How the response of high river flows to warming across the US is only loosely tied to precipitation

Efforts in enhancing resolutions of climate models are largely motivated by improving the representation of precipitation. Accurately capturing precipitation is key for understanding how a variety of natural hazards will change in the future, such as floods and droughts. Recent advances of climate models within HIGHRES-MIP to 50 km resolution and higher showed significant improvements in representing precipitation compared to CMIP6, particularly that associated with frontal systems of the mid-latitudes often manifesting as Atmospheric Rivers (ARs). Here, we leverage these new capabilities to study the sensitivity of high river streamflows on land to warming. We do so by utilizing the coupled atmosphere and land surface model AM4/LM4.0 developed at the Geophysical Fluid Dynamics Laboratory (GFDL). By applying a lagged correlation analysis between streamflows of the coupled river network and its upstream drivers, we identify major changes in drivers of high flows in response to a simple pseudo global warming experiment that yields a spatially homogeneous precipitation increase across most of the US.

We find that changes in high river flows show a strong dipole pattern across the US with increases in the East and decreases in most of the West. The increase in high-flows over the Eastern US is driven by an increase in precipitation-driven high-flows that exceeds the reduction in melt-driven high-flows. Among precipitation-driven high-flows, ARs contribute most to the increase. The reduction of high flows across the central and Western US is explained by significantly weaker snowmelt in spring. Here, increases in precipitation with warming, particularly from ARs, are counteracted by increased evaporation causing streamflows to dwindle. A Budyko-Analysis reveals that the disconnect of changes in precipitation and high flows can be explained by the energetic potential of the land-surface to evaporate the additional precipitation. While this potential is high over the central and Western US, it is low over the Eastern US. Finally, the overall reduction of snowmelt is found to alter the seasonality of high-flows with warming, for example in different sub-basins of the Mississippi where the month of peak high-flows shifts from March to May.