European Journal of Sustainable Agroecosystems

Background: Soil moisture (SM) is a fundamental state variable of the terrestrial water, energy, and carbon cycles, governing infiltration, evapotranspiration, crop water availability, and land-atmosphere feedbacks. Its high spatial and temporal heterogeneity, driven by topography, soil texture, land cover, and precipitation variability, makes conventional point-based monitoring inadequate for basin- to regional-scale water resource planning and precision agriculture. Advances in geospatial techniques, including satellite microwave remote sensing, Geographic Information Systems (GIS), geostatistics, and machine learning, have substantially enhanced the ability to characterise soil moisture across spatial and temporal scales.

Objective: This review synthesises peer-reviewed literature published between 2015 and 2025 on spatiotemporal soil moisture analysis using geospatial techniques. It aims to critically evaluate methodological approaches, identify convergences and contradictions across studies, and highlight key research gaps for future investigation.

Method: A structured literature search was conducted across Scopus, Web of Science, ScienceDirect, SpringerLink, IEEE Xplore, PubMed, and Google Scholar. The review was supplemented with reports from the Food and Agriculture Organization (FAO) and other international organisations. Study identification, screening, and selection followed PRISMA-guided procedures to ensure a systematic and transparent review process.

Results: The reviewed literature consistently demonstrates the superiority of active–passive microwave sensors, particularly SMAP, Sentinel-1, and SMOS, for large-scale soil moisture retrieval. Machine learning and deep learning models, including Random Forest, XGBoost, Long Short-Term Memory (LSTM), and Convolutional Neural Networks (CNNs), increasingly outperform traditional physical retrieval algorithms in predictive accuracy. Geostatistical interpolation methods, especially kriging variants, together with multi-source data fusion, are widely recognised as essential for converting coarse-resolution satellite products into field-scale information suitable for precision agriculture. However, significant disagreements remain regarding the optimal downscaling techniques, the transferability of trained models across different climatic regions and land-cover types, and the reliability of soil moisture retrieval under dense vegetation. Persistent research gaps include limited validation networks in the Global South, insufficient model explainability, inconsistent integration of geospatial soil moisture products into operational irrigation and fertiliser decision-support systems, and the lack of standardised accuracy-reporting metrics.

Conclusion: Geospatial soil moisture analysis has achieved substantial methodological advances but remains operationally fragmented. Future research should prioritise explainable artificial intelligence, cross-scale data fusion, affordable and extensive validation networks, and the integration of soil moisture products into precision agriculture and fertiliser recommendation systems to facilitate effective field-level decision support.