Isotopic evaluation of the National Water Model reveals missing agricultural irrigation contributions to streamflow across the western United States

The National Water Model (NWM) provides critical analyses and projections of streamflow that support water management decisions. However, the NWM performs poorly in lower elevation rivers of the western United States (U.S.). The accuracy of the NWM depends on the fidelity of the model inputs and the representation and calibration of model processes and water sources. To evaluate the NWM, we performed a water isotope (δ18O and δ2H) mass balance using long term mean summer 5 hydrologic fluxes between 2000 and 2019, and gridded precipitation and groundwater isotope ratios.We compared the NWM-flux-estimated (‘model’) river reach isotope ratios to 4503 in-stream water isotope observations in 877 reaches across 5 basins in the western U.S. A simple regression between observed and mass balance estimated isotope ratios explained 57.9% (δ18O) and 67.1% (δ2H) of variance, though observations were 0.5‰ (δ18O) and 4.8‰ (δ2H) higher, on average, than mass balance estimates. The unexplained variance suggest that the NWM does not include all relevant water fluxes to rivers. To infer10 possible missing water fluxes, we evaluated patterns in observation-model differences using δ18Odiff (δ18Oobs −δ18Omod) and ddiff (δ2Hdiff −8 ∗ δ18Odiff ). We detected evapoconcentration of observations relative to model estimates (negative ddiff and positive δ18Odiff ) at lower elevation, higher stream order, arid sites. The catchment aridity index, the fraction of streamflow estimated to be derived from agricultural irrigation, and whether a site was reservoir-affected were all significant predictors of ddiff in a linear mixed effects model, with up to 15.1% of variance explained by fixed effects. This finding is 15 supported by patterns in groundwater levels and groundwater isotope ratios, and suggests the importance of including irrigation return flows to rivers, especially in lower elevation, higher stream order, arid rivers of the western US.

Impact/Purpose

EPA relies on models for critical analyses and projections of important processes at the national scale to support decisions.  Independent data that can be used to test models is critical for identifying potentially important processes that models may not accurately represent.  Stable isotopes indicate key ecological processes and can be important for testing models when isotopes are measured on samples collected at the national scale such as EPA’s National Aquatic Resource Surveys (NARS).  In this manuscript, we use stable isotopes of water measured on NARS samples to help identify why the National Water Model, which predicts streamflow for the USGS, preforms poorly in western low elevation rivers in the US.  Water stable isotopes indicate both the origin of water within the rivers, and the extent that water has become evaporated.  Water isotopes indicated that the model was not accounting for return flows from agricultural irrigation.  Testing models using nationally distributed stable isotope data is a powerful tool for assessing the validity of model assumptions about key ecological processes.

Citation

Putman, A., P. Longley, M. McDonnell, J. Reddy, M. Katoski, O. Miller, AND Jacqueline Brooks. Isotopic evaluation of the National Water Model reveals missing agricultural irrigation contributions to streamflow across the western United States. EGS, 28(13):2895–2918, (2024). [DOI: 10.5194/hess-28-2895-2024]

DOI: Isotopic evaluation of the National Water Model reveals missing agricultural irrigation contributions to streamflow across the western United States