D2.4 Information on climate service needs and gaps

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 101037293 Deliverable D2.4 Information on climate service needs and gaps June 2025 This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 101037293 Innovating Climate Services through Integrating Scientific and local Knowledge Deliverable Title: Information on Climate Service Needs and Gaps Author(s): Katie Egan (ECMWF), Rebecca Emerton (ECMWF) Contributing Author(s): Ilyas Masih (IHE), Micha Werner (IHE), Christel Prudhomme (ECMWF), Calum Baugh (ECMWF), Paolo Mazzoli (GECOsistema), Stefano Bagli (GECOsistema), Valerio Luzzi (GECOsistema), Francesca Renzi (GECOsistema), Daniele Castellana (Netherlands Red Cross, 510), Gal Agmon (Netherlands Red Cross, 510), Vakho Chitishvili (CENN), Lotte Muller (VUA), Marije Schaafsma (VUA), Johannes Schnell (52N), Felix Voigtlander (52N), Ricardo Buitenhuis (52N), Benedikt Graler (52N), Lucia De Stefano (UCM), Lluis Pesquer (CREAF), Amanda Batlle (CREAF), Annelies Broekman (CREAF), Nuria Hernandez-Mora (UCM), Ester Prat (CREAF), Nikoletta Ropero (UCM), Alexandros Ziogas (EMVIS). Date June 2025 Suggested citation: Egan, K., Emerton, R., et al., 2025: Information on climate service needs and gaps, I-CISK Deliverable 2.4, available online at www.icisk.eu/resources Availability: ☒ PU: This report is public ☐ CO: Confidential, only for members of the consortium (including the Commission Services) Document Revisions: Author Revision Date Rebecca Emerton & Katie Egan Version 1 31/05/2025 Contributing Authors; Co - ordinators Internal Review 5 - 13 2025 Rebecca Emerton Final version 2 5 /06/2025 Rebecca Emerton Revisions following review 24/10/2025 [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 1 Executive Summary Co-exploring user-needs for climate data and information to support critical decision making is key to the design and development of useful and usable Climate Services (CS). It is important to understand barriers to the use of existing CS and how these issues can be addressed through effective design and communication. This report (Egan, Emerton, et al., 2025 [D2.4]) provides a detailed overview of the use of existing climate services (CS) at the start of the I-CISK project in early 2022 in each of the seven participating Living Labs (LLs; located in The Netherlands, Spain, Italy, Greece, Hungary, Georgia and Lesotho). It identifies user needs and barriers to the use of these existing services, then demonstrates how new, tailored CS, co-created through the I-CISK project, are addressing them. A preliminary version of this report, Moschini, Emerton, et al., 2022 [D2.1], published in April 2022, provided an initial understanding of decision-making and CS needs in each LL, obtained through the project scoping process, initial discussion meetings, the establishing of the LLs, reports on the characteristics of each LL (Masih, Van Cauwenbergh, et al., 2022 [D1.1]) and targeted questionnaires and interviews. Throughout the I-CISK project, further questionnaires and interviews, along with information from workshops within the LLs, were used at regular intervals (~annually) to hear from the LLs about additional barriers, decision-making contexts and user needs identified during the course of co-development. This final report summarises information from the preliminary deliverable and two further iterations completed in 2023 and 2024, drawing conclusions on the progress made towards providing usercentred CS that address the gaps identified in existing provisions. It further documents the experience gained and lessons learnt throughout the project. Some of the key challenges identified in the use of existing CS for decision-making include insufficient resolution (spatial and/or temporal), or forecasts that are aggregated in such a way that doesn’t allow users to identify key patterns and distributions. A lack of useful variables/indicators and issues related to the accessibility and usability of CS, including data availability, download difficulties, and challenges in communication, including language, and dissemination, were also highlighted. In response to these barriers, users expressed a need for services that provide forecasts at different and extended timescales (e.g., sub-seasonal and seasonal), tailoring to specific sectors and decision-making contexts, impactand action-based forecasts, additional variables such as streamflow or vegetation indices, higher resolutions, and improved access and communication channels. The evolving role of visualisation in CS was noted across all LLs, with map-based interfaces, time series plots, uncertainty ranges, alert thresholds, and summary dashboards all used and adapted based on user feedback. The co-development process played a central role in ensuring that visualisations were meaningful, accessible, and suited to real-world decision-making. Visual elements were iteratively improved through workshops, mock-ups, and user testing, with feedback informing everything from colour schemes and icons to interactivity and layering options. Efforts to communicate uncertainty and probabilistic information varied by context, with some LLs favouring simplified outputs and others incorporating bestand worst-case scenarios or user-adjustable percentile displays. A key area identified for future research is how best to incorporate both forecast uncertainty and skill (performance and accuracy) information into CS. Co-evaluation conducted as part of I-CISK demonstrates the benefit of helping users understand uncertainty, for example through serious games [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 2 or workshops. However, even when uncertainty is understood, it can still be difficult to know how it should be incorporated into user decision making - for example, how do we deal with probability when there is a set threshold that triggers an action? Not including information on forecast skill (performance and accuracy) may prove to be a gap in a number of the new CS developed here, for example if the underlying forecasts are not ‘good enough’ to support users’ decision-making, but this is not clear to the users. At this stage, only two of the new CS consider providing information on forecast skill. This report brings together these findings and highlights common themes across the LLs, while recognising the value of local context and knowledge. It offers a synthesis of the barriers, needs and co-developed solutions, and documents best practices and lessons learned in co-producing CS that are inclusive, practical, and locally relevant. Overall, this work demonstrates the value of participatory, user-centred approaches in the development of CS. Through sustained engagement with diverse users, the I-CISK project has helped to shift the design of CS from a one-size-fits-all approach to services that reflect real decision-making contexts, improve accessibility and trust, and build capacity for climate resilience. These lessons provide a strong basis for continuing to enhance the usefulness and impact of CS in the years ahead. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 3 Table of Contents Executive Summary ................................................................................................................................. 1 Table of Contents .................................................................................................................................... 3 List of Figures .......................................................................................................................................... 4 List of Tables ........................................................................................................................................... 6 1 Introduction .......................................................................................................................................... 7 2 Method ................................................................................................................................................. 9 3 Summary of findings .......................................................................................................................... 10 4 Use and development of climate services .......................................................................................... 11 4.1 Georgia ....................................................................................................................................... 11 4.2 Greece ......................................................................................................................................... 16 4.3 Hungary ...................................................................................................................................... 23 4.4 Italy ............................................................................................................................................. 27 4.5 Lesotho ....................................................................................................................................... 31 4.6 The Netherlands ......................................................................................................................... 37 4.7 Spain ........................................................................................................................................... 42 5 Overarching remarks .......................................................................................................................... 53 5.1 Barriers and needs ...................................................................................................................... 53 5.2 The new climate services - commonalities and differences ....................................................... 54 5.3 Uncertainty and skill ................................................................................................................... 56 5.3.1 Roadmap for integrating skill information into tailored climate services .......................................... 57 5.4 Effective visualisation and communication ................................................................................ 58 5.5 User-Informed Design and Iterative Feedback ........................................................................... 60 5.6 Remaining Challenges and Opportunities .................................................................................. 60 6 Conclusions ......................................................................................................................................... 62 References ............................................................................................................................................. 63 [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 4 List of Figures Figure 1 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in Georgia. ......................................................................................................................................... 11 Figure 2 Example screenshot from the Georgian LL climate service. ................................................... 15 Figure 3 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in Crete, Greece. ............................................................................................................................... 16 Figure 4 Example screenshot from the Cross-Sector Planning Service for Tourism, displaying a map of the seasonal forecast of the frequency of light precipitation. A map is provided for Crete, and clicking on a location indicates the frequency for the current month and shows a graph of the forecast frequency over the next seven months. ............................................................................................... 20 Figure 5 Example screenshot from the Cross-Sector Planning Service for Tourism, displaying a map of the seasonal forecast for the ‘tourism climatic index’. A map is provided for Crete, and clicking on a location indicates the current value and shows a graph of the forecast over the next seven months. .............................................................................................................................................................. 21 Figure 6 Example screenshot from the Cross-Sector Planning Service for Tourism, displaying a map of the seasonal hydrological forecast. A map is provided for Crete, and clicking on a subcatchment displays shows a box plot graph of the probabilistic forecast over the next seven months. ............... 21 Figure 7 Example screenshot from the Cross-Sector Planning Service for Tourism, displaying a map from the climatic part of the service, indicating the current and projected average summer temperature. A map is provided for Crete, and clicking on a location displays the current mean summer temperature, and displays a graph indicating the projected change in mean summer temperature out to 2090. The map can be changed to display the mean summer temperature in future decades. ..... 