Analyzing past and future droughts that induce clay shrinkage in France using an index based on water budget simulation for trees

Abstract. Clay shrinkage is the retraction of clayey soils under dry conditions, caused by the loss of adsorbed water molecules from clay minerals. This phenomenon called clay-shrinkage induced subsidence can cause permanent damage to buildings if the drying extends below the foundations. In France, soils with these characteristics are widespread, affecting 48 % of the mainland territory (MTES, 2021), resulting in damage amounting to 20.8 billion euros since 1989 (CCR, 2023b). The causes of clay shrinkage are not yet fully understood, particularly at large spatial scales that are critical for land management. In a previous study (Barthelemy et al., 2023), a drought index designed specifically for clay shrinkage was created. It is a yearly index called the year drought magnitude. This index measures the daily soil moisture anomaly over the course of a year, based on the Interactions between Soil, Biosphere, and Atmosphere (ISBA) land surface model of Météo-France. Its properties have been fine-tuned by comparing it to a sample of insurance data. As a continuation of this work, our aim is to analyze past and future soil moisture drought events that may cause subsidence by calculating yearly drought magnitudes for France. Prior to this, we refined the ISBA configuration by focusing solely on tree vegetation types. Historical and projected simulations were conducted with the main difference being the meteorological forcing provided to ISBA. The historical simulation covered the years 2000–2022 and used the SAFRAN atmospheric reanalysis, while the projected simulation covered the years 2006–2065 and used an ensemble of climate models under Representative Concentration Pathway (RCP) 4.5 and RCP 8.5. The historical simulation revealed significant soil moisture droughts in France in 2003, 2018, 2019, 2020, and 2022. In 2022, there were notably high index values throughout the country. The projected simulation indicated that drought conditions are expected to worsen in the future, particularly under RCP 8.5 compared to RCP 4.5. The scenarios diverged significantly after 2046, and both the north and south of the country were equally affected. Differences between historical and projected year drought magnitudes were observed: projections are more pessimistic on average and more optimistic regarding extreme events. This discrepancy can be explained either by differences in climate forcing or by differences in the vegetation response of the land surface scheme.