Evaluating incremental improvements in precipitation estimation using an ensemble of hydrologic simulations

Improvements in quantitative precipitation estimation (QPE) are often cast in terms of their impact on hydrologic simulations. This study addresses the following questions through the use of a physics-based, distributed parameter hydrologic model: 1) Is there a benefit to using radar data over gauge data as an input to the model? 2) Does the deployment of a gap-filling radar in complex terrain yield more accurate streamflow predictions than are possible using operational, NEXRAD data? 3) What is the benefit of optimizing the parameters in the reflectivity-to-rainfall (Z-R) equation? 4) How important is the radar's calibration on hydrologic simulations? 5) Should more effort be placed on increasing the density of rain gauge networks, deploying gap-filling radars, or improving techniques to adjust radar data based on observed or modeled vertical profiles of reflectivity (VPR)?

An ensemble is created by perturbing the sensitive parameters in the hydrologic model, and the skill of each model ensemble is assessed and compared to that of another ensemble that uses different QPE inputs. Experimental datasets from a portable, C-band Doppler radar that were collected over the American River Basin (ARB) in northern California during the winters of 2005-2007 as part of NOAA's Hydrometeorological Testbed Experiment are used in this study.