22 Figure 8 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in Hungary. ........................................................................................................................................ 23 Figure 9 Screenshot from the Hungary LL climate service. This display shows the measured thermal orthophoto for Budapest VI district. ..................................................................................................... 26 Figure 10 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in Italy. .............................................................................................................................................. 27 Figure 11 Example screen shot from the Italian LL climate service mock-up, displaying the map and station location (left) and ensemble seasonal forecast of river flow with daily timestep and with thresholds marked (right). .................................................................................................................... 30 Figure 12 Example screenshot from the Italian LL climate service mock-up, displaying the map and station location (left) and ensemble seasonal forecast from GloFAS with a monthly timestep, alongside climatological percentiles, displayed as a box plot (right). ................................................................... 30 Figure 13 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in Lesotho. ........................................................................................................................................ 31 Figure 14 Screenshot from the Lesotho LL CS, indicating what users will see if there is a warning for drought in several districts, based on the ECMWF SEAS5 forecast. Icons indicate warnings for the upcoming six months, and the interface highlights that a trigger is expected for drought, the exposed districts and the associated populations. This is based on mock data and does not indicate a real forecast. ................................................................................................................................................ 34 [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 5 Figure 15 Screenshot from the Lesotho LL CS, indicating what users will see when clicking on a specific district from the display shown in figure 14. The exposed population in the chosen district is shown, and an advisory to inform potentially exposed districts and follow local protocols. ........................... 35 Figure 16 Screenshot from the Lesotho LL CS, displaying an additional layer that users can choose to visualise if they are interested in the probability of drought according to the global forecast from ECMWF’s SEAS5 seasonal forecasting system. ..................................................................................... 35 Figure 17 Screenshot from the Lesotho LL CS, indicating what users will see if they choose the option to ‘set trigger’ – this option allows the user to select the districts where the Red Cross Early Action Protocol should be activated. ............................................................................................................... 36 Figure 18 Screenshot from the Lesotho LL CS, indicating what users will see when the trigger has been set. The districts highlighted in red are at risk of drought, according to the official seasonal forecast from LMS. .............................................................................................................................................. 36 Figure 19 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in the Netherlands. ........................................................................................................................... 37 Figure 20 Screenshot of the welcome page of the Netherlands LL CS. The user can select their sector (recreational boating, agriculture or water management), which provides a dropdown of CS options tailored to that sector – including drought alerts, discharge at Lobith, precipitation deficits and climate. .............................................................................................................................................................. 40 Figure 21 Screenshot of the Netherlands LL CS river discharge forecast at Lobith, displaying an ensemble forecast of river discharge out to 4 months ahead, and indicating the forecast ensemble mean, percentiles to display uncertainty, and a low water threshold. The user can toggle to show observed streamflow for the preceding month before the forecast, and any observations available since the start of the forecast. The forecast can also be displayed as individual ensemble members (a ‘spaghetti’ chart). .................................................................................................................................. 41 Figure 22 Screenshot of the Netherlands LL CS precipitation deficit information, indicating the predicted precipitation deficit in mm, displaying a map of the median of the forecast ensemble, and an ensemble forecast chart for the selected location. ......................................................................... 41 Figure 23 Screenshot of the Netherlands LL CS weekly drought alert. Options within each dropdown guide the user to other relevant pages within the CS for more information, such as the river discharge forecast or the map of precipitation deficit forecast. .......................................................................... 42 Figure 24 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in Spain. ............................................................................................................................................ 42 Figure 25 Screenshot of the Spain LL CS 2, showing a seasonal precipitation forecast. The map displays the median of the forecast ensemble, with seasonal bias correction and downscaling applied, showing a spatial resolution of 231m. A toggle allows the user to display the associated uncertainty (see Figure 26). ........................................................................................................................................................ 49 Figure 26 Screenshot of the Spain LL CS2, showing a seasonal precipitation forecast. The map displays the uncertainty in the forecast, which has been bias corrected and downscaled to a spatial resolution of 231m. Clicking on a point on the map shows the 5th to 95th percentiles of the forecast, providing an assessment of the forecast uncertainty at that location. ..................................................................... 50 Figure 27 Screenshot of the Spain LL CS 3, showing historical climate information. The map displays temperature, precipitation or drought indices, and allows comparison of two different time periods using the dropdown menus at the top, and the slider underneath the map. Clicking on a location displays a time series of the data at that location. Another map layer provides temperature and precipitation time series data at individual stations. ........................................................................... 51 [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 6 Figure 28 Screenshot of the Spain LL CS, showing agroclimatic indicators. The map displays the maximum number of consecutive dry days (<1 mm of rain) per season, for the years 2011 to 2040. This CS is still under development and the final service may be displayed differently based on the upcoming development and feedback from the stakeholders in the LL. ............................................. 52 List of Tables Table 1 Summary of user needs, gaps/barriers in existing CS, and how the new CS address these needs. .............................................................................................................................................................. 10 Table 2 Actors involved in the Spanish LL, their role, and the CS produced and used by each actor. . 45 Table 3 Overview of information provided by the new CS co-created under the I-CISK project. ........ 54 [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 7 1 Introduction Climate services (CS)1 are crucial in empowering citizens, stakeholders and decision-makers in defining resilient pathways to prepare for hazards and extreme events and adapt to climate change. Availability of CS has improved significantly in recent years, alongside advances in scientific knowledge and data, with CS such as those from Copernicus or GEOSS (Global Earth Observation System of Systems) providing a range of data, training, access to scientific knowledge and forecasts (e.g. the Copernicus Emergency Management Service (CEMS), Climate Change Service (C3S) and Atmosphere Monitoring Service (CAMS), www.copernicus.eu/en/copernicus-services; the GEOSS portal, earthobservations.org/geoss.php). Despite this, there remain challenges for end-users to make the best use of the potential of such CS and data, including accessibility, local applicability and the translation of scientific data into actionable information, social and behavioural factors and varying needs of decision-makers. The typical approach in the development of CS is the top-down approach and has often been “onesize-fits-all” (Jacobs and Street, 2020; WISER, 2020), but approaches to developing and providing CS are continually evolving and moving towards those that account for a broad range of societal challenges and potential users. A co-creation2 approach, used throughout the I-CISK project and incorporating co-exploration, codesign, co-production, co-implementation and co-evaluation, can help to overcome challenges that lead to a lack of usability of CS, and provides the opportunity to meet climate information needs at relevant spatial and temporal scales across a range of regions and sectors (WISER, 2020; Hirons et al., 2021). I-CISK has involved and engaged stakeholders3 (including CS providers, purveyors and endusers) at each step of the co-creation process, in order to co-produce tailored CS that integrate local knowledge and experiences with large-scale data and information. Co-exploring needs surrounding the value of CS, climate data and information is key in the design and development of CS. It is important to understand the decision-making context of CS end-users, the barriers to use of existing CS and how these issues can be addressed in the development of nextgeneration CS to provide CS that are useful, usable and effectively address user needs. 1 The I-CISK prototype framework on co-creating end-user centred climate services (MS10, 2022) includes discussion of ‘what do we mean by Climate Services?’, from which the following is adapted: “climate services” is broadly defined as “the transformation of climate-related data — together with other relevant knowledge — into customized products such as projections, forecasts, warnings, trends, economic analysis, and risk assessment, which allows to deliver information on best practices, to develop and evaluate solutions, and to provide any other service in relation to climate that may be of use for the society at large” (Street et al., 2015; MS10, 2022). 2 Co-creation is the interdisciplinary, interactive and iterative approach to developing CS, as a way to overcome the divide between climate science and decision-makers. It is often used interchangeably with co-production or co-design. In the I-CISK project, we use the term co-creation to describe the collaborative process encompassing the co-design, co-production, coimplementation, co-evaluation and dissemination of user-centred CS (MS10, 2022). 3 Stakeholders is the general term that encompasses all CS producers, intermediaries and consumers, or others who are affecting/affected by the decisions informed by CS (or absence thereof). Within I-CISK (MS10, 2022), the following stakeholder categories are defined: (1) actors – stakeholders that play an active role in the technology, institutional and investment readiness of CS. These are the stakeholders affecting decisions, by creating either drivers or barriers. They include, for example, the project team, scientists, practitioners, decision-makers, private sector, public authorities, providers, end-users, etc. (2) providers – actors who provide the necessary data, investment, regulatory context for the CS to be sustained; they supply climate information and knowledge, operating on a range of scales and in different sectors. (3) purveyors – act as knowledge brokers providing guidance on ways that CS can address regional problems. They also ensure that products, scientific results and business opportunities are adequately communicated to end-users. (4) end-users – actors who use CS at different levels of the decision chain. They employ climate information and knowledge for decisionmaking, and may or may not participate in developing the CS itself, or may also pass information on to others, making them both users and providers. They include civilians, companies, developers, private organisations, local communities, governments etc. