A national-scale tool integrating machine learning for rootzone soil moisture estimation, enhancing irrigation scheduling across diverse fields. It utilizes models like RF, LightGBM, and more to provide precise, scalable solutions for corn and other crops.
The AI-Infused Soil Moisture Integration into OASIS Tool represents a cutting-edge solution for irrigation scheduling. This innovative tool leverages advanced machine learning and deep learning models to estimate rootzone soil moisture across various depths and field conditions. By overcoming traditional barriers of cost, labor, and data intensity, this tool provides a scalable and accurate method to optimize water usage in agriculture, particularly for corn crops.
The tool integrates state-of-the-art machine learning models including Random Forest (RF), LightGBM, XGBoost, Long Short-Term Memory (LSTM), and Transformers (TF). These models are tested across 45 irrigated fields over five years and 120 non-irrigated fields spanning 30 years, covering 600 soil types. The tool provides soil moisture estimates with accuracy measured in root mean square error (RMSE) and mean bias error (MBE). This allows for precise field-level hourly soil moisture predictions, which are then aggregated to daily rootzone moisture values.
This solution is at Technology Readiness Level 9, indicating it has been successfully tested and is ready for full deployment across Kansas, Oklahoma, and Texas. The integration within the OASIS tool exemplifies a validated, market-ready technology that can significantly enhance irrigation strategies.
Oklahoma State University is a comprehensive public land‑grant research university with a statewide footprint anchored in Stillwater. A research and technology park connects faculty expertise with corporate partners through co‑located labs, flexible leases, and shared prototyping and pilot‑scale facilities. Integration with an academic health system in Tulsa and a veterinary teaching hospital supports clinical translation, field trials, and product evaluation. A statewide extension network and experiment stations provide industry with rapid access to real‑world testing sites and workforce pipelines. Research is supported by competitive federal funding from NSF, USDA, NIH, and DOE, alongside industry‑sponsored agreements. A dedicated technology transfer office and an affiliated incubator accelerate IP protection, licensing, and startup formation.