Investigating the Efficacy of the SMAP Downscaled Soil Moisture Product for Drought Monitoring Based on Information Theory

Soil moisture (SM) information can be routinely obtained from high-quality microwave retrievals at a global scale—such as datasets generated by the Soil Moisture Active Passive (SMAP) mission. In this article, using mutual information (MI) theory, we investigate the efficacy of the downscaled SMAP/Sentinel-1 L2 3-km EASE-Grid SM product (SPL2) for the detection of agricultural drought over northwestern China. The SPL2 is generated by merging SMAP enhanced radiometer data with Sentinel-1 radar observations. To evaluate the efficiency of the SPL2 downscaled algorithm, the SMAP Enhanced L3 Radiometer 9-km EASE-Grid SM product (SPL3) is also utilized as a non-downscaled baseline. Over croplands, comparing normalized MI (NMI) values sampled between the NDVI time series and 3-km Sentinel-1
C
-band backscatter coefficient (
σ
) from SPL2 with NMI values between NDVI and SPL3 radiometer brightness temperature (Tb; resampled to 3-km resolution), we find that the Sentinel-1
σ
explains more (3-km) NDVI information than the SPL3 Tb, as the NMI between
σ
_vh
(
σ
_vv
) and NDVI is 15% (8%), larger than that between SPL3 Tb and NDVI (5%). However, compared to the SPL3 Tb baseline, the information from downscaled SPL2 Tb on NDVI is reduced by approximately 3%, and the SPL2 algorithm extracts only 7% (10%) of the total information available from both enhanced SPL3 Tb and Sentinel-1
σ
_vh
(
σ
_vv
). Overall, the
C
-band
σ
signal provides valuable information for vegetation monitoring due to its frequency advantage. However, additional efforts should be focused on SPL2 merging algorithms to improve the value of the downscaled SPL2 product for agricultural applications.