Can we use the water budget to infer upland catchment behavior? The role of dataset error estimation and interbasin groundwater flow

Water budgets are essential for characterizing water supplies from snow-dominated upland catchments where data are sparse, groundwater systems are complex, and measurements are prone to error (epsilon). One solution is imposing water budget closure (CWB) by ignoring difficult-to-measure variables, including inter-basin groundwater fluxes (G) and epsilon. However, conventional CWB-based analyses, which derive evapotranspiration (ET) from precipitation (P) and streamflow (Q) (e.g., the Budyko hypothesis), are limited in their ability to take advantage of recent advances in ET products, physically-based frameworks for improving inferences about G, or tools to statistically characterize epsilon (Triple Collocation [TC]); all of which offer promise for improved water supply predictions via open water budgets (OWB). We clarify the value of these advances in upland settings by comparing standard land surface model, Ensemble Mean, and TC-Merged P and ET products in 114 upland catchments. When compared against a long-term OWB, we find that the CWB assumptions are unsupported in 75%-100% of our 114 catchments, depending on the product. We then show how applying these CWB assumptions in snowy, steep catchments where epsilon is large can inflate inferences about streamflow response to climate change by up 9 times more than independent (OWB) estimates of ET using TC. Finally, we demonstrate how advances in OWB analysis reveal that high, arid settings with deep permeable substrate are groundwater exporters while most other basins are groundwater importers. Our results highlight the advantages of OWB analyses that harness new products, tools, and frameworks for characterizing inter-basin groundwater fluxes in critical upland settings.