Remote Sensing and Deep Learning Based Soil Moisture Monitoring System Using HCSWO Optimization Technique and AGRL-RBFN

Soil moisture serves as a crucial factor in the hydrological cycle, supporting plant development, ecosystems heath and contributing to groundwater reserves. Consequently, it plays a significant role in the global climate system. Existing research has not sufficiently explored the impact of climatic changes on soil moisture patterns during monitoring, which has complicated prediction and management efforts. To tackle this issue, the proposed study employs seasonal mapping and grouping techniques to observe climatic variations and predict soil moisture utilizing the AGRL-RBFN method with IL. Initially, historical remote sensing data on soil moisture is gathered and subjected to a three-step preprocessing procedure: gaps are filled using the AdaK-MCC method, noise is minimized through the Savitzky-Golay Filter (SGF), and atmospheric interferences are corrected. Following this preprocessing phase, seasons are mapped, and the AdaK-MCC method is utilized for data grouping. A multivariate correlation analysis is subsequently conducted on the grouped data through Principal Component Analysis (PCA). The diverse patterns within the grouped data are further examined using the FWFCSD method. Features are then extracted from these patterns and correlation analyses, after which optimal features are selected via the Hierarchical Correlated Spider Wasp Optimizer (HCSWO). Ultimately, the AGRL-RBFN with IL is employed to predict soil moisture, resulting in a highly accurate prediction with an accuracy rate of 98.09%.