- Norwegian Meteorological Institute, Research and Developement, Oslo, Norway (rasmus.benestad@met.no)
Climate models are not designed to provide detailed information on local rainfall that may trigger an outbreak of diarrhoea, but are nevertheless able to reproduce large-scale climatic conditions, processes, and phenomena. Hence, they have a minimum skillful scale, and downscaling makes use of skilfully simulated large-scale aspects in addition to information about how local rainfall depends on those larger scale conditions. The SPRINGS project studies the link between climate change and diarrhoea outbreak through a chain of models, where one stage provides input to the next. It’s important to design such model chains so that they provide a flow of salient and relevant information. This framework also needs to ensure robust results, as different global climate model simulations may give a different regional outlooks. It also needs to involve proper evaluation, and it's important that it is designed for both how the end-results are being used in decision-making, and that the end-results are correctly interpreted in terms of what they really represent. Here, such a framework used in SPRINGS is presented.
How to cite: Benestad, R.: Using global climate model simulations for outlooks on how climate change affects future diarrhoea risks, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9608, https://doi.org/10.5194/egusphere-egu26-9608, 2026.
Comments on the supplementary material
AC: Author Comment | CC: Community Comment | Report abuse
Hello, my name is Jeeyoun Kim, a PhD student at Yonsei University. Since your presentation was scheduled just before mine yesterday, I listened to it with particular interest. As my background is in political science rather than climate modelling, my understanding may be limited. In my recent paper, I found that climate projection tools in Korea, such as MOTIVE and VESTAP, are not always meaningfully used in local adaptation planning. Based on the SPRINGS framework, what do you think is most important for making downscaled climate information actually usable for local adaptation or public health policy?
Thanks for an excellent question, Jeeyoun Kim.
The important point is to derive reliable and robust information, which does not depend on ad hoc choices (e.g. which model run is used or what downscaling approach). Globalclimate models are out main tool for computing what the future climate will look like, but they ae only designed to reproduce large-scale aspects, whereas impact studies and climate change adaptaion require local small-scale information. Such small-scale information can be derived through downscaling, and there are several approaches, all of them with some weaknesses and lmitations. However, they are based on different assumptions and have different strengths and weaknesses, so comining several will provide more reliable and robust results. Also, a global climate model run several times with slightly differentstarting points may give quite different regional outlooks, so it's also importantto downscale many runs - so-called ensembles. It's also importantto evaluate all these results and methods to make sure that they provide a realistic representation of the aspects that we need.
Another issue is what kind of information is nedded (e.g. time series of daily rainfall or the probability of getting more than 50 mm on a day). Hence, it may be necessary to make a synthesis of different outputs and boil it down to something that is useful for decision-making. There is no perfect template for all cases, but the analysis and modelling needs to be tailored for each specific case.
For more information about downscaling tailored for decision-making, see e.g. https://arxiv.org/abs/2411.02856.
Also see WCRP regional climate information for societies (RIfS).
Thanks for an excellent question, Jeeyoun Kim.
The important point is to derive reliable and robust information, which does not depend on ad hoc choices (e.g. which model run is used or what downscaling approach). Globalclimate models are out main tool for computing what the future climate will look like, but they ae only designed to reproduce large-scale aspects, whereas impact studies and climate change adaptaion require local small-scale information. Such small-scale information can be derived through downscaling, and there are several approaches, all of them with some weaknesses and lmitations. However, they are based on different assumptions and have different strengths and weaknesses, so comining several will provide more reliable and robust results. Also, a global climate model run several times with slightly differentstarting points may give quite different regional outlooks, so it's also importantto downscale many runs - so-called ensembles. It's also importantto evaluate all these results and methods to make sure that they provide a realistic representation of the aspects that we need.
Another issue is what kind of information is nedded (e.g. time series of daily rainfall or the probability of getting more than 50 mm on a day). Hence, it may be necessary to make a synthesis of different outputs and boil it down to something that is useful for decision-making. There is no perfect template for all cases, but the analysis and modelling needs to be tailored for each specific case.
For more information about downscaling tailored for decision-making, see e.g. https://arxiv.org/abs/2411.02856.
Also see WCRP regional climate information for societies (RIfS).