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 14 management in the Alazani and Iori River Basins, especially for agriculture and energy (hydropower). In the context of a changing climate, and with further hydropower plants planned, the pressure on water resources is likely to increase in the coming decades (Masih, van Cauwenburgh, et al., 2022 [D1.1]). To address the need for sector tailored streamflow predictions a, ‘Water Resource Planning Service’ has been developed, and a mock-up is available on the I-CISK Living Labs Server (https://icisk.dev.52north.org/living-labs/alazani--ge/). The primary function of the service is the planning and management of water resources in the Alazani-Iori River Basins at various lead times (LL survey). It will provide ensemble streamflow forecasts for the Alazani and Iori River Basins, covering lead times >3 days and including sub-seasonal and seasonal forecasts. Two new, tailored methods are in development, one for sub-seasonal and seasonal drought forecasts and the other a downscaling of hydrological forecasts to sub-catchments or points of interest (MS11). The CS uses historical rainfall data from meteorological stations and real-time river flow data from Shakriani hydrological stations as additional input. The development has considered multiple sources of streamflow forecasts for the CS, including from SMHI’s W-HYPE model and from the Copernicus Emergency Management Service’s (CEMS) Global Flood Awareness System (GloFAS) and European Flood Awareness System (EFAS). The CS also incorporates the Standardised Precipition Index (SPI) for drought monitoring, based on the ERA5-Land reanalysis dataset, and bias-corrected (using ERA5 data as a reference) seasonal forecasts of precipitation and temperature from ECMWF’s SEAS5 seasonal forecasting system. Precipitation forecasts are displayed as monthly accumulations, and temperatures as the average daily maximum and minimum over a month. Maintenance of the observation network and integration of observed data are also being addressed as part of the I-CISK project; the LL have undertaken field work to assess the amelioration (irrigation) system and streamflow monitoring sites, as well as the quality of streamflow data. The aim is to use these within the stream flow model, but high-quality calibration is challenging due to the limitations of the historical data. Work is currently underway to find a solution. How is the new information visualised and communicated? Figure 1 provides a snapshot from the online portal developed for the tailored CS, showing a map of the region and rivers covered by the CS, and an example of a streamflow forecast for one of the locations indicated by the red dots. The forecast is displayed using shading to indicate the forecast according to ensemble percentiles. The user can select from a range of time horizons, and can view observed data in the time series underneath the forecast. During the co-development of the CS, the LL team engaged in discussions on ensemble forecasting, probabilistic information and uncertainty. As the CS was planned to make use of ensemble forecasts, meetings were arranged to explain the benefits of probabilistic information in managing flood and drought risks. It was noted in the recent survey that initially, farmers were unfamiliar with probabilistic forecasts, posing challenges for conveying the concept effectively. Workshops were therefore arranged, aiming to simplify the concept by using visual aids and local examples to help farmers to understand and engage with the forecasting features of the new CS. A ‘serious game’ was also developed based around the user stories and decision-making context of farmers in Georgia (Rastogi et al., 2024). [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 15 Figure 2 Example screenshot from the Georgian LL climate service. The LL plans to gather feedback from the end users regarding the visualisation used in the CS, and to discuss new features and the portal’s usability. The feedback will be used to further refine the portal’s features for improved accessibility. Information on forecast skill (see Baugh, Egan, et al., 2025 [D3.4]) will also be incorporated into the CS in future, to help users understand the accuracy and usability of the forecasts. How does the new CS address the identified barriers and needs? Decision making by the National Environmental Agency (NEA), Georgian Amelioration, and the Rural Development Agency will be supported by the provision of streamflow forecasts at point locations of interest to them. The system will also support planning in the agricultural sector, flood control and hydropower operations. There has previously been no provision of streamflow forecasts in Georgia, and users instead rely on weather forecasts of precipitation. Whilst the NEA does produce runoff forecasts, these are available only for the spring period (D2.1 Iteration 3 survey). The provision of continuous streamflow forecasts at points of interest to the NEA will address these issues, making this a unique service within Georgia. Feedback from the LL from the most recent survey also indicates that this new service introduces drought-focussed CS with the possibility of downloading historical data and enabling users to access and analyse trends for the first time. New data provided by the service include historical rainfall data and ‘live’ hydrological data from the gauge at Shakriani on the Alazani River, with improved accessibility and integrated forecasting system. While the multi-hazard service under development through the Green Climate Fund will address some of the identified needs at short timescales, the new CS developed through I-CISK will address the identified need for monthly, sub-seasonal and seasonal forecasts (LL survey) throughout the year and support economic activity planning for multiple sectors. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 16 4.2 Greece Figure 3 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in Crete, Greece. What is the decision-making context? The Island of Crete is the focus area of the Greek LL where the main weather/climate hazards are heatwaves, floods, wildfires and droughts. In the I-CISK project, the hazards considered in the development of new and tailored CS include water scarcity, drought (indirectly), landslides, the combination of wind waves and swell, flooding, high winds and heatwaves. The sectors impacted and involved include tourism, water resource management, transport, infrastructure and accommodation. The main end users of the CS include tourism enterprises, tourists, citizens, road and port managers and the water supply and energy sectors. Decisions related to climate hazards are linked to water allocation in periods of droughts, planning of tourist activities and energy demand (due to higher usage of cooling systems) in relation to heatwaves, tourism and transportation disruption related to flash floods, and port traffic planning when strong winds are anticipated. Stakeholders take different decisions on a daily, weekly and seasonal basis in relation to the activities of their sector. What existing CS were already in use at the start of the I-CISK project, if any? At the start of the project, it was noted that information already in use included daily weather forecasts, short-term forecasts of reservoir water quantity and quality, and climate change impact assessments and vulnerability analysis at district level. It was found that in the tourism sector, decision-making was typically based on weather forecast information, whereas water management decisions may have been based on current conditions, experience, conditions during previous years, and past managing practices, rather than on forecast information. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 17 The existing CS identified included: ● Weather forecasts provided by national services via mobile apps and internet in the form of maps, charts and tables of precipitation, temperature and wind. These were being used by water management and tourism sectors for decisions on water allocation, and planning outdoor activities and construction. This information was seen to be trusted. ● Climate change impact assessment and vulnerability analysis information is available at national and regional scales, alongside climate change adaptation plans and research studies, including from private (consultancy) companies. These included information on river flow, precipitation, temperature, drought index and water exploitation index, with information feeding into longer-term strategies for public water management and tourism (e.g. long-term construction planning). ● Hindcasts were mentioned as being used by the water resources management sector for water allocation and analysis of climate impacts on water availability. ● A bespoke short-term forecast service for reservoir water quality and quantity was also in use by the water resources management for water allocation and information on water quality, in relation to both drought and extreme precipitation. What were the barriers to using existing CS? The barriers to using these existing CS effectively include the fact that climate change information lacked cross-sector links, the CS were not tailored for specific sectors, there was a lack of accessibility for non-expert users, and a lack of trust in the information. Lack of cross-sector links Climate change vulnerability assessments are available, however, they focus on single sectors and lack information on cross-sectoral-links. Studies were mostly designed for governmental and administrative level. Lack of tailored information While the climate change vulnerability assessments are available for specific sectors, in general there is a lack of CS information on different timescales that is tailored to support the needs of different sectors, for example with the most useful variables and indicators, and including a lack of information on the severity of predicted hazards or compound impacts of multiple hazards. Accessibility to non-expert users CS that use climate projections, seasonal and sub-seasonal information would be of great use in this LL to inform local administration, local authorities and local businesses to better plan development and management activities. At the moment, this information is accessible only by researchers and consultancies and therefore does not reach a wider audience of potential users and stakeholders. The main barrier is the lack of expertise/resources needed to extract and convert this data into useful information for the LL. Also related to accessibility and usability of CS by non-expert decision-makers, a lack of information/clarity regarding the reliability and uncertainty of CS was noted. Lack of trust In the recent survey, feedback was received that during the course of the project, it became apparent through various discussions within the LL context, that an additional barrier not reported in the preliminary deliverable was an element of lack of trust in the information. