Satellite‐based near‐real‐time estimation of irrigated crop water consumption

Abstract

Estimation of evapotranspiration using satellite sensors offers the potential for improved water management in irrigated areas. However, previous applications of remote sensing to estimate crop water use have been retrospective in nature, whereas the spatial resolution of sensors that provide information in real time has been thought to be too coarse for such purposes. We describe application of a variation of the Moderate Resolution Imaging Spectroradiometer (MODIS) standard evapotranspiration algorithm for near‐real‐time method for estimation of actual evapotranspiration based entirely on satellite data. The latency of the approach is typically 3 days to 1 week, although the lag could be reduced. The method works best over areas where there is substantial diversity in vegetation types within the remote sensing window; the contrast in vegetation between irrigated and adjacent unirrigated areas meets this requirement well. The satellite sensors/products are MODIS land cover (MOD12Q1), surface reflectance (MOD09GQ), vegetation indices (MOD13Q1), land surface temperature/emissivity (MOD11A1), and albedo (MCD43A3) and NOAA/NESDIS surface radiation budget (SRB) products derived from the Geostationary Operational Environmental Satellites. The MODIS/SRB evapotranspiration estimates agree favorably with ground flux tower observations and evapotranspiration estimates from a much higher resolution Landsat‐based (METRIC) method over irrigated areas of the Klamath River Basin, with instantaneous evapotranspiration biases less than 10% and daily evapotranspiration biases less than 15%. There is a tendency for the MODIS/SRB approach to underestimate, and for the METRIC‐based algorithm to overestimate, seasonal evapotranspiration relative to the tower flux observations.