In order to assess the ability of climate models to represent extreme droughts, simulations from 17 models of the CMIP6 project are compared with observations from the GRACE and GRACE-FO missions. On average, models and observations agree well, but individual simulations show strong regional deviations. Under a high-emission scenario (SSP5-8.5), the models project significantly more frequent and intense droughts by the end of the 21st century—especially in Africa, South America, and Asia. GRACE data are thus central for evaluating and improving climate models.
Text: Klara Middendorf, HCU Hamburg
How often and how severe will future droughts be? To answer this question, climate models are needed that simulate the physical processes of the Earth system. For reliable projections, these models must first be evaluated by comparing their simulations of the present and recent past with real-world measurements.
Comparing Climate Models with GRACE and GRACE-FO
The satellite missions GRACE and GRACE-FO enable measurements of large-scale mass changes on Earth. These observations reveal how the amount of water stored on land varies over time, accounting for all forms of water—soil moisture, groundwater, snow, ice, vegetation, and surface water—collectively referred to as terrestrial water storage (TWS). GRACE/GRACE-FO data thus make it possible to detect periods of unusually low water storage (drought) or exceptionally high storage (flood).
In our study, we used these observations to evaluate how well climate models reproduce the occurrence of extreme dry events. “Particularly dry” conditions are also defined as wet periods (e.g., the rainy season) that are unusually less wet. The climate models stem from the Coupled Model Intercomparison Project Phase 6 (CMIP6), a coordinated international effort to make model simulations developed by different institutions around the world comparable. Each numerical model can produce several simulations, which vary, for example, in terms of their initial conditions. Altogether, 245 simulations from 17 different models were included in this study.
To quantify the occurrence of droughts, extreme value statistics were applied to investigate the frequency with which values of a certain intensity occur, allowing the derivation of return levels. Each return level is associated with a return period, which is the average time interval between two extreme events: A 1-in-10-years return level thus represents the TWS value (or intensity) that is statistically expected to be exceeded once every ten years.
The results of individual climate models vary considerably for droughts, as small variations in model physics can lead to large deviations. However, when all models are combined to a joint multi-model result (here, using a scaled median), there is quite good agreement with the results from satellite data (see Fig. 1). Nevertheless, some models tend to underestimate drought severity in the Global South, while others overestimate it in Europe or Australia. GRACE data help reveal these discrepancies, creating opportunities to identify possibilities for improvement in model physics in collaboration with model developers.
Future Projections from Climate Models
It is particularly important to look also into the future, which climate models enable us to do, unlike the satellite data. Under a currently probable “business-as-usual” scenario with continuously high greenhouse gas emissions (SSP5-8.5), the models project on average a significant increase in drought extremes by the end of the 21st century—particularly in Central Africa, Southeast Asia, parts of South America, and northern Siberia (see Fig. 2). Overall, about 70 % of global land areas are projected to experience more intense droughts. More optimistic low-emission scenarios also indicate an increase in drought intensity when compared to present day, but to a lesser extent.
Our study highlights two points: First, satellites like GRACE/GRACE-FO are very valuable tools for validating and improving climate models. Second, the future of global water scarcity will strongly depend on which emission pathway humanity follows. ClimateIn contrast to weather, which refers to daily or very short-term events, climate refers to an average condition in the atmosphere over a longer period of 30 to 40 years. All processes such as average temperature, precipitation, wind direction, wind s... change is therefore not only reflected in rising temperatures but also in the growing risk of drought—with consequences for agriculture, drinking water supply, and ecosystems worldwide.
Reading tip
- Middendorf, K., Dobslaw, H., Jensen, L., & Eicker, A. (2025). Return levels of dry extreme events in terrestrial water storage from satellite gravimetry and CMIP6 global coupled climate models. Journal of Geophysical Research: Solid Earth, 130, e2024JB031011. https://doi.org/10.1029/ 2024JB031011