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 18 What needs were identified for improved and tailored CS? In the Greece LL (Crete), user requirements are focused on the tourism sector, where there is a need to alleviate drinking water and irrigation need conflicts in summer (peak of tourist period), cope with increased energy needs, ensure access to touristic destinations (i.e. a functional transport network), and to increase preparedness and adaptation (Table 1). Specific user needs within the sector are variable and pertain to a range of natural/socio-natural hazards, which affect different sectors that subsequently impact on tourism. Water scarcity, drought, flooding and heatwaves all affect tourism and the economy due to their impacts on guest experience, energy demand for cooling, and transportation via roads and ports. Climate projections for decreased precipitation and increased temperature suggest these impacts will worsen (Masih, van Cauwenburgh, et al., 2022 [D1.1]). As a result, the following needs were identified: Sectortailored information A lack of information on a range of timescales, tailored to support sector needs indicated the need for new sector-specific indicators and variables. In the case of climate change assessments, it was seen as potentially useful to identify links between sectors, providing information to support decision-makers outside of the government and administrative level and increase accessibility for non-expert users. Improved spatio-temporal resolution The stakeholders expressed the desire to have CS of least 10 km spatial resolution covering the Island of Crete, and access to information on monthly, subseasonal and seasonal timescales. Climate hazard severity An identified gap was in information on the predicted severity of an event, and therefore there is a need for the provision of forecasts that include severity. Reliability and uncertainty Both responses highlighted that there was a lack of information or clarity regarding reliability and uncertainty information related to the CS they were using, and such information would be useful for decision-making. Compound / multi-hazard information According to the LL report (Masih, I., van Cauwenburgh, N., et al., 2022 [D1.1]), the need was identified for CS that “help to assess synergistic effect of multiple climatic threats, water and energy needs (availability of resources) and infrastructure (e.g. resorts, ports, marinas, roads, etc.) physical security due to extreme events (flooding, surging, snow fall, icing etc.). The service should also be able to target a diversity of seasonal and spatial coverage (summer coastal activities / winter mountainous activities).” What information does the new CS co-created through I-CISK provide? To address these needs, a ‘Cross-Sector Planning Service for Tourism’ has been developed, available on the EmvisWater platform: https://icisk.emvis.gr This new CS includes sub-services for users involved in water, road, port and resort management. Data are provided in the form of 12 indices identified as helpful by users during the co-design process and decision time-lines exercise (Van den Homberg, Rastogi, et al., 2024 [D2.5]). While some of the indices included are based on essential variables that may be available from other CS, this new CS aims to provide access to all relevant information in one easy-to-access location. Other variables are newly-developed and not previously available through any existing CS. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 19 The water management dashboard includes wet period surface water availability for reservoir basins, with specific indices including, for example, surface water discharge for November to February (4 months) relative to historical conditions. This will support decisions on water allocations for public and private (agricultural) clients, and management of water reserves (flood control, storage or release to the environment). The service will aid decision-making by providing data at critical periods, which will allow water to be allocated and used more effectively. Decisions on water restrictions will also be better informed, when they are required. The hydrology part of the CS provides maps and graphs of forecast seasonal outflow for subcatchments with a lead time up to 7 months ahead. The climatic part of the CS provides information from climate projections based on two forcing scenarios (RCP 4.5 and RCP 8.5), including temperature and precipitation, broken down into the following indicators: mean annual, summer, winter and monthly temperature, hot days, cold days, cooling degree days, heating degree days, total annual precipitation, high precipitation days, extreme precipitation days and monthly precipitation profiles. Information is provided as both absolute values and as anomalies from a reference baseline (1980-2005). The seasonal indicators for tourism part of the CS provides seasonal forecasts of indices related to extreme weather conditions relevant for the tourism sector, including extended hot spells and longperiod precipitation. Indices include: ● temperature and precipitation (aggregated over 15 days out to 7 months) ● the ‘Tourism Climatic Index’ (a variable evaluating climate favourability for outdoor tourism through the combination of seven variables related to human comfort levels) ● the frequency of light/high/extreme precipitation (frequency of days where the daily precipitation exceeds 1/10/50 mm) ● the frequency of extreme hot days (number of days where the maximum temperature exceeds 35°C) ● the frequency of tropical nights (number of nights where the temperature does not drop below 20°C) ● cooling degree days (related to energy demand for cooling, this variable tracks by how much and for how long the temperature exceeds 25°C) ● ‘Instability Index (quantifying the susceptibility to landslides in a specific area, accounting for rainfall, soil properties, slope inclination, geology) ● the frequency of strong north, northwesterly, and southerly wind gusts (exceeding Beaufort Force 7). These will support decisions on which summer activities to provide, when to carry out development and maintenance works, and future investment/expansion, for example of hotels or resorts in the tourism sector. The CS makes use of a range of data, including observations of precipitation and temperature, forecasts from ECMWF’s seasonal forecasting system (SEAS5), regional climate projections (CORDEX, the Coordinated Regional Downscaling Experiment) through C3S, and seasonal river discharge forecasts from SMHI, produced by driving the E-HYPE model with seasonal meteorological forecasts. How is the new information visualised and communicated? Through the co-creation of the CS during the I-CISK project, the decision-makers and end users’ journey through the interface was considered. Early mock-ups of the CS were presented and [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 20 discussed, providing hands-on experience using the CS from the early stages of the concept. Feedback led to including the ability to customise the charts, with positive responses from the users, who also indicated that uncertainty information would be useful, alongside ‘infoboxes’ providing further information within the CS. Some screenshots are provided below. Figure 4 Example screenshot from the Cross-Sector Planning Service for Tourism, displaying a map of the seasonal forecast of the frequency of light precipitation. A map is provided for Crete, and clicking on a location indicates the frequency for the current month and shows a graph of the forecast frequency over the next seven months. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 21 Figure 5 Example screenshot from the Cross-Sector Planning Service for Tourism, displaying a map of the seasonal forecast for the ‘tourism climatic index’. A map is provided for Crete, and clicking on a location indicates the current value and shows a graph of the forecast over the next seven months. Figure 6 Example screenshot from the Cross-Sector Planning Service for Tourism, displaying a map of the seasonal hydrological forecast. A map is provided for Crete, and clicking on a subcatchment displays shows a box plot graph of the probabilistic forecast over the next seven months. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 22 Figure 7 Example screenshot from the Cross-Sector Planning Service for Tourism, displaying a map from the climatic part of the service, indicating the current and projected average summer temperature. A map is provided for Crete, and clicking on a location displays the current mean summer temperature, and displays a graph indicating the projected change in mean summer temperature out to 2090. The map can be changed to display the mean summer temperature in future decades. Feedback on the new CS suggests that it improves on the information available through existing CS by providing more comprehensive charts that are based on and tailored to user needs, alongside the ability to modify the charts using information the user has added, allowing for quick customisation of the information based on sectoror user-specific needs. Based on the identified need for uncertainty information, and feedback received during the cocreation process, the LL communicated with users regarding uncertainty and probabilistic information. This was found to be challenging to communicate, identifying the need to build a common understanding of uncertainty, with dedicated time for these discussions. The CS will provide probabilistic and uncertainty information, with particular interest from users in information on ‘best- ’ and ‘worst-case’ scenarios. Discussion with users indicates that the uncertainty information is generally considered to be very useful, but if the uncertainty is large it can discourage the user. Other users prefer not to have uncertainty information, and more work is needed around how to use the information. How does the new CS address the identified barriers and needs? The newly-developed CS provides tailored information in a multi-sectoral framework, making information available across sectors allowing for links between them, and provides a large range of sector-specific and tailored variables, addressing the barriers related to sector-specific information and information that was not tailored to the user needs. It provides a range of information on different forecast horizons and time periods, and incorporates elements of uncertainty information and hazard severity, while being designed in a way that is intended to be accessible for non-expert users. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 23 Overall, the service will support evidence-based decisions in the tourism, water and transport sectors, facilitating improved tourism planning, better informed decision making, and improved water resource planning and efficiency, leading to a more climate resilient tourism sector. 4.3 Hungary Figure 8 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in Hungary. What is the decision-making context? The focus of the Hungarian LL is on urban heat islands, in particular in Erzsébetváros, an inner district of Budapest and the most populated. This district has a low percentage of green spaces, with a high density of buildings, and therefore is particularly exposed to heatwaves, which impact a range of sectors in the city. In Erzsébetváros, small businesses are a key part of the economy, which is based primarily around tourism. Alongside the impacts of heatwaves on the tourism sector, a significant impact in Budapest is on the health sector. Key motivations for this LL include the consequences of climate change that are observed and projected, including an increase in the mean annual temperature and sunshine duration, alongside more frequent temperature and precipitation extremes. In Budapest, the urban heat island effect exacerbates the impacts of summer heatwaves, with temperatures in inner parts of the city reaching up to 7oC above the greener areas surrounding the city (Masih, I., van Cauwenburgh, N., et al., 2022 [D1.1]; Budapest SECAP, 2021). The district aims to implement adaptation strategies including increasing the percentage of green areas (such as green roofs, green walls), shading buildings and adding drinking fountains or other places to provide water during heatwaves, alongside developing a heatwave alarm system and educating the public on adaptation strategies (Masih, I., van Cauwenburgh, N., et al., 2022 [D1.1]). [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 30 Figure 11 Example screen shot from the Italian LL climate service mock-up, displaying the map and station location (left) and ensemble seasonal forecast of river flow with daily timestep and with thresholds marked (right). Figure 12 Example screenshot from the Italian LL climate service mock-up, displaying the map and station location (left) and ensemble seasonal forecast from GloFAS with a monthly timestep, alongside climatological percentiles, displayed as a box plot (right). During the iterative feedback sessions, it was highlighted that there were challenges around the complexity of ensemble forecasts, which were unfamiliar to some stakeholders. The decision was made to try to simplify the visualisation as much as possible, and focus on percentile information with options to view more information if needed. As such, the median is always shown as the default forecast information, and the user can choose to display uncertainty ranges, such as the 25th-75th percentiles of the forecast, when needed. Recent feedback suggests that there remains some confusion around the link between forecast data and historical information, and emphasises the need for further development to clearly explain this within the interface. How does the new CS address the identified barriers and needs? The new service will address a lack of streamflow forecasts for the region. Information is tailored to the sector and decision-maker, supplying the streamflow forecasts required at a point location [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 31 relevant to the end user and with decision-based thresholds and integrated local data. Forecasts with daily timesteps will be the main source of information, which were not previously available, complemented by data aggregated over longer (monthly) scales. The use of ensembles will provide an assessment of uncertainty, with the visualisation co-designed with stakeholders to reduce complexity compared to existing CS. Expected benefits include improved water allocation and reduced water shortages. In turn, this will benefit agricultural and industrial planning, as well as reducing agricultural production loss (LL survey). The CS will enable early warning and risk assessment, as users can evaluate the likelihood of river flows falling below critical ecological thresholds, helping with decisions on when to activate measures such as irrigation restrictions. It will support flexible and forecast-based decision-making that is both precautionary and proportional to the forecasted risk, with uncertainty information allowing planning for different scenarios (best-case, worst-case) – important information in managing drought risks and complying with regulatory requirements. 