based on historical data and model calibration. The model is based on the hypothesis that there are some land use changes that are driven by demands, since they are economically or socially interesting (croplands, forests, grasslands, solar land, urban, etc.) and other that are not demanded and only absorb the demand of the rest (other land and shrubland). In any case, all the land demands compete with each other and absorb the demand of other uses. Trend demands are calculated on the basis of historical land use trends, and in some cases have been adjusted to take account of evident changes in trends that cannot be extrapolated into the future (such as the sharp loss of agricultural land in the EU in recent decades due to agricultural policies, which does not appear to be continuing). In future releases of the model, a GIS-based analysis is planned to be used to determine based on historical data, the real shares of land use from other. This would determine what have really been the actual flows of land from one use to another and improve a lot the calibration of this model. In the meantime, this adjustment aims to stablish the most relevant trends of past land use changes for the most relevant uses. It is assumed that the primary forest cannot be increased, since it is defined as very mature forests whose creation goes back to centuries ago. When forest primary increases in the historical data, we assume it is due to changes in definition and assume the greatest value as the initial one. Solar land is the land under photovoltaic and concentrated solar power electricity appliances, since its historical values are very low, we do not take it into account in the calibration. For solar land, the initial shares have been obtained applying Geographic Information Systems (GIS) techniques analyzing the allocation of current solar power capacity. This analysis has been done for each of the 9 regions of WILIAM-TERRA module and it is based on data processed from the “Global Database of Power Plants” combined with land cover data (see [22] for a complete description). These hypothesis of land use trends are used to calculate the land use changes taken from other uses in each simulation time step according to equation 1 using initial values of the matrices of shares of land uses from other, ( ) obtained by the analysis of the literature described in [20, 21] (see Table 7) and the resulting land use changes are confronted to historical data. The discrepancy between estimated and historical data is used to accommodate the matrix . An initial computer calibration of these shares was done with Vensim Software calibration tools, but the final adjustment was made by hand, since the complexity of the task made automatic calibration worse than the human-made. The main efforts have been dedicated to the calibration of the most relevant and conflictive uses (croplands and forests), therefore the errors accumulate in shrubland and other land, whose historical data was not properly found (as described in section 3). Snow, ice and waterbodies and wetlands have not been calibrated at this stage of the model and they are left constant in the model. Table 7: Initial shares of land use changes from other as stated in Campano 2021 [21] INITIAL_SHARE_OF_CROPLAND_RAINFED_FORM_OTHER_LANDS_BY_REGION (REGIONS_I,LANDS_I) LANDS_I RAINFED IRRIGATED FOREST_M ANAGED FOREST_P RIMARY FOREST_P LANTATION S SHRUBLAN D GRASSLAN D WETLAND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_LA ND REGIONS_I [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] EU27 0 0 0.8 0 0 0.12 0.06 0 0 0 0 0.02 UK 0 0 0.8 0 0 0.12 0.06 0 0 0 0 0.02 CHINA 0 0 0.44 0 0 0.16 0.25 0 0 0 0 0.16 EASOC 0 0 0.19 0.66 0 0.15 0.01 0 0 0 0 0.01 INDIA 0 0 0.18 0.3 0 0.24 0.18 0 0 0 0 0.11 LATAM 0 0 0.18 0.63 0 0.18 0.01 0 0 0 0 0 RUSSIA 0 0 0.2 0 0 0.52 0.25 0 0 0 0 0.04 USMCA 0 0 0.11 0.38 0 0.23 0.27 0 0 0 0 0 LROW 0 0 0.18 