Estimation of the Spatiotemporal Variability of Surface Soil Moisture Using Machine Learning Methods - Integrating Satellite and Ground-based Soil Moisture and Environmental Data

Aug 5, 2026

Monitoring and quantifying drought extremes is critical for agriculture, water, and land management, and soil moisture acts as a key indicator. Real-time soil-moisture monitoring and forecasting remain difficult, so this study develops and tests a machine-learning framework that fuses ground-based measurements with Sentinel-1 satellite soil moisture, meteorological inputs, and environmental parameters. Four models are compared in an area exposed to hydrological extremes; extreme gradient boosting achieves the strongest performance and effective agreement across space and stations, while results indicate sensitivity to land use and soil types.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages18

Reading Time45 min

AccessFree to download

Abstract

Model setup and data integration

Evaluation of machine learning methods

Results and limitations

Estimation of the Spatiotemporal Variability of Surface soil Moisture Using Machine Learning Methods Integrating Satellite and Ground-based Soil Moisture and Environmental Data Viktória Blanka-Végi1,2 · ZalánTobak1,2 · György Sipos1,2 · Károly Barta1,2 · Brigitta Szabó2,3 · Boudewijn van Leeuwen1,2,4 Received: 16 August 2024 / Accepted: 14 December 2024 © The Author(s) 2025 Abstract Monitoring and quantifying the development of drought extremes is important to agriculture, water, and land management. For this, soil moisture (SM) is an effective indicator. However, currently, real-time monitoring and forecasting of SM is challenging. Thus, this study develops and tests a methodology based on machine learning methods that integrates ground-based data, Sentinel-1 satellite soil moisture (S1SSM) data, meteorological data, and relevant environmental parameters to improve the estimation of the spatiotemporal changes in SM. It also evaluates the relevance of the applied parameters and the applicability and limitations of S1SSM data in SM monitoring. Specifically, the performances of four machine learning methods (multiple linear regression, support vector machine regression, extreme gradient boosting, and a deep neural network) were evaluated in an area increasingly exposed to hydrological extremes. Overall, the extreme gradient boosting model provided the best result (R2 = 0.92) . In this case, the difference between the modeled and observed SM values at ground-based stations was below 3%, with only five stations reporting differences above 5%, indicating the effectiveness of this model for SM monitoring in larger areas. Additionally, the spatial pattern of the observed S1SSM values and the modeled values showed good agreement (with a difference below 10%) in the larger part (45.5%) of the area, while more than 20% difference occurred in 27.1% of the area, demonstrating the application potential of S1SSM data in areas with less heterogeneous land use. However, the results also suggest that the S1SSM data can be affected by land use and/or soil types. Keywords Soil moisture · Hydro-meteorological extremes · Machine learning · Sentinel-1 satellite data · Ground-based data Extended author information available on the last page of the article 1 3 1 Introduction The increasing frequency of drought extremes due to climate change poses a significant challenge to agriculture, water, and land management and society (Lesk et al. 2016; Blanka et al. 2017; Sharma et al. 2024). There are numerous ways to assess drought extremes, including evaluating meteorological and hydrological data, vegetation and crop-related information, and processing satellite/airborne data (Fiala et al. 2014; Leeuwen et al. 2020; Ladányi et al. 2021; Zarei et al. 2023) . Despite these approaches, real-time monitoring and forecasting of these extremes and the quantitative assessment of their effects on water resources are still a major challenge. The overall preparedness of farmers and water management can be increased if the indicators preceding the development of a drought are monitored and quantified. In this regard, one of the most important indicators is the depletion of soil moisture (SM), which is a key variable of the water cycle that links land surface and atmospheric processes (Vereecken et al. 2016). Detailed information regarding its spatiotemporal dynamics is important for numerous applications, including agriculture, water management, environmental monitoring, and forecasting of extreme climate events (Robinson et al. 2008; Seneviratne et al. 2010; Vereecken et al. 2016; Mladenova et al. 2020; Peng et al. 2021) . For example, since the amount of soil water shortage detected precedes the response of vegetation to drought, monitoring SM can forecast the need for additional water supply and reduce (or prevent) the negative ecological and economic effects of such events (Peng et al. 2021) . However, the large spatiotemporal variability of SM

FAQs

Why is soil moisture important for drought extreme monitoring?

Soil moisture is an effective indicator because it reflects depletion that precedes vegetation response, supporting forecasting and management decisions for drought impacts.

What data sources are integrated in the proposed methodology?

The study integrates ground-based soil moisture data, Sentinel-1 satellite soil moisture, meteorological data, and relevant environmental parameters to estimate spatiotemporal changes.

Which machine learning method performs best, and how is it evaluated?

Extreme gradient boosting provides the best result with R2 = 0.92, evaluated by comparing modeled versus observed soil moisture values at ground stations and assessing spatial agreement with Sentinel-1 patterns.

Aug 5, 2026

Upload by Theodore