Self-calibrated evaporation-based disaggregation of SMOS soil moisture: An evaluation study at 3km and 100m resolution in Catalunya, Spain

摘要

A disaggregation algorithm is applied to 40. km resolution SMOS (Soil Moisture and Ocean Salinity) surface soil moisture using 1. km resolution MODIS (MODerature resolution Imaging Spectroradiometer), 90. m resolution ASTER (Advanced Spaceborne Thermal Emission and Reflection radiometer), and 60. m resolution Landsat-7 data. DISPATCH (DISaggregation based on Physical And Theoretical scale CHange) distributes high-resolution soil moisture around the low-resolution observed mean value using the instantaneous spatial link between optical-derived soil evaporative efficiency (ratio of actual to potential evaporation) and near-surface soil moisture. The objective is three-fold: (i) evaluating DISPATCH at a range of spatial resolutions using readily available multi-sensor thermal data, (ii) deriving a robust calibration procedure solely based on remotely sensed data, and (iii) testing the linear or nonlinear behavior of soil evaporative efficiency. Disaggregated soil moisture is compared with the 0-5. cm in situ measurements collected each month from April to October 2011 in a 20. km square spanning an irrigated and dry land area in Catalunya, Spain. The target downscaling resolution is set to 3. km using MODIS data and to 100. m using ASTER and Landsat data. When comparing 40. km SMOS, 3. km disaggregated and 100. m disaggregated data with the in situ measurements aggregated at corresponding resolution, results indicate that DISPATCH improves the spatio-temporal correlation with in situ measurements at both 3. km and 100. m resolutions. A yearly calibration of DISPATCH is more efficient than a daily calibration. Assuming a linear soil evaporative efficiency model is adequate at kilometric resolution. At 100. m resolution, the very high spatial variability in the irrigated area makes the linear approximation poorer. By accounting for non-linearity effects, the slope of the linear regression between disaggregated and in situ measurements is increased from 0.2 to 0.5. Such a multi-sensor remote sensing approach has potential for operational multi-resolution monitoring of surface soil moisture and is likely to help parameterize soil evaporation at integrated spatial scales.