0.28 0 0.38 0.1 0 0 0 0 0.07 INITIAL_SHARE_OF_GRASSLAND_FORM_OTHER_LANDS_BY_REGION (REGIONS_I,LANDS_I) pondremos que grassland no tiene demanda salvo en LATAM LANDS_I RAINFED IRRIGATED FOREST_M ANAGED FOREST_P RIMARY FOREST_P LANTATION S SHRUBLAN D GRASSLAN D WETLAND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_LA ND REGIONS_I [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] EU27 0 0 0 0 0 0 0 0 0 0 0 0 UK 000000000000 CHINA 0 0 0 0 0 0 0 0 0 0 0 0 EASOC 0 0 0 0 0 0 0 0 0 0 0 0 INDIA 0 0 0 0 0 0 0 0 0 0 0 0 LATAM 0.34 0 0.12 0.44 0 0.04 0 0 0 0 0 0.05 RUSSIA 0 0 0 0 0 0 0 0 0 0 0 0 USMCA 0 0 0 0 0 0 0 0 0 0 0 0 LROW 0 0 0 0 0 0 0 0 0 0 0 0 INITIAL_SHARE_OF_FOREST_PLANTATIONS_FORM_OTHER_LANDS_BY_REGION (REGIONS_I,LANDS_I) LANDS_I RAINFED IRRIGATED FOREST_M ANAGED FOREST_P RIMARY FOREST_P LANTATION S SHRUBLAN D GRASSLAN D WETLAND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_LA ND REGIONS_I [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] EU27 0.23 0 0.61 0 0 0.1 0.6 0 0 0 0 0 UK 0.23 0 0.61 0 0 0.1 0.6 0 0 0 0 0 CHINA 0.23 0 0.61 0 0 0.1 0.6 0 0 0 0 0 EASOC 0.23 0 0.61 0 0 0.1 0.6 0 0 0 0 0 INDIA 0.23 0 0.61 0 0 0.1 0.6 0 0 0 0 0 LATAM 0.23 0 0.61 0 0 0.1 0.6 0 0 0 0 0 RUSSIA 0.23 0 0.61 0 0 0.1 0.6 0 0 0 0 0 USMCA 0.23 0 0.61 0 0 0.1 0.6 0 0 0 0 0 LROW 0.23 0 0.61 0 0 0.1 0.6 0 0 0 0 0 INITIAL_SHARE_OF_NEW_URBAN_FORM_OTHER_LANDS_BY_REGION (REGIONS_I,LANDS_I) LANDS_I RAINFED IRRIGATED FOREST_M ANAGED FOREST_P RIMARY FOREST_P LANTATION S SHRUBLAN D GRASSLAN D WETLAND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_LA ND REGIONS_I [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] EU27 0.75 0 0.08 0 0 0.04 0.06 0 0 0 0 0.06 UK 0.75 0 0.08 0 0 0.04 0.06 0 0 0 0 0.06 CHINA 0.76 0 0.03 0 0 0.06 0.14 0 0 0 0 0.02 EASOC 0.82950502 0 0.06475246 0 0 0.06306926 0.01297029 0 0 0 0 0.03 INDIA 0.84 0 0.03 0 0 0.07 0.05 0 0 0 0 0.01 LATAM 0.45 0 0.11 0 0 0.35 0.08 0 0 0 0 0.02 RUSSIA 0.67 0 0.08 0 0 0.12 0.09 0 0 0 0 0.04 USMCA 0.40465181 0 0.17046426 0 0 0.24418755 0.16313836 0 0 0 0 0.01244197 LROW 0.53574826 0 0.09093677 0 0 0.19739033 0.06602194 0 0 0 0 0.10978433 INITIAL SHARE_OF_NEW_SOLAR_FORM_OTHER_LANDS_BY_REGION (REGIONS_I,LANDS_I) LANDS_I RAINFED IRRIGATED FOREST_M ANAGED FOREST_P RIMARY FOREST_P LANTATION S SHRUBLAN D GRASSLAN D WETLAND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_LA ND REGIONS_I [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] EU27 0.125 0 0 0 0 0.125 0.125 0 0 0 0 0.625 UK 0.125 0 0 0 0 0.125 0.125 0 0 0 0 0.625 CHINA 0.125 0 0 0 0 0.125 0.125 0 0 0 0 0.625 EASOC 0.125 0 0 0 0 0.125 0.125 0 0 0 0 0.625 INDIA 0.125 0 0 0 0 0.125 0.125 0 0 0 0 0.625 LATAM 0.125 0 0 0 0 0.125 0.125 0 0 0 0 0.625 RUSSIA 0.125 0 0 0 0 0.125 0.125 0 0 0 0 0.625 USMCA 0.125 0 0 0 0 0.125 0.125 0 0 0 0 0.625 LROW 0.125 0 0 0 0 0.125 0.125 0 0 0 0 0.625 EU27 In Table 8 one can see the historical trends of land use change in EU. EU27 has had a decrease of rainfed cropland that shows a stagnation in the last years and a similar growth of irrigated cropland that have been maintained. Forest primary grows in the historical data and has been accommodated to be zero, as explained in previous section. Shrubland, snow ice and waterbodies and other land are assumed to have no demand. The historical demand of plantations and urban is maintained. Managed forest demand is set equal to the value of annual deforestation recorded in FAO data. Grassland shows a significant loss that is coherent with the abandonment of extensive farming seen in the EU and is maintained with a small increase to adjust the rest of the uses. Table 9 shows the calibrated shares. The error between the historical and the simulated land use areas after the calibration are shown in Figure 3. The average error is less than 0.4% and, although some land uses such as cropland rainfed and forest managed reach 4% in some years, this result is considered to be