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.
Keywords
Download Options
Introduction
The interaction between soil moisture and atmospheric processes significantly influences climate variability, hydrological cycles, and agricultural productivity. Semi-arid regions such as North Gujarat face persistent water stress, making soil moisture monitoring essential for sustainable farming and food security. The development of satellite-based sensing technologies and cloud-computing platforms, such as NASA’s SMAP mission and Google Earth Engine (GEE), provides unprecedented capabilities for continuous, high-resolution soil moisture monitoring.
The relationship between soil moisture and the interaction between soil and atmosphere has been extensively investigated by Seneviratne et al. (2010) providing a comprehensive picture of how soil moisture influences climate variability and extremes[1]. Their research shows that soil moisture acts as a key mediator in water and energy cycles, affecting in particular temperature and precipitation patterns by evapotranspiration. This basic understanding underpins the importance of monitoring soil moisture variations in order to predict climate change and agricultural performance in water-scarce environments. In semi-arid regions, soil moisture dynamics are characterized by high temporal and spatial variability, which significantly impacts agricultural productivity and water resource management. Porporato et al. (2004) developed a stochastic framework for understanding soil moisture dynamics in water-limited ecosystems, revealing how intermittent precipitation events interact with soil properties and vegetation to create complex moisture patterns [2]. Their mathematical modelling approach provides insight into the probabilistic nature of the availability of moisture in the soil, which is particularly important in semi-arid agricultural systems where crop performance is highly dependent on the ability of the soil to store water and the timing of rainfall events.
The use of remote sensing techniques has revolutionised the ability to monitor soil moisture, allowing for large-scale assessments of wetland conditions in a variety of landscapes. Dorigo and colleagues. (2017) presented the ESA CCI Soil moisture dataset, which provides a global long-term record of soil moisture from several satellite sensors[3]. This comprehensive dataset has proven invaluable for understanding regional soil moisture trends and their relationships with climate variability, offering crucial data for semi-arid regions where ground-based monitoring networks are often sparse or lacking.
Agro-systems in semi-arid regions face unique challenges in managing soil moisture, especially in the light of climate change and increasing weather variability. RockStrom et al. (2010) looked at water productivity in rain-fed agriculture and highlighted the crucial role of soil moisture management techniques in increasing crop yields and food security in dry-land farming systems[4]. Their analysis highlights various water harvesting and soil management practices that can improve moisture retention and utilization efficiency, providing practical solutions for farmers in regions like North Gujarat where rainfall reliability is a constant concern.
The Indian subcontinent, including regions like North Gujarat, presents specific challenges and opportunities for soil moisture research due to its monsoonal climate patterns and diverse agricultural practices. Singh et al. (2014) investigated soil moisture variability across different agro-climatic zones of India using satellite-based observations, revealing significant regional differences in moisture patterns and their correlations with monsoon intensity[5]. Their findings show the complex interaction of topography, soil characteristics and climatic factors in determining moisture availability and provide valuable insights for planning agriculture and managing water resources in semi-arid regions of India.
Recent advances in machine learning and data assimilation techniques have enhanced our ability to predict and model soil moisture dynamics in complex semi-arid environments. Feng and colleagues (2017) developed machine learning approaches to improve soil moisture forecasting using multiple satellite data sets and meteorological variables. [6]. Their research demonstrates how advanced computational methods can integrate diverse data sources to provide more accurate soil moisture estimates, which is particularly valuable for agricultural decision-making and drought monitoring in semi-arid regions where traditional monitoring approaches may be inadequate.
This study combines SMAP datasets with GEE-based spatial analysis and time-series forecasting techniques to address three major challenges: 1) Reliable monitoring of spatio-temporal changes in soil moisture. 2) Accurate short-term forecasting using rainfall data as an influencing parameter. 3)Integration of forecast results into an automated irrigation decision-support framework.
Conclusion
This study demonstrates the effectiveness of combining NASA’s SMAP data, GEE processing capabilities, and SARIMA-based forecasting for soil moisture monitoring in semi-arid regions. The integration with irrigation decision-support logic has the potential to optimize agricultural water use in North Gujarat. Future work will focus on extending the model with machine learning approaches and integrating crop growth models.
ACKNOWLEDGMENTS Thanks to NASA SMAP, UCSB CHIRPS teams, and Google Earth Engine for open data access. We express our sincere gratitude to the management of The Charutar Vidya Mandal (CVM) University and Sankalchand Patel University for their unwavering support and encouragement throughout this research. Their commitment to fostering academic excellence provided an invaluable foundation for this study. We are also deeply grateful for the access to state-of-the-art laboratory facilities, which enabled efficient data processing and model training.
CONFLICTS OF INTEREST The authors declare no financial involvement, competing financial interests, or personal relationships that could have influenced the work reported in this paper.
