PNNL

Abstract

Earth System Models (ESMs) serve as the primary basis for projecting future streamflow changes, but they are biased in terms of their ability to reproduce historical observations of streamflow. Rigorous downscaling and bias-correction procedures, which are time consuming and introduce uncertainties, are commonly used to address these biases in streamflow or runoff. However, the applicable limits of streamflow projections remain largely untested through direct comparisons with observations. In this work, we compare historical streamflow time series observations with a suite of Coupled Model Intercomparison Project, version 6 (CMIP6) ESMs to assess the space and time scales at which each ESM represents historic streamflow observations. We focus our analysis on streamflow across the Arctic given the higher degree of warming experienced over this region, underscoring its importance for understanding the impacts of future climate change. A series of metrics capturing streamflow variability, drought, and flooding is tested to assess ESM simulation skill for total volume, seasonality, extreme events, and overall distributions. Our results indicate that although improvements are necessary in ESM runoff and streamflow projections, the use of temporal averaging and targeting the best-fit model for comparison to observations when a suite of ESMs is considered can provide robust streamflow projections. Moreover, this work provides a basis for contextualizing the limits of ESM applicability for studies aiming to assess streamflow under a future climate.