Identifying the simplest predictive model of annual runoff ratio for quantifying the hydrologic impact of climate change in a Great Lakes river basin

Abstract:

A standard approach to predict hydrologic fluxes in a changing climate is to downscale climate model output and feed it into a process-based hydrologic model. However, it has been demonstrated that 1) uncertainty in climate model projections often overwhelms uncertainties in the structure and parameterization of hydrologic models and 2) multiple parameter sets and structures can be used to make a hydrologic model match historical discharge (implying actual processes are not always known). Thus, it makes sense to attempt to use the simplest hydrologic model possible, both to focus on better quantifying uncertainty in climate model input and to try to ensure that the described hydrologic processes can actually be confirmed with some confidence. In addition, the use of a simple model increases the transparency of the science and can be helpful in building public consensus in making decisions regarding climate adaptation.

As a case study, the Genesee River watershed was examined. The Genesee River is a 6475 sq km watershed extending from Northern Pennsylvania to Lake Ontario, running south to north. We focused on an important but simple measure of hydrologic function: the annual runoff ratio (annual average streamflow/annual average precipitation). The annual runoff ratio varies between 0.3 and 0.8 from 1927 to 2013. The obvious explanation for higher runoff ratio years – high precipitation in seasons with little evapotranspiration – does not readily explain the variations. Three simple attempts to explain this variation were explored: a statistical regression model, the Budyko Curve, and a hydrologic “bucket” model. All models were compared based on ability to replicate annual variations in runoff ratios. Preliminary analyses suggest that although the “bucket” model most closely predicts the annual runoff ratio values, none of the simple models sufficiently explain the annual runoff ratio variability. This indicates that a more complicated hydrologic model may be necessary, with work still remaining to determine an appropriate model for common use.