Towards medium-resolution brightness temperature retrieval from active and passive microwave

Given the importance of soil moisture for hydrological applications, including weather and flood forecasting, the Soil Moisture and Ocean Salinity (SMOS) mission was launched by the European Space Agency in 2009. This first-ever dedicated global soil moisture mapping mission has a target accuracy of 0.04 v/v. The passive microwave remote sensing approach has been adopted for this mission due to its high sensitivity to near-surface soil moisture, applicability to all weather conditions, direct correlation with the soil dielectric constant, and reduced effects by vegetation and roughness. However, passive microwave (radiometer) observations suffer from being relatively low spatial resolution, on the order of 40 km. It is proposed that this scale issue may be overcome by using active microwave (radar) observations, and this is the approach being taken by NASA’s Soil Moisture Active Passive (SMAP) mission, with a scheduled launch in late 2014. The rationale behind SMAP is that the synergy between active and passive observations can be used in a downscaling approach to overcome the individual limitations of each observation type, and ultimately provide a soil moisture data set at intermediate resolution (~10 km). The objective of this study is to test an existing downscaling approach, which has thus far received very limited testing, using airborne and satellite data, thus analyzing its viability for application. The downscaling approach tested in this paper is based on an observed near-linear relationship between active and passive observations, and theoretically proven to be an effective method. The rationale is to downscale low resolution (40 km) brightness temperature (Tb) to an intermediate resolution using high resolution (1 km) radar backscatter (σ), with soil moisture retrieval subsequently applied to the downscaled brightness temperature data. There are three components to this study: i) preliminary estimation of a slope parameter BC from a regression analysis of time-series Tb and σ data at 40 km resolution; ii) merging coarse resolution Tb and high resolution σ using a linear function with the slope BC; and iii) validate the downscaled Tb with airborne data. In this application, data from the C-band Advanced Synthetic Aperture Radar (ASAR) with approximately 1 km resolution are used to downscale L-band SMOS Tb data at the 40° incidence angle. The downscaled results are then evaluated using airborne Tb collected at 1 km resolution within the framework of the Soil Moisture Active Passive Experiments (SMAPEx) project over a 40 km × 40km area in south-eastern Australia. Results show that the Root-Mean-Square Error (RMSE) in Tb at 1 km resolution downscaled using SMOS and ASAR is 11.7 K at h-polarization and 10.3 K at v-polarization, respectively. When downscaled to 10 km resolution, the RMSE is reduced to 7.8 K and 6.9 K, respectively, showing an improvement in the RMSE of ~4 K. However, both results have errors larger than desired. This is mainly due to the ASAR_GM mode only being available in hh-polarization, while it has been shown that the best results from applying this downscaling algorithm can be expected from radar observations at vv-polarization. Some other reasons for such a result when using ASAR data may include: i) it is C-band that is more affected by vegetation, ii) the instrument is relatively noisy in Global Mode, iii) the incidence angle of the normalized backscatter is 30° rather than 40°. In addition, the SMOS data are relatively noisy (6 K-7 K) compared to that expected from SMAP (1.3K), and there is an offset in the overpass time of ASAR and SMOS of approximately 6 hrs. However, applying this downscaling algorithm similar to that being developed for SMAP shows little potential for the downscaling of SMOS with ASAR_GM data, unless the parameter BC, which was considered to be constant within the entire SMAPEx domain, can be derived at sub SMOS pixel scale.