Development of a Consistent GIS Based Method for Estimating the Groundwater Runoff Parameter for Regional Scale Precipitation-Runoff Models

Thursday, 18 December 2014
David M Bjerklie, USGS Connecticut Water Science Center, East Hartford, CT, United States
As part of a U. S. Geological Survey effort to (1) estimate river discharge in ungaged basins, (2) understand runoff quantity and timing for watersheds between gaging stations, and (3) estimate potential future streamflow, a national scale precipitation runoff model is in development. The effort uses the USGS Precipitation Runoff Modeling System (PRMS) model. The model development strategy includes methods to assign hydrologic routing coefficients a priori from national scale GIS data bases. Once developed, the model can serve as an initial baseline for more detailed and locally/regionally calibrated models designed for specific projects and purposes.

One of the key hydrologic routing coefficients is the groundwater coefficient (gw_coef). This study estimates the gw_coef from continental US GIS data, including geology, drainage density, aquifer type, vegetation type, and baseflow index information. The gw_coef is applied in regional PRMS models and is estimated using two methods. The first method uses a statistical model to predict the gw_coef from weighted average values of surficial geologic materials, dominant aquifer type, baseflow index, vegetation type, and the drainage density. The second method computes the gw_coef directly from the physical conditions in the watershed including the percentage geologic material and the drainage density. The two methods are compared against the gw_coef derived from streamflow records, and tested for selected rivers in different regions of the country. To address the often weak correlation between geology and baseflow, the existence of groundwater sinks, and complexities of groundwater flow paths, the spatial characteristics of the gw_coef prediction error were evaluated, and a correction factor developed from the spatial error distribution. This provides a consistent and improved method to estimate the gw_coef for regional PRMS models that is derived from available GIS data and physical information for watersheds.