Skip to main content
ARS Home » Plains Area » Fort Collins, Colorado » Center for Agricultural Resources Research » Water Management and Systems Research » Research » Publications at this Location » Publication #425282
Research Project: Improving Resiliency of Semi-Arid Agroecosystems and Watersheds to Change and Disturbance through Data-Driven Research, AI, and Integrated Models

Location: Water Management and Systems Research

Title: Fusion of field and satellite data for root zone moisture modeling

Author
item LAMICHHANE, MANOJ - South Dakota State University
item MEHAN, SUSHANT - South Dakota State University
item Mankin, Kyle

Submitted to: Vadose Zone Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/10/2025
Publication Date: N/A
Citation: N/A

Interpretive Summary: Precision agriculture management decisions depend upon knowing the root zone soil moisture level, especially in water-limited regions. We utilized PlanetScope data, which has a resolution of 3 meters, along with climate data, on-site soil characteristics, and a machine learning approach to estimate daily soil moisture. We found that raw data from remote sensing (like blue and near-infrared light), vegetation indices (such as NDVIgb and VARI), and climate factors (precipitation, evapotranspiration, and soil temperature) played significant roles. For deeper soil moisture layers, soil characteristics like sand, silt, clay, organic matter, and soil carbon were crucial indicators. The high-resolution, multi-layer soil moisture maps produced from this study effectively captured changes in soil moisture at the field level. This valuable tool can significantly help improve water management in agriculture and boost crop productivity in areas where water is scarce.

Technical Abstract: Precision agricultural management decisions require high-resolution root zone soil moisture (RZSM), particularly in water-limited semi-arid regions, but no study has yet quantified RZSM at the desired daily and multi-meter resolutions. This study aims to estimate multi-layer soil moisture (SM) at 30, 60, 90, 120, 150, and 180 cm depths by integrating multi-source data and applying machine learning algorithms. PlanetScope data (3-m resolution), climatic variables, and in-situ soil properties were used to estimate multi-layer daily SM at 3-m spatial resolution. Extreme Gradient Boosting (XGBoost) model was trained and tested using two different approaches: (1) using only the input features, and (2) by adding predicted SM from the adjacent upper layer. Raw remote sensing bands (blue and NIR), vegetation indices (NDVIgb and VARI) and climatic variables (precipitation, evapotranspiration, and soil temperature) were more influential in surface SM estimation, while soil properties (Sand, Silt, Clay, Organic Matter, and Soil Carbon) were key predictors for deeper SM layers. The R2, RMSE, and RRMSE values for SM predictions across six depths varied from 0.78 to 0.89, 0.021 to 0.028 cm3/cm3, and 11.7% to 14.6%, respectively. Incorporating adjacent upper layer SM information reduced RMSE by 10-27% and increased R2 by 8-24% at various depths. The generated high-resolution, multi-layer SM maps captured SM variation at the field scale, offering a valuable tool for improving agricultural water management and enhancing crop productivity in water-limited areas.