Assimilation of Metop ASCAT observations into the ISBA model: evaluation of the added-value of machine learning
Jean-Christophe Calvet presents a neural network that has been trained using the modelled surface soil moisture (SSM), soil temperature, rainwater interception by leaves, and satellite-derived LAI observations from Copernicus.
In the context of climate warming, the frequency and the intensity of extreme events such as droughts is increasing, and better modelling the response of vegetation to climate is needed. Monitoring the impact of extreme events on terrestrial surfaces involves a number of variables of the soil-plant system such as surface albedo, the soil water content and the vegetation leaf area index (LAI). These variables can be monitored by either using the unprecedented amount of data from the Earth observation satellite fleet, or using land surface models. Another solution consists in combining all available sources of information by assimilating satellite observations into models. In this work, level 1 ASCAT backscatter values (sigma0) are assimilated in the ISBA land surface model of Meteo-France using the LDAS-Monde tool. First, an observation operator is built using machine learning. A neural network (NN) is trained using the modelled surface soil moisture (SSM), soil temperature, rainwater interception by leaves, and satellite-derived LAI observations from Copernicus. The NN is then used for simulating sigma0, making LDAS-Monde capable of assimilating ASCAT sigma0 observations. It is shown that the assimilation of sigma0 alone is able to markedly improve the simulated LAI and soil moisture.