Land Surface Data Assimilation in the Big Data Era

Soil moisture is a key variable in the global energy, water and carbon cycle. Global soil moisture variations can be simulated with Land Surface Models (LSMs) or observed with space borne remote sensing instruments. However, both of these data sources are known to have various limitations and distinct error characteristics.
Data assimilation aims at leveraging the strengths of LSMs and satellite observations by merging them in a statistically optimal way, thus creating superior land surface information. To do this successfully, one needs to couple satellite observations to the model in a consistent manner and has to have accurate knowledge about the error structures in the model and the observations. Since this is far from trivial, to date, most studies have assimilated only single satellite data products into confined modelling systems, leaving the largest fraction of currently available and rapidly growing amounts of observational data unexplored.
Aim of this project is to develop – for the first time – a multi-satellite data assimilation framework that leverages novel radiative transfer modelling and adaptive filtering techniques to enable an optimal exploitation of state-of-the-art land surface models and current and future satellite missions for an improved soil moisture estimation across scales, thereby nurturing the study of the global water, energy, and carbon cycle. The added-value of the system will be further demonstrated in a numerical weather prediction system.