Error characterisation of the Soil Moisture ECV product over Europe
Pratola, Chiara1; Dwyer, Edward1; Barrett, Brian2; Gruber, Alexander3
1CMRC, UCC, IRELAND; 2UCC, IRELAND; 3TUW, AUSTRIA
Within the ESA funded Soil Moisture Climate Change Initiative project, both passive and active C-band microwave sensors are exploited to produce a complete and consistent global Essential Climate Variable (ECV) soil moisture record. Due to its dependency on meteorological conditions and soil physical characteristics, soil moisture exhibits a large spatial and temporal variation between different areas, seasons and years. For a better knowledge of the phenomenon, and aiming at a more accurate characterisation of the errors inherent in the coarser spatial resolution ECV product ENVISAT ASAR Wide Swath (WS) data have been used in this study due to their capability of providing information on short-term, seasonal and long-term variations in surface soil moisture at higher spatial resolution (150 m -aggregated to 1 km-). Such properties allow the monitoring of soil moisture over different land covers so that the ASAR soil moisture retrieved outputs may contribute to better understanding of the ECV products, through a comparison between ASAR derived soil moisture products and ECV data.
This paper focuses on three primary regions which are different in terms of climate, topography and land uses, namely Southern Ireland, Duero basin in Spain and Sodankylä in Finland. Soil moisture values have been retrieved from the ASAR WS data acquired during the period from 2002 and 2012 based on a simplified version of the TU-Wien change detection method. Preliminary results exhibit promising correlations between the two different soil moisture datasets for both the Irish and Spanish test sites; a very similar outcome has been obtained among the backscattering time series and the ECV product. The root mean squared deviation (RMSD) is slightly higher for the Spanish sites compared to the Irish ones. Moreover a similar seasonal soil moisture trend has been identified between the ECV and the ASAR derived soil moisture products. In general, for both the Irish and Spanish sites, it has been observed that the highest correlation occurs in autumn which corresponds to the lowest seasonal RMSD. In order to improve the understanding of each land cover contribution to the ECV soil moisture value, ASAR WS pixels belonging to each land cover type within an ECV pixel size are collected and comparisons are made between the temporal behaviour of the ASAR retrieved soil moisture and ECV soil moisture in the same area.