Short range operational streamflow forecasting in western washington state, usa

In 2006, 2007 and 2009, large floods in western Washington caused severe infrastructure damage and transportation interruptions, with costs in the millions of dollars. Reservoir operators depend at such times on flood forecasts with lead times of up to one week as a key input for managing their storage facilities and preventing downstream flooding. We describe and present results from an operational, short-range hydrologic prediction system for small watersheds in western Washington. The system provides hourly-timestep weather and streamflow forecasts that update 4 times daily to several reservoir-managing utilities in the region. The forecasts use mesoscale atmospheric model downscaling of Global Forecast System (GFS) predictions and statistical downscaling of the Global Ensemble Forecast System (GEFS) to generate the temperature and precipitation inputs for a distributed hydrologic prediction system that represents each watershed at a spatial resolution of 150 meters. Model outputs are then statistically adjusted to reduce systematic biases and estimate prediction intervals. We evaluate the performance of the streamflow forecast system during recent significant flooding events.