Water resource managers aim to ensure long-term water supplies for increasing human populations. Evapotranspiration (ET) is a key component of the water balance and accurate estimates are important to quantify safe allocations to humans while supporting environmental needs. Scaling up ET measurements from small spatial scales has been problematic due to spatiotemporal variability. Remote sensing products provide spatially distributed data that account for seasonal climate and vegetation variability. We used MODIS products [i.e., Enhanced Vegetation Index (EVI) and nighttime land surface temperatures (LST_n)] to create empirical ET models calibrated using measured ET from three riparian-influenced and two upland, water-limited flux tower sites. Results showed that combining all sites introduced systematic bias, so we developed separate models to estimate riparian and upland ET. While EVI and LST_n were the main drivers for ET in riparian sites, precipitation replaced LST_n as the secondary driver of ET in upland sites. Riparian ET was successfully modeled using an inverse exponential approach (r² = 0.92) while upland ET was adequately modeled using a multiple linear regression approach (r² = 0.77). These models can be used in combination to estimate ET at basin scales provided each region is classified and precipitation data is available.