What is Forthcoming for the Implementation of HRRR version 4 in 2020 and Beyond (Credit: NCEP Global Systems Division and Environmental Modeling Center)
Modeling is crucial to meteorology. We all rely on the output of the various computer models and their unique configurations, and subsequently the human forecaster skill and intellect, to provide the most accurate and reliable short and long-term forecasts. When it comes to the long-term models, the global forecast system (GFS) and European Center for Medium-range Weather Forecasting (ECMWF) systems are the two predominant long-term, large-domain systems in operations at the moment. The High Resolution Rapid Refresh model (hereafter HRRR), a model aimed at predicting weather at a shorter spatiotemporal scale, has undergone significant changes since its initial implementation into the operational meteorology world in 2012. The HRRR The newest version of HRRR, version 4, is expected to launch into operations beginning in spring of 2020.
So what is new with this upcoming version? For brevity, One of the more significant changes to the model lies in the configuration. The HRRRv4 is expected to run on the Advanced Weather Research and Forecasting dynamical core (WRF-ARW) version 3.9. Several physics changes are expected to be introduced, including enhancements to the Mellor-Yamada-Nakanishi-Niino planetary boundary layer scheme for improved resolution of sub-grid-scale clouds and cloud condensation and ice parameters. Aerosol source-and-sink terms for fire, dust, and smoke particulates. The land-surface/snow model and gravity-wave drag terms are also enhanced. The HRRR from the start utilized the Gridpoint Statistical Interpolator (GSI) data assimilation technique, and HRRRv4 enables the use of GOES-16 sky radiances. This addition is important because satellite-based radiance observations are critical for determining cloud-top temperatures which can potentially give clues to cloud top heights and thunderstorm intensities. HRRRv4 retains the 3-km grid spacing and 36-member configuration to generate a 1-hour pre-forecast with all available data assimilated but now will include the use of METAR cloud observations. Lastly, the MODIS albedo scheme will be replacing the 1-degree albedo. HRRRv4 was tested this past spring at NCEP during the 2019 Spring Forecasting Experiment.
Going forward into 2020 and beyond, another set of sweeping changes is expected to be rolled out for the HRRR. A fundamental change forthcoming will be the transition away from the WRF-ARW dynamical core to a unified modeling dynamical core known as the finite-volume cubed-sphere (FV3) worked on by the Geophysical Fluid Dynamics Laboratory. Intended to increase computational efficiency, it is already in full use within the global forecast system (GFS) since mid-summer 2019. It is expected that the HRRRv4 will be the last to use WRF-ARW in favor of FV3. Additionally, the phase-out of the GSI data assimilation scheme will occur in favor of a new unified DA framework known as the Joint Effort of Data Assimilation Integration (JEDI), with one key advancement being the integration of modernized programming techniques while retaining the speed and reliability of traditional lower-level programming code even with finer-resolution data which can take up far more space. Early tests with the JEDI framework show enhancements to several observation components with reduced observation-to-model errors. Also, the push towards finer resolution is allowing for some other model suites to take form, such as the National Severe Storms Laboratory Warn-on-Forecast system. Sub-hourly data assimilation and short-term forecasts to 6-hours will help to provide forecasters and researchers an enhanced insight into potentially high-impact, hazardous weather.
So as one can see, there is much to look forward to in the world of meterorological modeling that may enhance forecast accuracy and reliability. Within the next 3-5 years, it will be interesting to see what unfolds and how the modeling world will continue to grow and expand for the betterment of research, forecasting, and the general public.
Happy New Year’s Eve to all!
To learn more about weather research and other interesting topics, be sure to click here!
© 2019 Meteorologist Brian Matilla