Accurate estimation of reference evapotranspiration (ET_r) is important for irrigation planning, water resource management, and preserving agricultural and forest habitats. The widely used Penman–Monteith equation (ASCE-PM) estimates ET_r across various timescales using ground weather station data. However, discrepancies persist between estimated ET_r and measured ET_r obtained from weighing lysimeters (ET_r-lys), particularly in advective environments. This study assessed different machine learning (ML) models in comparison to ASCE-PM for ET_r estimation in highly advective conditions. Various variable combinations, representing both radiation and aerodynamic components, were organized for evaluation. Eleven datasets (DT) were created for the daily timescale, while seven were established for hourly and quarter-hourly timescales. ML models were optimized by a genetic algorithm (GA) and included support vector regression (GA-SVR), random forest (GA-RF), artificial neural networks (GA-ANN), and extreme learning machines (GA-ELM). Meteorological data and direct measurements of well-watered alfalfa grown under reference ET conditions obtained from weighing lysimeters and a nearby weather station in Bushland, Texas (1996–1998), were used for training and testing. Model performance was assessed using metrics such as root mean square error (RMSE), mean absolute error (MAE), mean bias error (MBE), and coefficient of determination (R²). ASCE-PM consistently underestimated alfalfa ET across all timescales (above 7.5 mm/day, 0.6 mm/h, and 0.2 mm/h daily, hourly, and quarter-hourly, respectively). On hourly and quarter-hourly timescales, datasets predominantly composed of radiation components or a blend of radiation and aerodynamic components demonstrated superior performance. Conversely, datasets primarily composed of aerodynamic components exhibited enhanced performance on a daily timescale. Overall, GA-ELM outperformed the other models and was thus recommended for ET_r estimation at all timescales. The findings emphasize the significance of ML models in accurately estimating ET_r across varying temporal resolutions, crucial for effective water management, water resources, and agricultural planning.

Downhole soil volumetric water content (VWC) sensors are used in access tubes to assess the soil water content at multiple depths. If sensor readings are spaced closely enough vertically and are accurate enough, then accurate soil profile water content storage and change in storage can be determined over the depth range of readings, leading to accurate estimates of evapotranspiration (ET) if readings extend to well below the root zone. Even if sensing only covers the active root zone, soil water depletion may be determined well enough to inform irrigation scheduling. While sensor accuracy is dependent on many factors, including the sensor's physical principle of operation, soil‐specific calibration is typically required for good accuracy. In soils with multiple horizons (layers) of different texture, bulk density (BD), or chemical composition, horizon‐specific calibrations may be necessary. We describe methods and equipment used for downhole sensor calibration to typical accuracy of <0.01 m ³ m ⁻³ with specific reference to calibration of 10 neutron scattering meters in a soil that required three different horizon‐specific calibrations. Our results contrast with the factory calibration, which would result in a 38‐mm error in water stored in a 1.5‐m deep profile of our soil. We describe variability of measured VWC and BD with depth, distance, and water content and the errors that result from using BD to convert mass basis (g g ⁻¹ ) water content data to VWC data, which can be as much as 35 mm (7.26% underestimation) for soil water storage in a 1.5‐m deep profile of our soil.

Evapotranspiration (ET) is one of the biggest data gaps in water management due to limited ET measurements, and further, spatial variability in ET is difficult to capture. Satellite-based ET estimation has great potential for water resources planning as it allows estimation of agricultural water use at field, landscape, and watershed scales. However, uncertainties with satellite data derived ET are a major concern. This study evaluates hourly satellite-based ET from 2001–2010 for the growing season (May–October) under irrigated and dryland conditions for both tall and short crops. The evaluation was conducted using observed ET from four large weighing lysimeters at the United States Department of Agriculture Agricultural Research Service (USDA-ARS) Conservation and Production Research Laboratory in Bushland, Texas. Hourly ET from satellite data were derived using the Mapping Evapotranspiration at High Resolution with Internalized Calibration (METRIC) model. Performance statistics showed that satellite-based hourly estimates compared to lysimeter measurements provided good performance with an root-mean-square error(RMSE) of 0.14 mm, Nash–Sutcliffe efficiency (NSE) of 0.57, and R² of 0.62 for ET for dryland crops, and RMSE of 0.16, NSE of 0.63, and R² of 0.65 for irrigated crops. METRIC provided accurate hourly ET estimates that may be useful for irrigation scheduling and other water resources management purposes based on the hourly assessment.