Evaluation of a Method for Estimating Irrigated Crop-Evapotranspiration Coefficients from Remotely Sensed Data in Idaho

Hydrologic modeling and agricultural water consumption analyses typically require some estimation of expected evapotranspiration (ET) from any given land cover or crop type. The main parameter used to make such estimations is the “crop coefficient” $\left(K_c\right)$ , also known as “reference ET fraction” (ETrF). An efficient remote-sensing methodology to obtain ETrF is mapping evapotranspiration at high resolution and with internalized calibration (METRIC), which requires thermal data, as obtained from the Landsat satellite. Unfortunately, the possibility of future Landsat satellites containing thermal sensors is uncertain, and other satellites’ data have access or spatial resolution concerns. Even when thermal-band data are available, energy-balance approaches can be costly. In this paper we continue ongoing efforts by the University of Idaho and others to estimate the evapotranspiration parameter $K_c$ (or $K_{cb}$ ) using the more readily available normalized difference vegetation index (NDVI) satellite-derived data product. Results from a case study in Idaho for irrigated agriculture indicate that the $\mathrm{NDVI}∕K_c$ method has significant potential to estimate $K_c$ because it is a fully objective and repeatable process, is comparably fast, easy, and less costly to apply, and does not require images from the thermal band. Preliminary work suggests that NDVI-based estimates produce results comparable to METRIC estimates over large spatial areas and full-season sample periods, even when the estimation equations were derived from different locations or crop type and management settings. While it is unclear whether empirical methods to derive the $K_c$ product are more robust than either remote-sensing estimation method, this ongoing $\mathrm{NDVI}∕K_c$ method is studied as an alternative, especially when empirical data are sparse or too costly to obtain, and when the METRIC approach cannot be taken. We find that NDVI-based estimation equations may be practically applied to areas other than the area of development.