Spatio-Temporal Analysis and Forecasting of Soil Moisture in North Gujarat using NASA SMAP Data and Google Earth Engine: An Integrated Approach for Agricultural Water Management

Abstract— Soil moisture plays a critical role in agricultural productivity, hydrological processes, and climate interactions, particularly in semi-arid regions. This study investigates soil moisture variability in North Gujarat by integrating NASA’s Soil Moisture Active Passive (SMAP) datasets with Google Earth Engine (GEE) capabilities. A modular Python-based analytical pipeline was developed for data acquisition, preprocessing, correlation analysis, anomaly detection, trend estimation, and time-series forecasting using SARIMA and ARIMA models. The SARIMA model, incorporating precipitation as an exogenous factor, achieved an RMSE of 0.0781 m³/m³ and an MAE of 0.0615 m³/m³ for surface moisture prediction. The system also integrates an irrigation decision-support logic that determines optimal ON/OFF irrigation sequences based on real-time and forecasted moisture levels. Results reveal stable soil moisture conditions with no anomalies in the last 30 days, enabling improved irrigation scheduling for water-constrained agro-systems in North Gujarat.