Detection of Soil Moisture Variations with Fusion-Based Change Detection Algorithm for MODIS and SCATSAT-1 Datasets

Soil moisture is a vital parameter in the study of hydrology, agriculture and meteorology. The estimation of soil moisture is important for crop yield estimation, crop growth analysis and water resource management. Remote sensing is a significant way of mapping and monitoring crop fields’ soil moisture content globally, using optical and microwave satellite datasets. In previous literature, many attempts have been made to compute soil moisture using optical and microwave-based remote sensing datasets. However, the applicability of optical data is limited due to the presence of atmospheric/cloud effects, while microwave applications are restricted due to limited resolution. In this article, a fusion-based change detection approach has been proposed to detect the soil moisture variation with multispectral and microwave satellite datasets. This study has been conducted in three stages i.e., (a) image-fusion of moderate resolution imaging spectroradiometer (MODIS) and scatterometer satellite (SCATSAT-1) at HH and VV polarization using different fusion algorithms i.e., nearest neighbour-based fusion (NNF), Gram–Schmidt (GS), Brovey transformation (BT) and principal component (PC) spectral; (b) Neural Net based classification of fused datasets to deliver the thematic maps, and (c) perform the post-classification change detection (PCD) to develop the change maps. The classified and change maps have been further utilized to detect the level of soil moisture. From the experimental outputs, it has been evaluated that the NNF-based PCD performed well enough in the development of the change maps as compared to other methods i.e., GD, BT and PC spectral. The present work can aid crop yield estimation, agricultural water and precision irrigation management.