Disaggregation of SMOS soil moisture in southeastern Australia

DisPATCh (Disaggregation based on Physical And Theoretical scale Change) is an algorithm dedicated to the disaggregation of soil moisture observations using high-resolution soil temperature data. DisPATCh converts soil temperature fields into soil moisture fields given a semi-empirical soil evaporative efficiency model, and a first order Taylor series expansion around the field-mean soil moisture. In this study, the disaggregation approach is applied to SMOS (Soil Moisture and Ocean Salinity) data over the 500 km by 100 km AACES (Australian Airborne Calibration/validation Experiments for SMOS) area. The 40 km resolution SMOS surface soil moisture pixels are disaggregated at 1 km resolution using the soil skin temperature derived from MODIS (Moderate resolution Imaging Spectroradiometer) data, and subsequently compared with the AACES intensive ground measurements aggregated at 1 km resolution. The objective is to test DisPATCh under various surface and atmospheric conditions. It is found that the accuracy of disaggregation products varies greatly according to season: while the correlation coefficient between disaggregated and in situ soil moisture is about 0.7 during the summer AACES, it is approximately zero during the winter AACES, consistent with a weaker coupling between evaporation and surface soil moisture in temperate than in semi-arid climate. Moreover, during the summer AACES, the correlation coefficient between disaggregated and in situ soil moisture is increased from 0.70 to 0.85, by separating the 1 km pixels where MODIS temperature is mainly controlled by soil evaporation, from those where MODIS temperature is controlled by both soil evaporation and vegetation transpiration. It is also found that the 5 km resolution atmospheric correction of the official MODIS temperature data has a significant impact on DisPATCh output. An alternative atmospheric correction at 40 km resolution increases the correlation coefficient between disaggregated and in situ soil moisture from 0.72 to 0.82 during the summer AACES. Results indicate that DisPATCh has a strong potential in low-vegetated semi-arid areas where it can be used as a tool to evaluate SMOS data (by reducing the mismatch in spatial extent between SMOS observations and localized in situ measurements), and as a further step, to derive a 1 km resolution soil moisture product adapted for large scale hydrological studies.