Use of an Evapotranspiration Toolkit for Near-Real-Time Irrigation Management: Advantages and Limitations Dr. Kyle Knipper United States Department of Agriculture, Agricultural Research Service, Hydrology and Remote Sensing Lab, USA

Accurate and timely evapotranspiration (ET) information is crucial for optimizing irrigation scheduling and efficiency in vineyards, especially those located in regions prone to limited water supplies. Advances in satellite-based ET estimates provide an ability to monitor vine water use and stress across large areas of acreage not previously afforded by traditional ground-based methods and instrumentation. However, these recent advances are generally rooted in retrospective analysis, where satellite imagery is available for the entirety of the study period and can be selected based on the degree of quality (i.e. cloud cover) deemed appropriate by the investigator. This luxury is not available in an operational setting, where ET information is required in a timely and consistent manner, regardless of the availability or quality of the satellite image. In the proposed talk we will discuss the advantages and limitations of employing a satellite-based ET model in an operational (near-real-time) setting over vineyards located in California, USA. This investigation is part of the Grape Remote sensing Atmospheric Profile and Evapotranspiration eXperiment (GRAPEX) and uses the Disaggregated Atmosphere Land Exchange Inverse (ALEXI/DisALEXI) surface energy balance model, with the addition of a data fusion approach (STARFM; Spatial and Temporal Adaptive Reflectance Fusion Model), to produce high spatial (30m) and temporal (daily) ET estimates in near-real-time, defined as having a 2-3 day latency from last satellite image acquisition to final product. The operational ET product is compared to 1) eddy covariance flux tower estimates to determine the robustness and reliability of the operational product, 2) a retrospectively completed ALEXI/DisALEXI + STARFM model to better understand the degrees of model sensitivity to satellite latency and quality control, and 3) a ‘business-as-usual’ approach currently implemented operationally in the vineyards under study, allowing us to determine if a more sophisticated approach is required to better manage vineyard irrigation. Initial results indicate high model sensitivities to Thermal Infrared (TIR) satellite latency times, with errors between observed and modeled ET estimates increasing with increasing satellite acquisition time. Regardless of TIR latency, we find the ALEXI/DisALEXI + STARFM model out performs the ‘business-as-usual’ approach by utilizing TIR imagery and not relying exclusively on vegetative indices such as Normalized Difference Vegetation Index (NDVI), which is shown here to be less sensitive to vine stress. Discussion of this operational satellite-based ET model and the steps currently being taken to mitigate limitations will allow for improved strategies for integrating ET into operational irrigation management frameworks.