Evaluation of MODIS NDVI and LST for indicating soil moisture of forest areas based on SWAT modeling

This study examined the capability of remotely sensed information gained using the terra moderate resolution imaging spectroradiometer (MODIS) normalized difference vegetation index (NDVI) and land surface temperature (LST) to explain forest soil moisture. The soil and water assessment tool (SWAT) was used for the analysis. Nine years (2000–2008) of monthly MODIS NDVI and LST data from a 2,694.4 km² watershed consisting of forest-dominant areas in South Korea were compared with SWAT simulated soil moisture. Before the analysis, the SWAT model was calibrated and verified using 9 years of daily streamflow at three gauging stations and 6 years (2003–2008) of daily measured soil moisture at three locations within the watershed. The average Nash–Sutcliffe model efficiency during the streamflow calibration and validation was 0.72 and 0.70, respectively. The SWAT soil moisture showed a higher correlation with MODIS LST during the forest leaf growing period (March–June) and with MODIS NDVI during the leaf falling period (September–December). Low correlation was observed in the year of frequent rains, regardless of the leaf periods.