Soil moisture retrieval based on C-band Imagery of SAR and rainfall detection over soccer fields: The case of Ghana

By Dimos T. Touloumidis, supervised by Prof. dr. ir. Nick van de Giesen

Precipitation over Africa can be estimated in three different ways, with ground-based rainfall products, with weather radars or with satellite rainfall products. Rain gauge is the most accurate way of measuring rainfall but there is an important limitation about the stations’ density. The low coverage of the in situ stations and the limitations of the weather radars led to solution of satellite products; the satellite rainfall and soil moisture products are available globally with lower accuracy. The recent satellite products GSMaP and IMERG (JAXA/Japan, NASA/USA, etc.) are providing a spatial resolution of 0.1o and temporal resolution of 30 minutes. Apart from precipitation, soil moisture data can be derived using the backscatter radiation (ASCAT, AMSR-E, SMOS, SMAP). Recent study, proposed a “bottom-up” approach for the estimation of precipitation, developing an algorithm to derive precipitation using soil moisture products and the results showed a significant Pearson correlation (R) between soil moisture and precipitation reaching a global median of 0.542 (using ASCAT-derived products). The three major limitations of the aforementioned algorithm are the fact that the backscatter signal corresponds to the upper surface of the ground (< 7 cm.), the resolution the pixels which are about 20 kilometres and the heterogeneity of the soil.

The imaging radar mission Sentinel-1, which was launched by 2016, is equipped with the Synthetic Aperture Radar (SAR) which provides C-band imagery with a typical resolution of 10 meters (for level 1 products) and temporal resolution of six days. The high resolution of SAR imageries can offer the retrieval of soil moisture inhomogeneous patches, overcoming the limitation of the heterogeneity. At the same time, a careful selection of specific patches, with known properties, shape and size can play an important role to retrieve reliable values of soil moisture.

The objective of the study is to evaluate whether the homogeneous patches in the landscape of Ghana (e.g., soccer fields) can be used as rainfall detectors using a “bottom-up” algorithm based on soil moisture (backscatter radiation imageries).