4.5 Lesotho Figure 13 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in Lesotho. What is the decision-making context? The Lesotho LL is focussed on Disaster Risk Reduction (DRR), Anticipatory Action (AA) and humanitarian aid for drought, water scarcity and cold waves. The ambition of the LL within I-CISK has been to support a move from reactive to proactive drought and cold wave impact management in the humanitarian sector. Lesotho is reliant on rainfed agriculture, which is heavily affected by recurring droughts. Agricultural and socio-economic droughts are the most impactful, going beyond rainfall patterns to impact water availability for agriculture and human consumption. Rainfall projections under a changing climate are variable, however temperature is expected to increase (Masih, Van Cauwenburgh, et al., 2022 [D1.1]). [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 32 In addition, people living and working in highland areas are vulnerable to cold and snow impacts; although temperature is likely to rise with climate change, a transition to less frequent snow/cold events could mean that individuals and communities are less aware of the risk and hence become more vulnerable. What existing CS were already in use at the start of the I-CISK project, if any? The Lesotho Meteorological Service (LMS) has access to global and regional meteorological forecasting products, global hydrological models and crop monitoring services. From these, LMS provides customised forecasting at a range of lead times. The most frequently mentioned CS during the initial interviews were the seasonal forecasts from LMS. The CS already in use at the start of the I-CISK project include: ● Meteorological forecasts of precipitation and temperature for the next 6 to 24 hours from LMS, provided by radio, television, social media and mailing list. ● Seasonal forecasts of precipitation and temperature for the rainy season from LMS. Released in September with an update in January using observed data for the season so far, distributed through workshops, press and mailing list. ● Agromet bulletins from LMS providing rainfall, temperature and vegetation index (NDVI) with both 10-day and 3-month forecast horizons. ● Climate projections from LMS covering the periods 2011-2040, 2041-2070 and 2071-2100. ● Crop Monitor for Early Warning providing maps of crop performance and yield through GeoGLAM. ● Seasonal soil moisture outlooks from NHyFAS (the NASA hydrological forecast and analysis system), for maps of root zone soil moisture out to three months ahead. ● River flow forecasts from GloFAS with daily time steps and seasonal outlooks. ● Socioeconomic outlooks from FEWSNET (the famine early warning systems network) IPC providing bulletins on food security classification and the seasonal soil moisture forecasts, with information on the current situation and a forecast out to 4 months ahead. What were the barriers to using existing CS? Service discontinuity and infrequent forecast updates A key challenge discussed during the initial interviews is the rapid change in weather patterns observed in Lesotho. Through further discussion, it was understood that weather forecasts are not updated frequently enough and are not detailed enough to capture some events. Reliance on external information Participants mentioned that a barrier to the use of CS can be a lack of in-house information, which results in the stakeholders relying on external data and services. Visualisation The user experience of CS was a concern expressed during the initial interviews. Two participants found that user interfaces and product visualisations were difficult to use and required expert knowledge and training. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 33 What needs were identified for improved and tailored CS? Longer lead times For cold wave forecasts, longer lead times were required to aid decision-making and anticipatory humanitarian action ahead of an event, based on forecast information. Existing lead times were not found to provide enough time for some relevant actions to be taken before an event. For drought, information on potential drought during the rainy season is required earlier than is currently available to enable preparedness activities. Centralised information The need for users to be able to find all the relevant information from a centralised location was identified, as much of the existing information required finding information from a range of sources, adding complexity to the decision-making process. What information does the new CS co-created through I-CISK provide? The new CS for drought risk aims to provide an interface for users, with a focus on tailored information, and is expected to contribute to informed decision making for DRR and agriculture by supporting early action and adaptation policies. The new CS makes use of seasonal precipitation forecasts from ECMWF’s SEAS5 forecasting system. These forecasts are processed to calculate the tercile probabilities (below, near or above average) of precipitation throughout the rainy season, and these probabilities are translated into a binary classification of ‘drought risk’ or ‘no drought risk’, based on thresholds defined in the Lesotho Red Cross’ Early Action Protocol. The forecasts are also spatially aggregated per district, to provide information at relevant scales. Expert users are also able to access the probability information as a separate layer. Beyond the forecasts, the new CS also uses population data aggregated over the same districts and information on the location of Red Cross branches. To tailor the information further, users can manually input data based on seasonal precipitation forecasts from LMS using the same binary classification of drought risk or no drought risk. This allows comparison of two seasonal outlooks. The information is tailored to the needs of the users. For example, some layers, such as population and Red Cross branch locations, were specifically requested by users. The CS is also designed to align with existing decision-making processes, particularly as outlined in the Early Action Protocol, which defines who has the authority to issue forecasts and act on them. While forecasts data from a global model are used for comparison, the required input of the official seasonal outlook from LMS ensures that actions are confirmed and triggered based on locally-relevant information, in line with national protocols. This CS does not directly include information on uncertainty, as it is not currently incorporated into the existing decision-making processes such as the Early Action Protocol. While the forecast data used to compute the tailored information is an ensemble forecast and the lower tercile of precipitation probability is considered for assessing the drought risk, the drought classification is binary as this is the current need of the users, although as noted above, ‘expert users’ can choose to access probability information. How is the new information visualised and communicated? The new CS provides an intuitive map-based interface that visualises both forecast and impact data. Additionally, the CS sends notifications by email (and potentially WhatsApp), ensuring timely and wider dissemination. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 34 The navigation of the CS by users was considered throughout the co-creation process. Interviews with users allowed the CS developers to understand existing workflows and specific needs. Once a prototype CS was available, it was tested with users to ensure it aligns well with real-word processes and effectively supports users. Accessibility of the information was also considered, with the CS using accessible font sizes and colour schemes. For effective and accessible communication, a combination of icons and colours is used to convey key messages, such as alerts. Feedback from users confirms that the visualisations used are generally clear and easy to interpret, the different sources of information are both understood and trusted, and the various user roles and their needs are well-defined. It also identified some areas for further improvement before the CS is finalised, such as improved colours of rainfall data layers, and clearer instructions for use. Future updates will also see the addition of early action information in the CS. A user guide will also be developed, which will be used for training purposes, and users of the new CS will receive training. Figure 14 Screenshot from the Lesotho LL CS, indicating what users will see if there is a warning for drought in several districts, based on the ECMWF SEAS5 forecast. Icons indicate warnings for the upcoming six months, and the interface highlights that a trigger is expected for drought, the exposed districts and the associated populations. This is based on mock data and does not indicate a real forecast. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 35 Figure 15 Screenshot from the Lesotho LL CS, indicating what users will see when clicking on a specific district from the display shown in figure 14. The exposed population in the chosen district is shown, and an advisory to inform potentially exposed districts and follow local protocols. Figure 16 Screenshot from the Lesotho LL CS, displaying an additional layer that users can choose to visualise if they are interested in the probability of drought according to the global forecast from ECMWF’s SEAS5 seasonal forecasting system. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 36 Figure 17 Screenshot from the Lesotho LL CS, indicating what users will see if they choose the option to ‘set trigger’ – this option allows the user to select the districts where the Red Cross Early Action Protocol should be activated. Figure 18 Screenshot from the Lesotho LL CS, indicating what users will see when the trigger has been set. The districts highlighted in red are at risk of drought, according to the official seasonal forecast from LMS. How does the new CS address the identified barriers and needs? A significant benefit is that the new CS brings together all relevant data in one integrated platform, addressing the need for centralised information. A key aspect of the new CS is that it will provide information in a timely, easy-to-understand fashion that indicates not only the weather but its potential impacts, enabling early action to be taken. The CS adds a new layer of forecast information from another forecast centre (SEAS5 from ECMWF), in addition to the existing seasonal forecast information from LMS which can be input manually, allowing comparison of multiple forecasts to aid in assessing confidence in the predictions, and extending the lead time of forecasts to ensure there is information available throughout the rainy season – a key gap and need identified in the existing services. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 37 Previously, there was also no visualisation associated with the seasonal forecasts from LMS. The new CS addresses this gap by allowing input of the seasonal forecast from LMS to display it as a map, alongside other relevant information aggregated over the same districts. The new CS is expected to contribute to informed decision-making processes that also include impact data, for disaster management agencies and the agricultural sector, and support the transition from reactive to proactive disaster risk management. This will support early action and adaptation strategies, and reduce the impacts of drought on food, water and livelihoods in local communities. While a new CS has not been developed for cold waves, support has been provided through the I-CISK project to co-explore ways to improve the existing cold wave forecasts provided by LMS. For example, research undertaken through I-CISK has assessed the forecast performance of cold wave forecasts from ECMWF at different lead times. Forecasts from ECMWF were already in use by LMS, and other potentially useful forecast products were discussed and identified through this collaboration. The results of this work help to address the need for cold wave forecasts at longer lead times, by better understanding the ability of forecasts to predict these events, and at which lead times the forecasts can be reliably used for decision-making. Baugh, Egan, et al., 2025 [D3.4], ‘assessment of existing and tailored CS using a range of user-driven evaluation metrics’, provides further information. 