GLERL-CILER Joint Seminar Series

Tuesday, May 12

10:30 am

Location:

GLERL,
Lake Superior Lecture Hall

Title: "Estimation of overlake precipitation and basin runoff uncertainty in the Great Lakes"

Speakers:
Qiang Dai
Department of Civil and Environmental Engineering, University of Michigan, Ann Arbor, Michigan
Mary E. Mello
Cooperative Institute for Limnology and Ecosystem Research, University of Michigan, Ann Arbor, Michigan
Dr. Carlo DeMarchi
Department of Geological Sciences, Case Western Reserve University, Cleveland, Ohio

Abstract:
The Net Basin Supply (NBS) for one of the Great Lakes represents the net amount of water entering the lake, not counting the supply of water from upstream lakes and can be computed as the sum of the actual inputs (over-lake precipitation, runoff from the drainage basin, and net groundwater flux) and outputs (evaporation) to/from each lake, but not the connecting-channel flows or change in storage. The reliability of NBS is a key parameter for managing Great Lakes water balance and for understanding long-term trends in the hydrological conditions of the Lakes. As part of the International Upper Great Lakes Study, we analyzed the algorithm used by NOAA (GLERL) to estimate tributary flow to the Great Lakes and evaluated its uncertainty using a Monte Carlo technique. We also computed over-lake precipitation by merging an operational multisensor product combining radar-based precipitation estimates and rain gage data, the National Centers for Environmental Prediction Multisensor Precipitation Estimates (MPE) with data from a larger network of rain gages for 2002-2007 and we analyzed the differences with the GLERL estimates for over-lake precipitation.

Results indicate that runoff uncertainty is relatively small for Lake Erie and Lake Michigan, but quite larger for lakes Superior, Huron, Saint Clair, and Georgian Bay. Further, GLERL estimates seem affected by a small bias (2 to 7%). Due to limitations of MPE, the comparison of overlake precipitation is limited to lakes Ontario, Erie, Michigan, and Huron. For these lakes, the Lake version of GLERL estimates show a better agreement with our results than the Land version of GLERL estimates. For the former product, no substantial bias is detected, while monthly uncertainty is limited for the smaller lakes, but substantial for the larger lakes.