Preliminary results from a multi-decadal analysis of root-zone soil moisture applying the exponential filter with AMSR-E and Soil Moisture CCI data

This study applied the exponential filter to produce an estimate of root-zone soil moisture (RZSM). Four types of microwave-based, surface satellite soil moisture products were used. The core remotely sensed data for this study was derived from NASA's long lasting AMSR-E mission. Additionally, three other products were obtained from the European Space Agency Climate Change Initiative (CCI). These datasets were blended based on all available satellite observations (CCI-Active; CCI-Passive; CCI-Combined). These daily products have a quarter degree resolution. We applied the filter to produce a soil moisture index (SWI). Others successfully have used this method to estimate RZSM. The only unknown variable in this approach was the characteristic time of soil moisture variation (T). We examined five different eras (1997-2002; 2002-2005; 2005-2008; 2008-2011; 2011-2014) that represented periods with different satellite data sensors. SWI values were compared with in situ soil moisture data from the International Soil Moisture Network at three depths (20-30 cm; 50-60 cm; 80-100 cm). Selected networks included ARM, SCAN, USCRN, OZNET, and others. We chose in situ stations that had reasonable completeness including ancillary temperature and precipitation data. These datasets were used to filter out periods with freezing temperatures and rainfall. Additionally, we only examined sites where surface and RZSM had a reasonably high correlation (r>0.5). Two separate optimizations of the T value based on Nash-Suffcliffe score (NS) and root mean square error (RSME) were explored. Both approaches yielded comparable results. Best results were noted at stations that had a bias within 10% and an r-value between SWI and in situ data of greater than 0.5. At depths >30 cm the SWI values generated by CCI-Passive data exhibited a decrease in performance (NS became highly negative) compared with other satellite products. Warm season (May-September) NS and RSME scores consistently outperformed entire era results from CONUS stations. In situ network average NS values exceeded 0.4 in some instances indicating that this approach has some skill in providing an estimate of RZSM. In the future, we plan to explore other approaches where results from the exponential filter could be leveraged.