Event-based Streamflow Simulation Using The Trex Distributed Model and Ensemble Forecasts From The Hmt-west Project

Abstract

Short-range quantitative precipitation forecasts (QPF) not only provide useful weather information, they also add value to river forecasts. This study investigates the usefulness of high-resolution QPF in simulating streamflow over the North Fork of the American River Basin in Northern California through a spatially distributed hydrologic model. Event-based streamflow simulation is conducted for selected heavy rainfall events using the Two-Dimensional Runoff Erosion and Export (TREX) model. During the Hydrometeorological Testbed (HMT)-West campaign, ensemble forecasts at 3-km resolution were implemented for three winters (2005/06, 2006/07, and 2007/08) in the Northern California region, including four mesoscale model configurations based on the Weather Research and Forecasting (WRF) model. All models were diabatically initialized by the local analysis and prediction system (LAPS), which greatly reduces the “spin-up” problem for the 0-6-h QPF. The 0-6h ensemble-mean QPF and the 6-h, Stage IV quantitative precipitation estimation product (QPE) are disaggregated to hourly rainfall at 150 x 150 m resolution and used as independent inputs to the TREX model. The TREX parameters are calibrated for one heavy rainfall event using the QPF and QPE products and applied to the rest of the events for validation. Statistical scores are computed based on the timing of peak discharge, magnitude of peak discharge, and integrated discharge volume of the simulated discharge compared to the USGS observations. For some events, the simulation results using QPF inputs show increased value compared to the simulations using QPE inputs. The TREX model also shows deficiencies in simulating snowmelt events.