Assessing the suitability of global evapotranspiration products over irrigated areas

Abstract. Reliable estimation of evapotranspiration (ET) over irrigated croplands is crucial for agricultural water management, hydrological modeling, and monitoring of land–atmosphere exchanges. Yet, the reliability of global ET datasets in these environments remains insufficiently assessed. Here, we evaluate six widely used global ET products (FLUXCOM RS, GLEAM4.3a, PMLv2, ERA5-Land, MOD16A2, and SSEBop v6.1), covering a wide range of modeling approaches, to assess their ability to capture irrigation-related ET signals. The assessment combines spatial and seasonal evaluations across diverse agro-climatic regions, using three complementary references: a map of area equipped for irrigation, the OpenET ensemble, and eddy covariance measurements from irrigated croplands. Results reveal marked differences in how the products represent irrigation-related ET variability. PMLv2, SSEBop v6.1, and FLUXCOM RS generally show stronger and more consistent agreement with the reference datasets in terms of spatial patterns and seasonal dynamics, although differences in ET magnitude remain between them. MOD16A2 and GLEAM4.3a capture temporal irrigation-related variability across most regions but show more variable spatial agreement with irrigation extent and generally lower ET magnitudes. ERA5-Land exhibits the weakest correspondence, reflecting limitations linked to its precipitation-driven soil-water-balance and stress formulations. Differences among products mainly reflect how vegetation water stress is represented and the extent to which irrigation-sensitive variables (e.g., land surface temperature and vegetation properties) are incorporated. This multi-scale evaluation provides guidance for selecting ET products in irrigated regions and highlights priorities for improving the representation of irrigation in global ET models.