GREENCYCLESII

T6.1 Jana Kolassa

When considering the overall amount of water available on Earth and in its atmosphere, soil moisture only accounts for a small percentage of the total budget. Nonetheless, it is one of the key variables in the hydrological cycle, not only because it governs the water uptake by vegetation, but also because it drives processes such as evaporation, infiltration and runoff. These processes largely influence meteorological developments and ultimately the climate, so in order to be able to provide accurate weather predictions and climate models it is essential to determine the soil moisture content with a high degree of accuracy. While it is possible to accurately measure soil moisture directly, these in-situ measurements do not provide the global coverage required to improve meteorological and climate models. A solution to this is to use data acquired from satellites in order to obtain the required coverage. The main objective of this thesis work is thus to provide global estimates of the soil moisture derived from satellite observations in order to validate and improve meteorological and climatological models.
To achieve the research objective, the work is split into three separate stages:
1. The optimization of soil moisture estimation at global scales using satellite data. In particular, it is intended to develop a neural network and train it using data from various satellite sensors as well as in-situ measurements campaigns.
2. The development of a global soil moisture dataset ranging from the 1990's until now with the aim of quantifying soil moisture variability on different temporal scales. The generation of this dataset will be achieved using the retrieval methodology developed under 1. and applying it to different satellite data for the time period of interest.
3. The development of methodologies to assimilate the retrieved satellite data into land surface models in order to improve weather forecast and climate models. At this stage it would also be possible to first use the developed retrieval algorithm in order to validate model predictions.