acceptable taking into account the big discrepancies that are always present in land use data at this level of aggregation. Table 8. EU27 initial and calibrated land use trend demands RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND EU27 Initial trends of land demand (km2/Year) -4471.5 517.9 -303.2 233.6 3515.1 749.0 -3280.0 0.0 691.9 0.0 23.6 2323.4 RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND EU27 calibrated trends of land demand (km2/Year) -4471.5 517.9 1127.0 0.0 3515.1 0, -4000.0 0, 691.9 0.0 0.0 0.0 Table 9. EU27 calibrated matrices of shares of land use changes from others Calibrated shares of land use changes from others (EU27) share of --> that comes from: RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND RAINFED 0.00 1, 0.3, 0, 0.23 0, 0, 0, 0.75, 0.13, 0, 0; IRRIGATED 0.00 0, 0, 0, 0.00 0, 0, 0, 0, 0, 0, 0; FOREST_MANAGED 0.06 0, 0, 0, 0.61 0, 0, 0, 0.08, 0, 0, 0; FOREST_PRIMARY 0.00 0, 0, 0, 0.00 0, 0, 0, 0, 0, 0, 0; FOREST_PLANTATIONS 0.00 0, 0, 0, 0.00 0, 0, 0, 0, 0, 0, 0; SHRUBLAND 0.80 0, 0.3, 0, 0.10 0, 0, 0, 0.04, 0.13, 0, 0; GRASSLAND 0.12 0, 0.4, 0, 0.06 0, 0, 0, 0.06, 0.13, 0, 0; WETLAND 0.00 0, 0, 0, 0.00 0, 0, 0, 0, 0, 0, 0; URBAN 0.00 0, 0, 0, 0.00 0, 0, 0, 0, 0, 0, 0; SOLAR 0.00 0, 0, 0, 0.00 0, 0, 0, 0, 0, 0, 0; SNOW_ICE_WATERBODIES 0.00 0, 0, 0, 0.00 0, 0, 0, 0, 0, 0, 0; OTHER_LAND 0.02 0, 0, 0, 0.00 0, 1, 0, 0.06, 0.63, 0, 0; Figure 3. Percent of error between historical and simulated values of land uses in EU27 after the calibration. UK In Table 10 one can see the historical trends of land use change in UK. UK shows no significant change of forests and shrublands and loss of irrigated cropland (though the absolute value of irrigated cropland in UK is very small). Historical trends for cropland rainfed and plantations have been reduced a bit to adjust the loss of other land. Table 11 shows the calibrated shares. In general, land use changes are small in UK and the error between the historical and the simulated land use areas after the calibration are less than 6% for most land uses (Figure 4). The relative error of cropland irrigated is not considered important because the small area of this land use in UK makes it negligigle. Table 10. UK initial and calibrated land use trend demands RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND UK Initial trends of land demand (km2/Year) 352.0 -98.0 0.0 0.0 125.0 0.0 150.0 0.0 12.0 0.0 -1.0 -540.0 UK calibrated trends of land demand (km2/Year) 254.0 0.0 0.0 0.0 94.0 0.0 0.0 0.0 12.4 0.0 0.0 0.0 Table 11. UK calibrated matrices of shares of land use changes from others Calibrated shares of land use changes from others (UK) share of --> that comes from: RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND RAINFED 0.00 1.00 0.30 0.00 0.30 0.00 0.00 0.00 0.75 0.45 0.00 0.00 IRRIGATED 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 FOREST_MANAGED 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 FOREST_PRIMARY 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 FOREST_PLANTATIONS 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 SHRUBLAND 0.92 0.00 0.30 0.00 0.30 0.00 0.00 0.00 0.13 0.12 0.00 0.00 GRASSLAND 0.06 0.00 0.40 0.00 0.40 0.00 0.00 0.00 0.06 0.40 0.00 0.00 WETLAND 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 URBAN 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 SOLAR 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 SNOW_ICE_WATERBODIES 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 OTHER_LAND 0.02 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.06 0.00 0.00 0.00 Figure 4. Percent of error between historical and simulated values of land uses in UK after the calibration. CHINA In Table 12 one can see the historical trends of land use change in China. China shows a large increase of forests, plantations and croplands that seems to come from other land. Irrigated land is much larger than in other regions. Urban