Soil moisture monitoring at kilometre scale: assimilation of Sentinel-1 products in ISBA

In a warming climate where the frequency and intensity of extreme events (such as droughts and floods) are increasing, a better representation and estimation of land surface variables remains a crucial step to study their response to climate change. Soil moisture is a key variable of the water cycle. Monitoring soil moisture, either by in situ measurements or by satellite observations allows better prediction and anticipation of droughts and floods, especially in agricultural regions. In order to fully exploit the growing number of satellite observations data, assimilation techniques can be used to integrate these data into land surface models.

In this work, surface soil moisture (SSM) observations from Sentinel-1 (S1) satellite are assimilated into the ISBA model at the kilometer scale. The main objective is to evaluate the added value of the SSM assimilation and its impact on the ISBA model simulations, driven by atmospheric variables from the AROME weather forecast model. The Land Data Assimilation System tool (LDAS-Monde) of Météo-France is used. The SSM S1 product covers the period 2017-2019, over two regions in south of France and one in Spain. The native resolution of the S1 product is 10 m, and the aggregated 1 km product only covers areas where radar signal interpretation is possible. The two areas of interest in France are the Toulouse and the Montpellier regions. In these two areas, in situ soil moisture measurements are available (SMOSMANIA network and Meteopole-Flux stations of Meteo-France). The area of interest in Spain is located between Salamanca and Valladolid, where the REMEDHUS network of in-situ soil moisture measurements is located. In situ SSM observations at a depth of 5 cm were gathered from all stations at an hourly temporal resolution. The S1 SSM shows a good agreement with the in situ observations, including over the Météopole-Flux site which is located in a semi-urban area.

The impact of assimilating SSM products is evaluated over three surface variables: SSM at the 1 – 4 cm soil deph layer (WG2), at the root zone at 30 cm soil depth (WG5) and on the Leaf Area Index (LAI). Three experiments are then carried out over the three regions: assimilation of the S1 SSM product alone, assimilation of the LAI retrieved from the Copernicus Global land Service (CGLS), and one last experience where S1 SSM is jointly assimilated with LAI.

The results of these experiments on one hand show that when SSM alone is assimilated, almost no improvement is observed on WG2 between the ISBA model outputs and the assimilation outputs when compared to in situ measurements. On the other hand, when SSM is jointly assimilated with LAI, there is a stronger impact on WG2 and thus the outputs are closer to the in situ observations. Concerning WG5, the impact of assimilating SSM and LAI is found to be even stronger.