Performance of Sentinel-2 SAFER ET model for daily and seasonal estimation of grapevine water consumption

Abstract

Assessment of water consumption is a crucial task for irrigation management in grapevines, especially in areas with limited water resources, which is the case of California Central Valley. This study evaluated the utility of the Simple Algorithm for Evapotranspiration Retrievement (SAFER) model to estimate daily and seasonal actual evapotranspiration (ETa) using Sentinel 2 images at 10-m spatial resolution and 5-day revisit time in 3 vineyards located at two sites in California. A unique characteristic of this model is the estimation of “synthetic” temperature maps, which are used as part of the estimation of ET and energy balance. The SAFER energy balance results were validated with six eddy covariance (EC) flux towers as part of the Grape Remote-sensing Atmospheric Profile and Evapotranspiration eXperiment (GRAPEX). The estimated surface temperature derived from upwelling longwave radiation measurements was closely correlated with the observed sensor surface temperature with R2 higher than 0.86 for the analyzed EC towers. After performing an internal calibration, SAFER root mean square error (RMSE) values on daily ETa were between 0.64 and 0.75 mm day−1. Additionally, the seasonal ETa was estimated and compared with the EC observations showing an average R2 ranging from 0.64 to 0.52 mm/season. Spatial patterns of ETa showed variability between sites and producer management activities. The results found indicate both limitations and potential utility of SAFER for irrigation management in vineyards using daily or seasonal ETa under different irrigation treatments.