expansion is large and irrigated land demand is increased to cope with the demands from urban. Table 13 shows the calibrated shares and the error between the historical and the simulated land use areas after the calibration are shown in Figure 5. The average error is less than 6% for all uses. Table 12. China initial and calibrated land use trend demands RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND China Initial trends of land demand (km2/Year) 1336.0 0.0 6991.0 0.0 13933.0 -29.0 0.0 0.0 3876.0 0.0 149.0 -26256.0 China calibrated trends of land demand (km2/Year) 4772.1 49056.1 6990.6 0.0 13933.1 0.0 0.0 0.0 3875.9 0.0 0.0 0.0 Table 13. China calibrated matrices of shares of land use changes from others Calibrated shares of land use changes from others (China) share of --> that comes from: RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND RAINFED 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.76 0.15 0.00 0.00 IRRIGATED 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 FOREST_MANAGED 0.40 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.01 0.00 0.00 FOREST_PRIMARY 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 FOREST_PLANTATIONS 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 SHRUBLAND 0.60 0.00 0.50 0.00 0.50 0.00 0.00 0.00 0.20 0.04 0.00 0.00 GRASSLAND 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.09 0.00 0.00 WETLAND 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 URBAN 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 SOLAR 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 SNOW_ICE_WATERBODIES 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.04 0.00 0.00 OTHER_LAND 0.00 0.00 0.50 0.00 0.50 0.00 1.00 0.00 0.01 0.65 0.00 0.00 Figure 5. Percent of error between historical and simulated values of land uses in China after the calibration. EASOC In Table 14 one can see the historical trends of land use change in EASOC. Both croplands experiment important increases that seem to be compensated with the decrease of forest managed and primary. Table 15 shows the calibrated shares. The error between the historical and the simulated land use areas after the calibration are shown in Figure 6. There is a relevant error for shrubland and other land that we cannot compensate with the calibration. It seems to come from the fact that shrubland and other land areas have not been obtained from real historical data but from and approximation (assuming constant proportions between them) and this assumption might not hold. In any case, these uses are of very little importance for out model. Table 14. EASOC initial and calibrated land use trend demands RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND EASOC Initial trends of land demand (km2/Year) 15692.0 -155.0 -6669.0 -1477.0 3309.0 466.0 ###### 0.0 1341.0 0.0 -53.0 25710.0 EASOC calibrated trends of land demand (km2/Year) 15691.9 -155.0 0.0 0.0 3640.3 0.0 ###### 0.0 1341.3 0.0 0.0 0.0 Table 15. EASOC calibrated matrices of shares of land use changes from others Calibrated shares of land use changes from others (EASOC) share of --> that comes from: RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND RAINFED 0.00 1.00 0.30 0.00 0.30 0.00 0.00 0.00 0.75 0.63 0.00 0.00 IRRIGATED 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.09 0.00 0.00 FOREST_MANAGED 0.40 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.08 0.06 0.00 0.00 FOREST_PRIMARY 0.14 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 FOREST_PLANTATIONS 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.06 0.00 0.00 SHRUBLAND 0.46 0.00 0.30 0.00 0.30 0.00 0.00 0.00 0.04 0.06 0.00 0.00 GRASSLAND 0.00 0.00 0.40 0.00 0.40 0.00 0.00 0.00 0.06 0.08 0.00 0.00 WETLAND 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 URBAN 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 SOLAR 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 SNOW_ICE_WATERBODIES 