Identification of climatic extremes by multi-fractal analysis of long climate data series

Shrinking and swelling of clays (SSC), occur as a result of water content fluctuations in expansive clayey soils, governed by seasonal cycles of precipitation and drought. This hazard causes ground movement, which can affect foundations and infrastructures. In France, where 54% of constructions are exposed to this hazard, SSC is the second largest category for natural disaster compensation.

With climate change, modification in the intensity and frequency of droughts, heat waves and precipitation are likely to exacerbate this phenomenon. In this context, further research is needed to anticipate the influence of climatic changes on the evolution of the SSC hazard and its impact on constructions in the next decades.

In particular, it is crucial to understand soil-atmosphere interactions on some appropriate spatial and temporal scales, but also through scales. Climate impact studies use hydrological or agricultural models, fed by global climate data adapted locally by statistical adjustments or downscaling. These methods improve local accuracy but increase bias and uncertainty, as they are often based on stationarity assumptions, which are not always valid in the context of climate change. The modeling of extreme values, essential for risk management, thus becomes more complex.

In response to the difficulties of climate models in representing extreme events at high spatio-temporal resolutions, and in understanding hydro-climatic interactions with clay soil, several geostatistical approaches are proposed.

An in-depth study of the existing literature has enabled us to compare the various downscaling methods. This state of the art is complemented by the study of data (extreme meteorological phenomena, humidity, soil displacements, etc.) acquired by various organizations concerned by the SSC problem (sources: BRGM, INRAE, SNCF, Météo-France, etc.).

This presentation will include the results of geostatistical analyses based on (multi)fractals conducted on this data (spatiotemporal variability, scale breaks, estimation of extreme values, spectral analysis, etc.). The data analyzed will cover the main parameters influencing soil moisture, i.e., precipitation and temperature.

These analyses may reveal the statistical signatures of climatic extremes. By identifying them, it will then be possible to research the different climate scenarios, and thus represent the extremes with precision. This step is essential to understanding SSC phenomena.

The final objective of this research work is to propose a soil-atmosphere interaction model, capable of generating the input data required for a numerical SSC behavior model. This model will take into account the various hydro-climatic parameters mentioned above, focusing mainly on evaporation and infiltration processes, as well as soil heterogeneity.