Enhancing evapotranspiration estimates in orchards with the Surface Energy Balance for Partially Vegetated surfaces (SEB-PV) model through combined use of gridded soil moisture and temporal upscaling methods.

Abstract

The Surface Energy Balance for Partially Vegetated surfaces (SEB-PV) model provides accurate evapotranspiration (ET) estimates for orchard crops. However, it faces two operational limitations: requiring specific input data unavailable from conventional agro-meteorological stations and lacking an evidence-based algorithm for upscaling instantaneous ET to daily values. This study addresses these limitations by evaluating SEB-PV performance under three conditions: (1) using measured soil moisture with in-situ meteorological equipment; (2) using gridded soil moisture products (Climate Forecast System and Soil Moisture Active Passive) with in-situ meteorological equipment; (3) using gridded soil moisture products with agro-meteorological stations' data. Seven temporal upscaling methods were compared for ET estimation in commercially-produced, micro-irrigated hazelnut (Chile) and pistachio (California) orchards. A Model Decision Making Indicator (MDMI), combining Kling-Gupta efficiency and normalized root mean square error (NRMSE), is proposed to enhance parameter optimization sensitivity. SEB-PV performance using gridded soil moisture products demonstrated comparable accuracy to configurations using measured soil moisture after parameter adjustment (MDMI values >70 for hazelnuts, NRMSE ∼ 21%; >59 for pistachios, NRMSE ∼ 29%). Transitioning from in-situ meteorological measurements to agro-meteorological stations minimally impacted hazelnut orchards but required careful consideration for pistachios. Methods that upscale instantaneous ET to daily values on the basis of net radiation performed optimally for hazelnut orchards grown in Mediterranean climatic conditions (NRMSE ∼ 15%), while meteorological inputs-based methods were preferable for semi-arid pistachio orchards (NRMSE ∼ 30%). These findings show that SEB-PV can maintain acceptable accuracy using globally available datasets, improving operational applicability through guidance for input selection and temporal upscaling tailored to orchard characteristics.