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 OTHER_LAND 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.07 0.02 0.00 0.00 In Table 24 one can see the historical trends of land use change in LROW, that shows a large cropland expansion that can explain the losses of managed and primary forests. Table 25 shows the calibrated shares. The errors are below 6% and can be assumed. Table 24. LROW initial and calibrated land use trend demands RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND LROW Initial trends of land demand (km2/Year) 28841.0 -7048.0 -50912.0 -12302.0 3430.0 -4258.0 ###### 0.0 2872.0 0.0 -2231.0 66945.0 LROW calibrated trends of land demand (km2/Year) 14000.0 -7048.2 -37000.0 -7000.0 3429.7 0.0 ###### 0.0 2872.4 0.0 0.0 0.0 Table 25. LROW calibrated matrices of shares of land use changes from others Calibrated shares of land use changes from others (LROW) share of --> that comes from: RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND RAINFED 0.00 1.00 1.00 1.00 0.30 0.30 0.00 0.00 0.00 0.00 0.00 0.00 IRRIGATED 1.00 0.00 0.00 0.00 0.54 0.54 0.08 0.08 0.00 0.00 0.00 0.00 FOREST_MANAGED 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 FOREST_PRIMARY 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 FOREST_PLANTATIONS 0.70 0.70 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 SHRUBLAND 0.00 0.00 0.00 0.00 0.09 0.09 0.02 0.02 0.00 0.00 0.00 0.00 GRASSLAND 0.30 0.30 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 WETLAND 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 URBAN 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 SOLAR 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.02 0.00 0.00 0.00 0.00 SNOW_ICE_WATERBODIES 0.00 0.00 0.00 0.00 0.30 0.30 0.00 0.00 0.00 0.00 0.00 0.00 OTHER_LAND 0.00 0.00 0.00 0.00 0.20 0.20 0.08 0.08 0.00 0.00 0.00 0.00 Figure 11. Percent of error between historical and simulated values of land uses in LROW after the calibration. The future trends of land expansion used in the model when the historical period ends are not necessarily the same as the historical ones, since some uses show clear rupture of the past trends. The trends used after the historical data are shown in table 26. Table 26. Trends of land expansion used in the simulation of the model after the historical period TRENDS OF FUTURE LAND DEMAND BY REGION RAINFED IRRIGATED FOREST_M ANAGED FOREST_ PRIMARY FOREST_ PLANTATI ONS SHRUB LAND GRAS SLAND WETL AND URBAN SOLAR SNOW_ICE _WATERB ODIES OTHER_ LAND REGIONS_I|LANDS_I [Mm2/Year] [Mm2/Year] [Mm2/Year] [Mm2/Year] [Mm2/Year] [Mm2/Year] [Mm2/Year] [Mm2/Year] [Mm2/Year] [Mm2/Year] [Mm2/Year] [Mm2/Year] EU27 0 0 0.00112702 0 0.003515 0 0 0 0.0007 0 0 0 UK 0 0 0 0 0.000125 0 0 0 1E-05 0 0 0 CHINA 0.004 0.00342736 0.01 0 0.013933 0 0 0 0.0039 0 0 0 EASOC 0.015551 9.0001E-05 0 0 0.00364 0 0 0 0.0013 0 0 0 INDIA 0 0 0.000484 0 0.001533 0 0 0 0.0007 0 0 0 LATAM 0.008 0.00193252 -0.03 0 0.004762 0 0 0 0.0006 0 0 0.04 RUSSIA 6.29E-05 -0.0001621 0.00122584 0.001404 0.0011 0 0 0 0.0003 0 0 0 USMCA 0 0.00074409 0 -0.001057 0.006813 0 0 0 0.0021 0 0 0 LROW 0.028841 0 -0.0509123 -0.012302 0.001715 0 0 0 0.0029 0 0 0.044 REFERENCES [1] van Schrojenstein Lantman, J., Verburg, P.H., Bregt, A., Geertman, S., 2011. Core Principles and Concepts in Land-Use Modelling: A Literature Review, in: Koomen, E., Borsboom-van Beurden, J. (Eds.), [2] Heistermann, M., Müller, C., Ronneberger, K., 2006. Land in sight?: Achievements, deficits and potentials of continental to global scale land-use modeling. Agric. Ecosyst. Environ. 