8.6 Addressing systematic biases in RAP/HRRR physics for WFIP2

Olson, Joseph B., Jaymes S. Kenyon, Cooperative Institute for Research in Environmental Sciences (CIRES) and National Oceanic and Atmospheric Administration/Earth Systems Laboratory/Global Systems Division (NOAA/ESRL/GSD), John M. Brown, NOAA/ESRL/GSD, Wayne M. Angevine, CIRES and NOAA/ESRL/GSD, Curtis Alexander, Georg Grell, NOAA/ESRL/GSD, Tanya Smirnova, CIRES and NOAA/ESRL/GSD, David D. Turner, Stan Benjamin, and Michael Toy, NOAA/ESRL/GSD

The Rapid Refresh (RAP) and High-Resolution Rapid Refresh (HRRR) are real-time operational hourly updating forecast systems, run at 13- and 3-km grid spacing, respectively. Both systems use the Advanced Research version of the Weather Research and Forecasting (WRF-ARW) model, which allows WRF-community users to utilize any model components improved for forecasting in regions of complex terrain. During the second installment of the Wind Forecast Improvement Project (WFIP 2), the HRRR has been targeted for the improvement of low-level wind forecasts in the Columbia River Basin (CRB), which requires much finer grid spacing to resolve important terrain features in the Cascade Mountains and the CRB. This project provides an opportunity to set up and test a high-resolution nest (dx = 750 m) within the HRRR over the northwestern U.S. Special effort is made to incorporate scale-aware aspects into the RAP/HRRR physics to improve wind forecasts not only for operational-scales, but also for this higher resolution application.

Many wind profiling and scanning instruments were deployed in the CRB in support the WFIP 2 field project, which began on 01 October 2015 and ended 31 March 2017. During the project, several forecast error modes have been identified; the most systematic are the low-bias in the depth of cold pools during the winter and the low wind speed bias in thermal trough-induced gap flows during the summer. Most effort has been focused on developing the Mellor-Yamada-Nakanishi-Niino (MYNN) turbulent mixing scheme to improve upon these biases, but investigating the effects of horizontal mixing has also helped improve some of the common model biases. This presentation will highlight the testing and development of these model components, showing the improvements over original versions for temperature and wind profiles. Examples of case studies and retrospective periods will be presented to illustrate the improvements. Ongoing and future RAP/HRRR physics development will be noted.