The Great Lakes Runoff Intercomparison Project (GRIP)

As a continuation of investments in the development of alternative methods for estimating major components of the Great Lakes water budget through the recently-completed International Joint Commission (IJC) International Upper Great Lakes Study (IUGLS), representatives from a variety of United States and Canadian agencies have formed a bi-national collaboration to assess alternative methods for modeling runoff within the Great Lakes basin. The project is based on assessing and comparing simulated runoff across the watersheds of both Lake Michigan and Lake Ontario, with an emphasis on understanding the different sources of data needed to support these models, and a comparison between both total runoff and estimated runoff at individual gauging stations. Models, or modeling frameworks (and contributing agencies) participating in the project include (but are not limited to) Analysis of Flows in Networks of Channels (or AFINCH, from USGS), the Community Hydrologic Prediction System (or CHPS, from NOAA's National Weather Service), the MESH system (from Environment Canada), the Large Basin Runoff Model (or LBRM, from NOAA's Great Lakes Environmental Research Laboratory) as well as a series of empirical methods for extrapolating historical gauge measurements to ungauged portions of each Lake basin. This presentation will also explore alternative methods for comparing runoff estimates over broad spatial scales, and for understanding potential sources of bias and uncertainty within and between these estimates. For models generating probabilistic estimates (i.e. with an explicit expression of uncertainty) we provide a comparison based on posterior predictive p-values (similar to rank histograms), an approach which, unlike conventional deterministic metrics, provides an indication of the relative importance of uncertainty in large-scale hydrological model assessment and how expressions of that uncertainty propagate into model-based water resources management planning decisions.