114, 141–158. doi:10.1016/J.AGEE.2005.11.015 [3] Schaldach, R., Priess, J.A., 2008. Integrated Models of the Land System: A Review of Modelling Approaches on the Regional to Global Scale. Living Rev. Landsc. Res. 2, 1. doi:10.12942/lrlr-2008-1 [4] van Soesbergen, A., 2016. A review of land-use change models. UNEP World Conservation Monitoring Centre, Cambridge UK. [5] https://www.environmentalgeography.nl/site/data-models/models/ [6] Stehfest, E., van Vuuren, D., Bouwman, L., Kram, T., 2014. Integrated assessment of global environmental change with IMAGE 3.0: Model description and policy applications. Netherlands Environmental Assessment Agency (PBL). [7] https://www.pik-potsdam.de/en/institute/departments/activities/biosphere-watermodelling/lpjml [8] van Asselen, S., Verburg, P.H., 2013. Land cover change or land-use intensification: simulating land system change with a global-scale land change model. Glob. Chang. Biol. 19, 3648–3667. doi:10.1111/gcb.12331 [9] Dietrich, J.P., Bodirsky, B.L., Humpenoder, F., Weindl, I., Stevanovic, M., Karstens, K., Kreidenweis, U., Wang, X., Mishra, A., Klein, D., Ambrosio, G., Araujo, E., Yalew, A.W., Baumstark, L., Wirth, S., Giannousakis, A., Beier, F., Chen, D.M.-C., Lotze-Campen, H., Popp, A., 2019. MAgPIE 4 - a modular open-source framework for modeling global land systems. Geosci. Model Dev. 12, 1299–1299. [10] Wise, M., Calvin, K., Kyle, P., Luckow, P., Edmonds, J., 2014. Economic and physical modeling of land use in GCAM 3.0 and an application to agricultural productivity, land, and terrestrial carbon. Clim. Chang. Econ. 05, 1450003. doi:10.1142/S2010007814500031 [11] Wise, M., Calvin, K., Thomson, A., Clarke, L., Bond-Lamberty, B., Sands, R., Smith, S.J., Janetos, A., Edmonds, J., 2009. Implications of limiting CO2 concentrations for land use and energy. Science 324, 1183–6. doi:10.1126/science.1168475 [12] Havlík, P., Schneider, U.A., Schmid, E., Böttcher, H., Fritz, S., Skalský, R., Aoki, K., Cara, S. De, Kindermann, G., Kraxner, F., Leduc, S., McCallum, I., Mosnier, A., Sauer, T., Obersteiner, M., 2011. Global land-use implications of first and second generation biofuel targets. Energy Policy 39, 5690–5702. doi:10.1016/J.ENPOL.2010.03.030 [13] Havlík, P., Valin, H., Mosnier, A., Frank, S., Lauri, P., Leclère, D., Palazzo, A., Batka, M., Boere, E., Brouwer, A., Deppermann, A., Ermolieva, T., Forsell, N., Di Fulvio, F., Obersteine, M., Herrero, M., Schmid, E., Schneider, U., Hasegawa, T., 2018. GLOBIOM Documentation. Laxenburg, Austria. [14] Christoph Müller, Elke Stehfest, Jelle G van Minnen, Bart Strengers, Werner von Bloh, Arthur H W Beusen, Sibyll Schaphoff, Tom Kram and Wolfgang Lucht. Drivers and patterns of land biosphere carbon balance reversal. Environ.Res.Lett. 2016. 11 044002 [15] Martin von Lampe, Dirk Willenbockel, Helal Ahammad, Elodie Blanc, Yongxia Cai, Katherine Calvin, Shinichiro Fujimori, Tomoko Hasegawa, Petr Havlik, Edwina Heyhoe, Page Kyle, Hermann Lotze-Campen, Daniel Mason d'Croz, Gerald C. Nelson, Ronald D. Sands, Christoph Schmitz, Andrzej Tabeau, Hugo Valin, Dominique van der Mensbrugghe, Hans van Meijl. Why do global long-term scenarios for agriculture differ? An overview of the AgMIP Global Economic Model Intercomparison. Agricultural Economics, Volume45, Issue1, January 2014. https://doi.org/10.1111/agec.12086 [16] Meadows DH, Meadows DL, Randers J, Behrens WW. The limits to growth. New York: Universe Books; 1972. [17] Mapping SPAM (Spatial Production Allocation Model). https://mapspam.info/ [18] https://www.fao.org/faostat/en/#data/RL/metadata FAO definitions and standards land use [19] F. N. Tubiello et al., “Measuring the world’s cropland area,” Nat Food, vol. 4, no. 1, Art. no. 1, Jan. 2023, doi: 10.1038/s43016-022-00667-9. [20] Modelado de los cambios de uso de la tierra mediante Dinámica de Sistemas para su integración en el modelo WILIAM. Trabajo Fin de Grado, Universidad de Valladolid. [21] Marina Campano-Méndez, Land use Chnages Literatura Review. GEEDS internal report, July 2021. https://uvadoc.uva.es/handle/10324/75091 [22] Ferreras-Alonso, N., Capellán-Pérez, I., Adam, A., de Blas, I., Mediavilla, M. 2024. Mitigation of land-related impacts of solar deployment in the European Union through land planning policies. Energy, 302, 131617.