平特五不中

Evaluating Precipitation Forecasts from a High-Resolution Ensemble Kalman Filter (HReNKF) Over the Pacific Northwest

Abstract

The Pacific Northwest endures some of the heaviest precipitation on the planet - parts of this region receive more than three meters a year. Therefore, producing accurate precipitation forecasts for these areas is essential to warn different sectors of society of the timing and intensity of precipitation events. However, forecasting precipitation remains difficult as it is a chaotic and non-linear process, and is even harder over regions of mountainous terrain. In addition, observational precipitation estimation is difficult due to sparseness of rain gauge data and limitations of radar coverage in complex terrain. In an effort to improve current forecasts, Environment and Climate Change Canada are developing a HReNKF for ensemble prediction over this region, with a 2.5km resolution and an enhanced ability to assimilate observations. To evaluate these forecasts we develop a gridded observational precipitation product that merges radar and rain gauge data, called 'kriging with external drift'. This verification product is used quantitatively to evaluate the performance of the HReNKF and of a lower resolution (15km) regional ensemble (ReNKF) using a variety of verification methods.