Global variable resolution simulations offer an opportunity for affordable high-resolution modeling without the boundary condition challenges typically associated with regional model downscaling. Hundreds of years have been simulated using version 2 of the Energy Exascale Earth System Model (E3SMv2) with regional refinement over North America. These experiments include both forced (“historical”) and unforced (“piControl”) scenarios, building upon earlier work with E3SMv1, which compared globally uniform low- and high-resolution simulations. E3SMv1 indicated a significant slowdown in the Conterminous United States (CONUS) water cycle with increasing resolution, characterized by reductions in precipitation, evapotranspiration (ET), atmospheric moisture convergence, runoff, and terrestrial water storage (TWS).

In contrast, initial findings with E3SMv2 suggest a less pronounced and less uniform sensitivity in response to increased horizontal resolution, despite the resolution differences over CONUS being the same in both versions. These discrepancies prompted a deeper exploration of the uncertainty in resolution sensitivity. This study aims to address questions about the CONUS water cycle’s response to increasing resolution in E3SMv2: Is there a robust response? If so, how many years of data are necessary to capture it confidently?

Leveraging the unforced piControl simulations at low resolution (LR) and the North American regionally refined model (NARRM), both of which include 500 years of coupled simulations, we tested the uncertainties in the water cycle response to increasing resolution. Results show that 30-year periods are often insufficient to capture even the sign change of the true 500-year mean difference. By utilizing bootstrap resampling techniques, we determined the sample sizes needed to achieve 95% confidence in the sign of the mean differences, with ET requiring the fewest years (32 on average). Our analysis also considered differences between coupled and prescribed sea surface temperature (SST) simulations and their sensitivity to global temperature changes. In this presentation, we will provide an in-depth look at the resolution sensitivity of the water cycle at the watershed level, comparing the impacts of different modeling approaches.