4.6 The Netherlands Figure 19 Overview of the barriers to the use of existing CS identified at the start of the project, the needs for improved CS, and how the newly-developed CS addresses these barriers and needs, for the LL in the Netherlands. What is the decision-making context? The Netherlands LL is focussed in Rijnland, a sub-region of the Rhine delta in the Netherlands where the Rijnland water authority is responsible for governance of the surface water system and waste water treatment. Surface water is highly controlled in Rijnland and is used for both irrigation and drainage. The main climate hazards in this region are pluvial flooding and drought, with drought being the key focus for this LL. Droughts strongly impact water tourism (by limiting recreational shipping) and agriculture (by affecting irrigation water management and water quality). In particular, during spring and summer, local meteorological drought in combination with low flow in the Rhine can cause issues with salinity. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 38 Examples of the decision-making required include whether to limit ship-lock operation, and management decisions such as changing drought alert levels, inspecting dikes, planning for staff availability, and starting or increasing the frequency of drought committee meetings regarding the activation (or not) of the alternative fresh surface water supply route due to the primary source of freshwater becoming too saline. Other relevant decisions are those involving long-term investments for ensuring the water system is robust in a changing climate. What existing CS were already in use at the start of the I-CISK project, if any? At the start of the project, CS were already being used by the water board for monitoring and operating the water system according to current and future forecast state up to 15 days ahead. These were used to support decisions for a range of activities including emergency management, short-term planning and daily decision-making, and weekly monitoring and prediction. Information already in use included: ● Drought bulletins provided by the water board of Rijnland, using forecasts from KNMI, ECMWF, MeteoGroup and the HydroLogic water balance model. The bulletins assess the drought condition and provide drought warnings colour-coded according to water use restrictions, as well as graphs and forecast plumes. An associated monitoring network monitors precipitation, potential evapotranspiration, water levels, pumped discharge, inflow and salinity. Forecasts include spatially distributed potential rainfall deficit and river discharge for the Rhine. The bulletins are produced every week and provide forecasts out to 14 days ahead. ● Streamflow monitoring and prediction for the Rhine at Lobith. Rijkswaterstaat provides river discharge monitoring and 15 day predictions at this location. What were the barriers to using existing CS? Stakeholder engagement The CS available is produced for the water authority of Rijnland, lead stakeholder of the LL. Water tourism and agricultural sector representatives were not necessarily aware of the various CS and climate information available. Limited lead times Current CS provide forecasts out to 14 days ahead, resulting in challenges for decision-making that relies on information on timescales longer than 2 weeks ahead. Manual reporting The drought monitor reports are circulated and used widely by water authorities, key water-related stakeholders in the area, and the general public. A challenge identified for this CS is that the report is generated manually, and is typically a long report aimed at including all relevant information for a wide range of stakeholders. Uncertainty communication In general, information provided by the CS is ensemble-based, and presented using uncertainty plumes. A challenge highlighted through the initial surveys is that guidance on the use of probabilistic forecasts, for example for the alert levels, is not available. Negative consequences of action While not a barrier to the use of a CS itself, participants from the Netherlands LL highlighted that a challenge associated with decision-making based on CS is that some drought management measures [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 39 negatively affect some of the stakeholders (for example, recreational water use and shipping), while taking measures too late can negatively impact other stakeholders (such as agriculture). What needs were identified for improved and tailored CS? Seasonal, sub-seasonal and climate change information Longer lead times would allow stakeholders to prepare better for upcoming drought events and, by incorporating climate change information, develop climate adaptation plans that account for the frequency and severity of droughts anticipated in the future. Forecast skill information Information on forecast skill is limited beyond 2 days, and additional information on forecast skill at a range of lead times would be beneficial, to supplement the general knowledge that forecasts are uncertain and that misses and false alarms can occur. Stakeholder engagement and co-creation Within the I-CISK project, the water authority of Rijnland aims to strengthen communication and engagement with other sectors and stakeholders such as those from the agriculture and water tourism/recreation sectors, in order to co-create tailored CS that are informative for their decisionmaking processes. What information does the new CS co-created through I-CISK provide? The Netherlands LL has developed a ‘Drought Awareness Service’ (https://icisk.dev.52north.org/living-labs/rijnland--nl/app/) with two sub-services: one for upcoming drought forecasts (including streamflow and potential precipitation deficit forecasts) and the other assessing drought in climate change projections (covering various parameters of interest to users e.g. streamflow, temperature, precipitation). Two new sub-seasonal to seasonal drought forecasts are included, one based on potential precipitation deficit and the other on River Rhine discharge. Streamflow forecasts in the new service are provided for the Lobith station, where The Rhine enters The Netherlands, which is used as the key indicator for low flows and drought in Rijnland. The new CS makes use of a range of data, including observed precipitation and temperature. The seasonal river discharge forecasts are based on ECMWF SEAS5 meteorological (precipitation and temperature) forecasts, which have been bias-corrected using HydroGFD and used to drive the E-HYPE hydrological model. The bias-corrected meteorological forecasts are also used for predicting the cumulative precipitation deficit. The bias correction step is seen to increase trust in the forecast by limiting the deviation at the initial time. The CS users will also receive information on forecast performance/skill, including the maximum lead time of positive skill (using climatology as a reference), contingency tables and graphs of re-forecasts compared to historical observations. The stakeholders in the LL were presented with a range of forecast evaluation metrics and graphs, and these choices represent the preference of the stakeholders for the skill information most relevant to their decision-making. How is the new information visualised and communicated? A range of visualisations are used in the new CS, which has been developed with the user’s journey through the CS interface in mind. For example, the welcome page of the CS provides a simple selection process allowing the user to select their sector, and then the type of information they are interested in, including drought alerts, river discharge forecasts, precipitation deficit forecasts and climate [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 46 Actors involved Role Type of organization Sector Climate value chain CS produced (publicly available) CS used Network of meteorological stations RIA (Red Información agroclimática de Andalucía) ADROCHES Local rural development association for the Pedroches region Not for profit association Rural development End-User Not applicable CSIC drought monitor: monitordesequia.csic.es Traditional knowledge AEMET 15-day predictions OLIPE Olive growers cooperative (organic and conventional) Farmer organization and food processing industry Agriculture & agroindustry End-User Not applicable Cabañuelas (traditional predictive system) Experience AEMET Weather forecasts Regional and local weather stations COVAP Cooperative of ranchers & farmers that produces milk and meat products Rancher organization and food processing industry Livestock, feed production, and agroindustry Provider End-User Network of meteorological stations and a network of environmental stations on livestock farms TV and radio weather reports 15 day predictions – AEMET and eltiempo.es Traditional knowledge Data from own meteorological network CICAP Research and milk production quality control organization R&D associated with COVAP Agriculture and livestock Provider End-User Network of meteorological stations and a network of environmental stations on livestock farms 15 day predictions – AEMET and eltiempo.es TV and radio weather reports Traditional knowledge Data from own meteorological network Federación Andaluza de Caza (Andalucía Hunting Federation) Federation of small game hunters in Andalucía Hunting not for profit association Forestry / hunting End user Not applicable TV and radio weather reports 15 day predictions – AEMET and eltiempo.es INFOCA fire risk reports Monthly drought reports from RBA [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 47 What were the barriers to using existing CS? Dissemination and communication, tailored information During initial interviews conducted when establishing the LL, REDIAM, who integrate and disseminate environmental information from various providers, highlighted concerns that their CS are not reaching the target audiences as desired – potentially due to a combination of lack of effective dissemination and communication, and lack of tailored information for the needs of local users. Stakeholders in the agriculture sector noted that a lack of specialised information, and slow communication regarding drought, cause challenges for decision-making. Another limitation is the inaccessibility of information from other sources, such as river basin authorities or drought reports. Forecast uncertainty The uncertainty of forecasts is noted as a challenge in all questionnaire/interview responses, from the forestry, agriculture and rural development sectors; some forecasts/CS are too uncertain to support decision-making processes. Insufficient spatio-temporal resolution While the 2-week forecasts from AEMET and elTiempo.es, for example, are considered to be generally reliable with sufficient resolution, other CS, particularly longer-range predictions, do not have spatial or temporal resolutions adequate to support decision-making. Access to historical data Lack of access to historical data relevant to the decision-making was also noted as a key challenge at the start of the project. Limited knowledge of groundwater dynamics The Pedroches aquifer system is a highly fractured low productivity aquifer but critical for extensive livestock production and the health of the dehesa agroforestry system. There is little information available and a sparse monitoring network. Aquifer dynamics and its relationship to surface water and climate are, however, poorly understood. What needs were identified for improved and tailored CS? Sector-tailored information The climate information currently available is not tailored to stakeholders needs and the information in forecast and projections does not take impacts into consideration (e.g. expected acorn production in the coming year, or water availability under different drought scenarios). Tailored CS would help farmers adjust their plans and estimate productivity, manage water availability and plan management of activities such as livestock stocking rates or purchase of additional feed. Key variables of interest include rainfall patterns (seasonal and monthly distribution, yearly accumulations), and the start and duration of summer and winter seasons. Improved spatio-temporal resolution Existing forecasts are available up to 7-14 days; however, stakeholders noted that the addition of CS covering longer timescales (sub-seasonal, seasonal, annual) would be useful to make informed decisions. For example, longer-range forecasts would allow farmers to adapt (reducing numbers of livestock, when to harvest etc), and would assist in planning forest management activities. In addition, the spatial resolution of existing information from e.g. climate projections is seen as too coarse (e.g. only international information is available) to be informative at the scale of the LL, and would benefit from being downscaled to more local levels / regions given existing spatial variability. