ECOSTRESS Evapotranspiration Estimates Across Temporal and Spatial Scales in Arid and Semi-Arid Southern Arizona, USA

Evapotranspiration (ET) is the largest outgoing water flux in the hydrologic cycle but the most difficult component to quantify. In arid and semi-arid regions, ET can be up to 99% of precipitation and thus critical for land and water managers to quantify accurately. We evaluate ECOSTRESS ET at various temporal scales in the arid and semi-arid region of southern Arizona for 2019–2022 including all four seasons annually and using in situ observations from six eddy covariance flux towers. Our results demonstrate that ECOSTRESS ET estimates have the highest accuracy at seasonal (R² = 0.61, RMSE = 0.76, p-value < 0.05) and annual (R² = 0.77, RMSE = 0.42, p < 0.05) temporal scales, indicating stronger performance over longer integration periods. Our results also indicate that OpenET estimates more closely align with in situ data trends compared with ECOSTRESS ET. Furthermore, this study demonstrates locally-adjusted ECOSTRESS ET estimates by integrating Sentinel-2A spectral indices. The locally-adjusted ECOSTRESS ET estimates perform well for the arid and semi-arid southern Arizona (improving R² from 0.28 to 0.85) highlighting the need for site-specific and finer spatial resolution data inputs for estimating ET in these challenging environments. Enhanced accuracy of ET measurements enables land and water managers to make more informed decisions regarding limited natural resources and conservation and deepen their understanding of hydrologic dynamics in regions where ET dominates the water balance.