Impact of day/night time land surface temperature in soil moisture disaggregation algorithms

Since its launch in 2009, the ESA's SMOS mission is providing global soil moisture (SM) maps at similar to 40 km, using the first L-band microwave radiometer on space. Its spatial resolution meets the needs of global applications, but prevents the use of the data in regional or local applications, which require higher spatial resolutions (similar to 1-10 km). SM disaggregation algorithms based generally on the land surface temperature (LST) and vegetation indices have been developed to bridge this gap. This study analyzes the SM-LST relationship at a variety of LST acquisition times and its influence on SM disaggregation algorithms. Two years of in situ and satellite data over the central part of the river Duero basin and the Iberian Peninsula are used. In situ results show a strong anticorrelation of SM to daily maximum LST (R approximate to-0.5 to -0.8). This is confirmed with SMOS SM and MODIS LST Terra/Aqua at day time-overpasses (R approximate to-0.4 to -0.7). Better statistics are obtained when using MODIS LST day (R approximate to 0.55 to 0.85; ubRMSD approximate to 0.04 to 0.06 m(3)/m(3)) than LST night (R approximate to 0.45 to 0.80; ubRMSD approximate to 0.04 to 0.07 m(3)/m(3)) in the SM disaggregation. An averaged ensemble of day and night MODIS LST Terra/Aqua disaggregated SM estimates also leads to robust statistics (R approximate to 0.55 to 0.85; ubRMSD approximate to 0.04 to 0.07 m(3)/m(3)) with a coverage improvement of similar to 10-20%.