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 48 Historical data and climatology Easy access to historical data (such as precipitation, temperature, runoff generation, vegetation phenology etc.) would allow stakeholders to identify trends / confirm observations of trends and help them to make decisions with or without forecast information. Alongside this, information on climate and hydrological characteristics based on historical data would be useful. Uncertainty and skill information Questionnaire/interview responses from the forestry and agriculture sectors highlighted a need for reducing the uncertainty in the forecast information, and the response from the rural development sector indicated an interest in improved availability of uncertainty and skill information for available CS. What information does the new CS co-created through I-CISK provide? To address these needs, the Spanish LL has developed a ‘Climate Planning Service’ (https://icisk.dev.52north.org/living-labs/guadalquivir--es/), which has five main sub-services: 1) climate projections, 2) seasonal forecasts, 3) historical climate information, 4) an olive phenology service and 5) groundwater characterisation. This information is designed to be tailored to the decision-makers and intended use, and as such, climate projections for monthly temperature and precipitation out to 15 years are provided at a spatial resolution of 1km, seasonal forecasts are provided monthly out to 6 months ahead for temperature and precipitation at a spatial resolution of 250 m. The olive phenology service models the impacts of climate change on olive production, using indicators of relationships between phenology of olive trees and climate. Historical data is also rendered into various formats of relevance to the end user. For example, 10-day precipitation accumulations are provided, which give a measure of possible water shortage, and the number of days on which the temperature exceeds 25°C provides information on heat stress or conditions for optimal growth of crops. CS 1 to 4 (climate projections, subseasonal to seasonal predictions, historical data and olive phenology) make use of historical monthly values of temperature and accumulated precipitation, from AEMET, and CS 4 also uses data on olive tree production from the local cooperative OLIPE, while CS 5 (hydrogeology characterisation) uses piezometric data from two public piezometers managed by the Guadiana River Basin Authority, data gathered through five fieldwork campaigns for well measuring, as well as data from grey literature, past research projects, and local knowledge. In addition to the observed data, seasonal forecasts of temperature and precipitation from ECMWF’s SEAS5 forecasting system are used for CS 2 (subseasonal to seasonal forecasts), with a bias correction methodology applied using observed meteorological data, followed by a linear regression and residual interpolation processing method to create higher-resolution (250m) maps, driven by the user needs for improved spatial resolution of information due to a limited network of meteorological stations in the Pedroches region. CS 1 (climate projections) uses climate projection data from the CODREX-EUR11 dataset, using the RCP 4.5 scenario, and the information is downscaled from 11km spatial resolution to a 1-2km spatial resolution. Detailed information on these methodologies can be found in the deliverables from Work Package 3. How is the new information visualised and communicated? Throughout the co-creation of the new CS for the Spanish LL, online and in person meetings and workshops were held with stakeholders, either bilaterally or in a group setting, to present the CS and discuss the visualisation. Input and feedback were gathered at each meeting to guide the development and improvement of the CS throughout the process. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 49 The discussions and feedback covered topics such as uncertainty and how to convey and visualise it effectively, the need to ensure sustainability of the CS beyond the project, the need to homogenise the language of the CS with those used in other CS from AEMET, and the need for open data. On the topic of uncertainty, feedback from some users indicated that when there are high levels of uncertainty in the data, the information should not be presented or made available. Figure 25 Screenshot of the Spain LL CS 2, showing a seasonal precipitation forecast. The map displays the median of the forecast ensemble, with seasonal bias correction and downscaling applied, showing a spatial resolution of 231m. A toggle allows the user to display the associated uncertainty (see Figure 26). [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 50 Figure 26 Screenshot of the Spain LL CS2, showing a seasonal precipitation forecast. The map displays the uncertainty in the forecast, which has been bias corrected and downscaled to a spatial resolution of 231m. Clicking on a point on the map shows the 5th to 95th percentiles of the forecast, providing an assessment of the forecast uncertainty at that location. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 51 Figure 27 Screenshot of the Spain LL CS 3, showing historical climate information. The map displays temperature, precipitation or drought indices, and allows comparison of two different time periods using the dropdown menus at the top, and the slider underneath the map. Clicking on a location displays a time series of the data at that location. Another map layer provides temperature and precipitation time series data at individual stations. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 52 Figure 28 Screenshot of the Spain LL CS, showing agroclimatic indicators. The map displays the maximum number of consecutive dry days (<1 mm of rain) per season, for the years 2011 to 2040. This CS is still under development and the final service may be displayed differently based on the upcoming development and feedback from the stakeholders in the LL. How does the new CS address the identified barriers and needs? These new CS address the lack of access to information with sufficient level of detail, spatiotemporal resolution, and tailored information. It addresses the lack of access to historical data and, alongside the development of bespoke services such as olive phenology indices, linking phenological data and climate data for the region for the first time. The Los Pedroches region had limited climate information and relied on 15-day forecasts from the National Met Service (AEMET) or online weather services. These are also based on a limited number of stations in the area, so the provision of historical, extended-range forecast and climate projection data with increased resolution will address this. The service will also provide or account for forecast/projection uncertainty. Overall, the service is expected to help reduce the impacts of climate variability/change, building resilience and reducing vulnerability by aiding evidenced based decision making and supporting sustainable management. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 53 5 Overarching remarks 5.1 Barriers and needs Across the seven LLs—The Netherlands, Spain, Italy, Greece, Hungary, Georgia, and Lesotho— common barriers to the effective use of existing climate services (CS) emerged despite the geographic, climatic, and sectoral diversity. These included: ● Lack of sector-specific tailoring: A predominant challenge was the provision of generic CS that did not cater to the specific needs of sectors like agriculture, water management, tourism, and health. This was noted in almost all LLs, where users found existing CS to lack actionable detail and relevant variables. ● Limited spatial and temporal resolution: Many stakeholders noted that forecasts were either too coarse spatially (e.g., country-level rather than regional or catchment-level) or temporally (e.g., aggregated weekly or seasonally, missing important intra-period variability). ● Poor accessibility and usability: CS interfaces often required technical expertise, lacked intuitive visualisation, and were not always openly accessible. Barriers such as language, absence of centralised platforms, and non-continuous service provision further hindered access, especially among local and non-expert users. ● Lack of trust and understanding: Trust was undermined by unclear communication of uncertainty and skill. Users struggled to interpret probabilistic forecasts or understand how CS linked to historical conditions, leading in some cases to avoidance of or scepticism towards CS. ● Fragmentation and over-reliance on external sources: In some contexts (e.g., Georgia, Lesotho), stakeholders relied on fragmented sources or external tools, making consistent decision-making difficult These barriers led to the following set of overarching needs that apply to most, if not all, of the LLs, and can be considered as a list of recommendations to CS producers to consider in the creation of new CS and information to ensure relevance to decision-makers, depending on the context. The newlydeveloped CS would need to be: ● Locally relevant: Stakeholders expressed a need for CS with higher spatial resolutions for improved local relevance, and that include variables directly relevant to their decisions, such as streamflow, soil moisture, heat indices, and sector-relevant indicators. ● Multi-hazard and cross-sectoral: There was a need to capture compounding and cascading effects of hazards (e.g., drought, heatwaves, flooding) and their cross-sectoral implications. There remain challenges associated with addressing this need, but incorporating a range of information on all relevant hazards in one centralised CS can be useful as a starting point, which also addresses the identified barrier of fragmented information from various external sources. ● Proactive and early warning-oriented: There was a consistent need for CS that could support a transition from reactive to anticipatory action, and often this means providing improved temporal resolution / more frequent and regular updates and relevant temporal aggregations, and longer lead times. ● Integrated with local data and workflows: Across LLs, stakeholders stressed the need for integration of local observations and compatibility with existing planning frameworks and policies (e.g., drought management protocols). ● Accompanied by guidance and capacity building: Users highlighted the importance of documentation, training, and support to ensure CS usability, particularly around interpreting uncertainty. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 54 ● Clear and effective visualisation and communication: Visualising information clearly and providing clear information on forecast uncertainty and skill can help users to build trust in the information through confidence in the understanding of the information presented and the benefits and limitations of the CS for the intended use. Considering accessibility and language of the information provided is also essential. 