Mapping Walnut Water Stress with High Resolution Multispectral UAV Imagery and Machine Learning

Effective monitoring of walnut water status and stress level across the wholeorchard is an essential step towards precision irrigation management ofwalnuts, a significant crop in California. This study presents a machinelearning approach using Random Forest (RF) models to map stem water potential(SWP) by integrating high-resolution multispectral remote sensing imagery fromUnmanned Aerial Vehicle (UAV) flights with weather data. From 2017 to 2018,five flights of an UAV equipped with a seven-band multispectral camera wereconducted over a commercial walnut orchard, paired with concurrent groundmeasurements of sampled walnut plants. The RF regression model, utilizingvegetation indices derived from orthomosaiced UAV imagery and weather data,effectively estimated ground-measured SWPs, achieving an $R^2$ of 0.63 and amean absolute error (MAE) of 0.80 bars. The integration of weather data wasparticularly crucial for consolidating data across various flight dates.Significant variables for SWP estimation included wind speed and vegetationindices such as NDVI, NDRE, and PSRI.A reduced RF model excluding red-edgeindices of NDRE and PSRI, demonstrated slightly reduced accuracy ($R^2$ =0.54). Additionally, the RF classification model predicted water stress levelsin walnut trees with 85% accuracy, surpassing the 80% accuracy of the reducedclassification model. The results affirm the efficacy of UAV-basedmultispectral imaging combined with machine learning, incorporating thermaldata, NDVI, red-edge indices, and weather data, in walnut water stressestimation and assessment. This methodology offers a scalable, cost-effectivetool for data-driven precision irrigation management at an individual plantlevel in walnut orchards.