Spectral signatures in satellite soil moisture reveal irrigation patterns across the contiguous United States

Christian Massari , Sara Modanesi, Zdenko Heyvaert, Louise Busschaert, Pierre Laluet, Martina Natali, Jacopo Dari, Luca Brocca, Wouter Dorigo, Clement Albergel, Gabrielle De Lannoy

Irrigation profoundly alters the terrestrial water cycle, yet its spatial distribution and temporal variability remain poorly constrained. Here, we introduce a new approach to detect irrigation in space based on spectral differences between modelled and satellite-observed soil moisture time series. Using wavelet decomposition, we isolate irrigation-induced variability at sub-annual scales by comparing Noah–MP simulations with and without irrigation. We then analyse differences between Soil Moisture and Ocean Salinity (SMOS) soil moisture retrievals and model simulations without irrigation to identify irrigated areas.
We apply this method over the Contiguous United States (CONUS) for the period 2010–2023. Results show that irrigation enhances the power of seasonal and sub-seasonal spectral components of soil moisture, enabling identification of irrigated and non-irrigated areas when comparing observed and modelled soil moisture. Comparison with LGRIP30, an irrigation dataset for the United States, shows moderate-to-good agreement in the spatial patterns of irrigation. Major irrigated regions across CONUS largely coincide with those identified in LGRIP30, while discrepancies increase in areas with weaker irrigation signals or dense vegetation.
The resulting SMOS-derived irrigation map suggests spatial patterns of irrigated areas that differ from those represented in static reference maps such as LGRIP30. When used to constrain Noah–MP simulations, SMOS-derived irrigation map highlights alternative irrigation distributions that better reflect actual irrigation spatial patterns and water use.
Our method provides a scalable, all-weather pathway to detect where irrigation occurs, offering new observational constraints for representing irrigation in land and Earth-system models.