Recalibration of Hydrologic Models for Use with Wsr-88d Precipitation Estimates

With the deployment of the WSR-88D weather radar, the National Weather Service will provide improved estimates of hourly precipitation accumulations for most of the United States on roughly a 4 by 4 km grid. This information will be used as input to operational hydrologic models to improve streamflow prediction for flood forecasting and water resources management. A critical issue that hydrologists face is the calibration of models for use with WSR-88D precipitation estimates. The traditional procedure for calibration is to: (1) obtain long records of precipitation and streamflow, (2) use a portion of the record to adjust the model parameters to minimize the prediction errors in streamflow for the calibration period, and (3) use the remaining portion of the record to verify the predictive ability of the model with the estimated model parameters. Clearly, the shortness of existing radar precipitation records prohibits the use of traditional approaches for calibration. For operational hydrologic models in use today, calibration has already been carried out using precipitation data from gage networks. Therefore, the same estimated model parameters could be used in model simulation with the improved precipitation estimates from the WSR-88D. However, Troutmann (1982) has demonstrated the importance of using consistent precipitation information for both calibration and prediction. During calibration, model parameters adjust to compensate for the biases and sampling characteristics of areal precipitation estimates (Troutmann, 1982; Bradley, 1997). Therefore, model parameters estimated using gage information may no longer be appropriate for use with radar precipitation estimates — even though the radar estimates may be better (e.g., less bias and lower sampling errors). In this paper, we present an experimental framework for examining the impact of a change from Corresponding author address: A. Allen Bradley, The University of Iowa, Iowa Institute of Hydraulic Research and Department of Civil and Environmental Engineering, Iowa City, IA 52242-1585; e-mail gage precipitation to WSR-88D precipitation estimates in model calibration and runoff simulation. The framework will be used to examine runoff simulation errors using WSR-88D precipitation estimates with (1) model parameters obtained from a calibration with gage precipitation, and (2) model parameters obtained from a recalibration technique. Preliminary results are presented for a hypothetical model subcatchment.