Adjustment of Radar-Based Precipitation Estimates for Great Lakes Hydrologic Modeling

Abstract

For the Great Lakes region, radar-based precipitation estimates provide valuable data for over-lake areas and may help to improve mean areal precipitation estimates over land areas, particularly where precipitation gauges are sparse. To this end, potential biases in daily radar-based precipitation data are evaluated for several Great Lakes watersheds. Statistical analysis of gauge data (MAP) and radar-based data (MAPX) for the period 1996–2004 shows that, except for some summer months, MAPX values were consistently lower than MAP values. Possible reasons for this underestimation include truncation error, beam blockage and overshoot, and variability in radar reflectivity-surface precipitation relationships. Application of the precipitation data to continuous simulation and forecasting models of the Great Lakes shows that the observed biases are generally magnified by the hydrology models, i.e., even larger discrepancies are observed between simulated runoff values generated with MAP and MAPX inputs. Discrepancies in observed runoff and simulated runoff with MAPX inputs are deemed unacceptable for hydrologic prediction, and procedures for adjusting the MAPX data are developed and verified.