5.2 The new climate services - commonalities and differences Table 3 Overview of information provided by the new CS co-created under the I-CISK project. LL Variables Type Temporal Resolution Spatial Resolution Forecast lead time Forecast update frequency Georgia - Streamflow - Standardised precipitation index - Temperature - Precipitation Historic and Forecast Daily, monthly, bi-monthly and 3-month Sub-catchment, point locations Up to seasonal Monthly Greece - Large range of tailored indices based on temperature, precipitation, surface water discharge and wind speed/direction Historic and Forecast Daily, fortnightly, monthly, 4 month Sub-catchment, point locations, gridded Up to climate projection Monthly Hungary - Urban mapping and thermal imaging - Heat source information Historic N/A Individual property level N/A N/A Italy - Streamflow Historic and Forecast Daily Sub-catchment, point location Next weeks, up to 3-month lead Monthly to start, weekly optimal Lesotho - Precipitation (droughts) - Population exposed Forecast Seasonal (droughts) District level Up to 6 months Monthly Netherlands - Streamflow - Temperature - Precipitation Historic and Forecast Daily, weekly, monthly, TBC Point location and gridded data Up to climate projection Monthly Spain - Temperature - Precipitation - Agroclimate indicators - Groundwater (spatial description of the aquifer) and groundwater level monitoring - Olive phenology Historic (scientific and local knowledge) and Forecast Monthly (Temperature and Precipitation) Seasonal and 10-day (agroclimatic indicators) Yearly (olive phenology) Point locations and gridded data (250 m for historical data and seasonal forecast, 1-2 km climate projections) 50 km (agroclimatic indicators) Groundwater modelling n/a Up to climate projection Monthly for forecasts [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 55 Within most LLs, end user needs are diverse and two main strategies have been employed to deal with this. Some LLs, for example Georgia, Italy and The Netherlands, focus primarily on one main information type (streamflow), and a primary end user who makes decisions that impact upon other ‘downstream’ beneficiaries. These ‘beneficiary’ users gain from a service that supports decision making by the primary user but may not use the service directly themselves. For example, better planning and management of water resources benefits users ranging from farmers to hydropower plants and recreational boaters (Table 1 and Table 3). In other LLs, for example Greece and Spain, several bespoke indices are brought together under one ‘umbrella’ service for a given sector or sectors. This alternative approach to diverse needs provides streams of data that may be used directly by multiple end users. For example, data provided in the Greece LL service range from bespoke wind indices for port management through to surface runoff and reservoir storage for water management, and in the Spain LL service examples include bespoke olive phenology and agroclimate indicators, and groundwater data. CS information provided by the Georgia and Italy LLs is very similar, and has strong parallels with the Netherlands’ service (Table 4). These services also show commonality with streamflow provision in other locations and for other uses (e.g. flood forecasting), in that they provide ensemble streamflow forecasts overlaid with various thresholds of interest. This convergent evolution of services, based on consultation with different stakeholders, in some cases in different fields (e.g. flood vs. drought), suggests there may be scope for a transferable framework that others in the field, and perhaps even other fields, could utilise. However, some caution needs to be exercised to avoid influencing end user ‘needs’ with existing provision. Parallels exist between the Lesotho and Spanish LLs, in terms of natural hazard and aspects of vulnerability. In both locations, there is a reliance on rainfed agriculture, which increases vulnerability to drought and rising temperatures. As a result, there may be similarities in the types of information that could be of use. However, user requirements in the two LL are different; the focus in Lesotho on preparedness for DRR and Anticipatory Action would suggest that a smaller number of main users (The Lesotho Meteorological Service, The Lesotho Red Cross Society) will make decisions that benefit downstream ‘beneficiary’ users. Conversely, the Spanish LL CS is more directly usable by farmers and co-operatives. A key need for existing CS improvement identified across the LLs was the use of appropriate spatial and temporal resolutions, for both forecast and historical data. Methods of addressing this within the new CSs include downscaling of data to a more useful gridded resolution, aggregation of data over appropriate regions of interest or more useful time periods or providing increased resolution timesteps. Another common user requirement is access to relevant historical data. Within the new CS, provision of such data not only directly addresses this, but is also one of three main methods used to incorporate, or facilitate the use of, local knowledge. Historical data is important in this respect because it allows the user to contextualise current or forecast events using their own lived experiences. Two other methods for integrating local knowledge, demonstrated in the new CS, are the inclusion of user-defined decision/indicator thresholds (e.g. streamflow values or Early Action Protocol trigger thresholds) and provision of bespoke indicator forecasts identified by the user (such as the tourism indicators in the Crete CS). Both methods are used to indicate when events might be impactful, and therefore help users assess when they need to act. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 62 6 Conclusions This deliverable has synthesised the insights gained throughout the I-CISK project regarding the use, needs, and development of climate services in seven diverse Living Labs. The work confirms that while climate services have expanded in availability and technical sophistication, significant gaps remain in usability, relevance, and accessibility particularly at the local and sectoral level. Through iterative engagement, stakeholder participation, and co-development of new CS prototypes, I-CISK has made significant strides in addressing these gaps. The newly developed services are tailored to specific user needs, integrate local data, consider user workflows and decision contexts, and incorporate feedback on visualisation and interface design. Key contributions of this work include: ● A systematic identification of shared and unique barriers to climate service use across multiple contexts. ● A set of concrete user-driven needs that inform the development of future-generation CS. ● Demonstrated success in applying participatory methods to co-design usable, contextsensitive, and actionable climate services. ● Valuable lessons on communicating uncertainty and supporting decision-making through fitfor-purpose visualisation. Despite the progress made, further efforts are needed to: ● Expand the integration of uncertainty and skill metrics in ways users can understand and apply to their specific decision-making context. ● Continue capacity building and support for both providers and users to foster effective use and trust in CS. ● Establish stronger links between CS and policy/action frameworks at local and national levels. The findings and tools developed through I-CISK provide a strong foundation for future climate service innovation, and a clear pathway to achieving more user-centred, impact-oriented, and inclusive climate adaptation solutions. [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 63 References Bagli, S., Demmich, K., Gräler, B., Mazzoli, P., et al., 2023 [D5.1]. I-CISK platofrm - technical specification, I-CISK deliverable 5.1, available online at www.icisk.eu/resources Bagli, S., et al., 2024 [D5.2]. Climate data and back-end components and front-end components, I-CISK deliverable 5.2, available online at www.icisk.eu/resources Baugh, C., Egan, K., et al., 2025 [D3.4]. Assessment of existing and tailored climate services using a range of user-driven evaluation metrics, I-CISK deliverable 3.4, available online at www.icisk.eu/resources Budapest SECAP, 2021. Budapest climate strategy and sustainable energy and climate action plan, The Municipality of Budapest, Department for Climate and Environmental Affairs, available online at https://archiv.budapest.hu/sites/english/Documents/BP_klimastrategia_SECAP_EN_final.pdf [last accessed 25 June 2025] Crameri, F., Shephard, G. E., & Heron, P. J., 2020: The misuse of colour in science communication. Nature Communications, 11(1). https://doi.org/10.1038/s41467-020-19160-7 Egan, K., Emerton, R., et al., 2025 [D2.4]. Information on climate service needs and gaps, I-CISK deliverable 2.4, available online at www.icisk.eu/resources Emerton, R., et al., 2018: Developing a global operational seasonal hydro-meteorological forecasting system: GloFAS-Seasonal v1.0, Geoscientific Model Development, 11, 3327-3346, https://doi.org/10.5194/gmd-11-3327-2018 Fuglsang, L., Hansen, A. V., Gago, D., Mergel, I., Liefooghe, C., Gallouj, F., Røhnebæk, M., Rønning, R., Lepczynski, S., Mureddo, F., & Garbasso, G., 2019: Co-VAL D5.1 Report on cross-country comparison on existing innovation and living labs (Issue Lc). Hemingway, R., & Gunawan, O., 2018: The Natural Hazards Partnership: A public-sector collaboration across the UK for natural hazard disaster risk reduction. International Journal of Disaster Risk Reduction, 27, 499–511. https://doi.org/10.1016/j.ijdrr.2017.11.014 Hirons, L., et al., 2021: Using co-production to improve the appropriate use of sub-seasonal forecasts in Africa, Climate Services, 23, 100246, https://doi.org.10.1016/j.cliser.2021.100246 Hossain, M., Leminen, S., & Westerlund, M., 2019: A systematic review of living lab literature. Journal of Cleaner Production, 213, 976–988. https://doi.org/10.1016/j.jclepro.2018.12.257 Jacobs, K. L. and Street, R. B., 2020: The next generation of climate services, Climate Services, 20, 100199, https://doi.org/10.1016/j.cliser.2020.100199 Masih, I., Van Cauwenburgh, N., et al., 2022 [D1.1]. Characterisation of the I-CISK Living Labs, I-CISK deliverable 1.1, Available online at www.icisk.eu/resources Moschini, F., Emerton, R., et al., 2022 [D2.1]. Preliminary report: Information on climate service needs and gaps, I-CISK deliverable 2.1, available online at www.icisk.eu/resources MS10, 2022: A prototype framework on co-creating end-user climate services, I-CISK Milestone Report MS10 [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ 64 Murphy, A. H., 1993: What Is a Good Forecast? An Essay on the Nature of Goodness in Weather Forecasting. Weather and Forecasting, 8, 282–293. Palmer, T. N., & Hardaker, P. J., 2011: Handling uncertainty in science. In Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences (Vol. 369, Issue 1956, pp. 4681– 4684). Royal Society. https://doi.org/10.1098/rsta.2011.0280 Rastogi, S., Werner, M and van den Homberg, M., 2024: Harnessing decision timelines to improve understanding and integration of local and scientific knowledges across the Climate Services value chain, Abstract, EGU, 24-13096, https://doi.org/10.5194/egusphere-egu24-13096 Schroeter, S., Richter, H., Arthur, C., Wilke, D., Dunford, M., Wehner, M., & Ebert, E., 2021: Forecasting the impacts of severe weather. Australian Journal of Emergency Management, 36(1), 1–76. Slingo, J., & Palmer, T., 2011: Uncertainty in weather and climate prediction. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 369(1956), 4751–4767. https://doi.org/10.1098/rsta.2011.0161 Street, R., Parry, M., Scott, J., Jacob, D., Runge, T., & European Commission. Directorate-General for Research and Innovation, 2015: A European research and innovation roadmap for climate services. Publications Office. Terrado, M., Calvo, L., & Christel, I., 2022: Towards more effective visualisations in climate services: good practices and recommendations. In Climatic Change (Vol. 172, Issues 1–2). Springer Science and Business Media B.V. https://doi.org/10.1007/s10584-022-03365-4 Van Andel et al., 2025 [D3.5]. Categorisation and evaluation of visualisation practices for communication uncertain predictions in climate services, I-CISK deliverable 3.5, available online at www.icisk.eu Van den Homberg, M., Rastogi, S., Hernandez-Mora Zapata N., et al., 2023 [D2.2]. Concepts and methods to characterise and integrate local and scientific knowledge, I-CISK deliverable 2.2, Van den Homberg, M., Rastogi, S., et al., 2024 [D2.5]. User-centred validation of climate risk knowledge integration: using decision timelines for collecting, understanding and integrating local knowledge, I-CISK deliverable 2.5, available online at www.icisk.eu WISER, 2020: Manual for Co-production in African Weather and Climate Services, 2nd Edition, Weather and Climate Information Services for Africa (WISER) and Future Climate for Africa (FCFA), Available online at: https://futureclimateafrica.org/coproduction-manual/downloads/WISER-FCFAcoproduction-manual.pdf (Last accessed 20/04/2022) [D2.4 – Climate service needs and gaps] __________________________________________________________________________________ This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 101037293 Colophon: This report has been prepared by the H2020 Research Project “Innovating Climate services through Integrating Scientific and local Knowledge (I-CISK)”. This research project is a part of the European Union’s Horizon 2020 Framework Programme call, “Building a low-carbon, climate resilient future: Research and innovation in support of the European Green Deal (H2020-LC-GD-2020)”, and has been developed in response to the call topic “Developing end-user products and services for all stakeholders and citizens supporting climate adaptation and mitigation (LC-GD-9-2-2020)”. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 101037293. This four-year project started November 1st 2021 and is coordinated by IHE Delft Institute for Water Education. For additional information, please contact: Micha Werner ([email protected]) or visit the project website at www.icisk.eu