Seasonality and autocorrelation of satellite-derived soil moisture products

This study presents an analysis of temporal behaviour of in situ and satellite-derived soil moisture data. The main objective is to evaluate the temporal reliability of the satellite products, comparing them with in situ data, for applications that would benefit from the use of consistent time series of soil moisture, such as studies on climate and hydrological cycle. The time series, seasonalities, and anomalies of Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E) soil moisture and European Remote Sensing (ERS) satellite soil wetness index data sets were analysed over five test sites. The agreement of temporal behaviours and autocorrelation functions and the correlation with in situ data were investigated. A good agreement between the seasonalities of both satellite data sets and in situ data with high correlations (i.e. 0.9) was found over the sites with a large soil moisture variability range and short vegetation cover. Noisier seasonalities were found over sites with small soil moisture variability ranges, affected by radiofrequency interference (RFI) and characterized by croplands. In spite of ERS soil moisture being characterized by a longer time series, the seasonality is much noisier than the AMSR-E products due to the numerous gaps in the data set. The correlation among the anomalies is lower than 0.6, mainly due to the noise in the satellite products. However, the autocorrelation functions show that the anomalies are not random, although noisy. Although the stability of the anomaly correlograms is affected by the relatively short time series available for this study, the analysis shows that there are statistical similarities between the satellite soil moisture anomalies and the in situ data anomalies. The results show that AMSR-E and ERS products are consistent over long time periods and do contain useful information about soil moisture seasonality and anomaly behaviour, although they are affected by noise. © 2013 Copyright Taylor and Francis Group, LLC.