Large variability in satellite-based estimates of irrigation water use

As competition for water resources intensifies, especially in water-scarce regions, there is a growing need to manage water usage effectively, particularly in irrigated agriculture. However, data on agricultural water use and abstractions are often unavailable or only at coarse spatial resolutions. Water use by crops can be physically characterised by the landscape’s rate of evapotranspiration (ET). Satellite-based monitoring of actual ET provides one potential solution to this data gap, but significant knowledge gaps remain about the uncertainty in satellite-based estimates of irrigation water use (IWU) and their associated implications for policy and management. In this model-intercomparison study, we attempt to address the relevance of model choice on satellite-based estimates of IWU by assessing the variability resulting from different model estimates of satellite-based IWU. We utilised six satellite-based ET datasets from OpenET and five precipitation datasets to estimate field-level IWU over 6 years in the high plains aquifer, United States. Results reveal substantial variability in IWU estimates, particularly at field and seasonal scales, which reduces when aggregated spatially or temporally. ET, rather than precipitation, was the primary driver of variability in IWU estimates. These findings highlight the challenges of using satellite data to estimate IWU at fine spatial and temporal scales or in areas where irrigation supplements rainfall. Aggregating IWU estimates reduces variability but emphasises the importance of model choices when monitoring irrigation water usage both at the farm and at regional levels.