Quantifying Alaska Pacific River Forecast Hydrologic Model Performance Relative to Different Precipitaiton Forcings

Accurate hydrologic forecasts are essential to ensuring the safety and livelihood of those who live near, and rely on water for their needs. The Community Hydrologic Prediction System (CHPS) is used at NOAA’s NWS River Forecast Centers throughout the country to predict the hydrologic activity in the region. CHPS can be driven with precipitation estimates directly from weighted gauges or from gauges interpolated over a grid. This study tests which precipitation forcing provides the most skillful model simulation in Alaska, where high quality gauge observations are sparse. At the Alaska-Pacific River Forecast Center, a stand-alone version of CHPS was used to compare streamflow forecasts from both forcings through the use of three metrics: Nash-Sutcliffe Efficiency, Percent Bias, and RMSE-observation Standard Deviation Ratio. Over the years 2016-2018, nine of the twelve basins studied were modeled most accurately using a gauge-weighted forcing while the other three were modeled most accurately using a grid-derived forcing. After looking at the characteristics of each basin, it can be concluded that grid-derived forcings for basins larger than 200 mi2 may be more accurate, although there is not enough evidence to know for certain. Otherwise, a gauge-weighted forcing is likely to provide a better simulation. But there is no perfect input in hydrologic modeling, as it depends on many factors of the basins, such as size and location. Additionally, given that most of the metric values fall outside of what is considered an acceptable range in hydrologic modeling, it is clear that these forcings alone are not enough to ensure accurate predictions. It is necessary for human forecasters to evaluate the simulations before they are published.