The trade-off between increasing agricultural production and groundwater conservation, which is often the only irrigation source in arid regions, is a major challenge for water managers, decision-makers and the scientific community. Monitoring of the water use efficiency (WUE) provides a measure of the water lost from a canopy to the atmosphere relative to the biomass produced by the plants, and can be used to inform efforts that seek to reduce irrigation without impacting upon the yield. In this study, WUE was computed as the ratio between the gross primary productivity (GPP) and the evapotranspiration (ET). Both GPP and ET can be measured at the local scale using eddy covariance (EC) systems. However, in order to spatially evaluate WUE for larger areas or regions, local measurements are rarely representative. The spatiotemporal covariation of ET, GPP and WUE remains understudied, particularly at the scale of operational agroecosystems. In this study, Sentinel-2 optical bands were used to provide high spatiotemporal resolution estimates of ET via application of the Priestley-Taylor Jet Propulsion Laboratory (PT-JPL) model. A key soil water stress constraint in PT-JPL was linked to the Modified Soil-Adjusted Vegetation Index (MSAVI), while estimates of the GPP were delivered via the Vegetation Photosynthesis Model (VPM). In order to evaluate the water consumption and irrigation efficiency over an irrigated plantation of very high-density olive trees, ET, GPP and WUE were computed for a 1.5 year period, covering January 2019 to June 2021. Model estimates of both ET and GPP were compared to the EC measurements, with results providing a root mean square error (RMSE) of 63.3 W/m2 and a coefficient of determination (R2) of 0.74 for ET and 2.3 g C/m2/day and 0.73 for GPP, respectively. Based on these estimates, WUE was then calculated as the ratio of the GPP and ET. Relative to in situ measured data, the results were biased, with ET overestimated and GPP underestimated. An alternative measure for WUE was determined based on a linear regression model of meteorological data derived from ERA5-Land and vegetation indices derived from Sentinel-2 data. Overall, the best results were identified using an NDVI-based linear regression model, with an RMSE of 0.9 g C/Kg H2O and an R2 of 0.76. These results show the potential of high-resolution satellite data to assess WUE over agroecosystems, helping to overcome the limitations presented by in-situ measurements and supporting agricultural production and water resources management in water-limited regions.