Dashboard

HWRF

HWRF HURRICANE MODELING AT EMC

The HWRF group at EMC works in conjunction with members of all three EMC branches to continually make improvements to the Hurricane Weather Research and Forecast (HWRF) modeling system for each hurricane season. Extensive testing and evaluation is completed for each new model configuration before these changes are incorporated into the operational model. Collaboration from organizations like GFDL, DTC, NHC, and HRD is also essential to the hurricane team's progress.

The atmosphere-ocean coupled Hurricane Weather Research and Forecast (HWRF) modeling system runs in the NCEP production suite on the NOAA Central Computer System. This system is developed and supported by the Environmental Modeling Center (EMC) and operated by NCEP Central Operations (NCO) since 2007. HWRF consists of multiple movable two-way interactive nested grids that follow the projected path of the storm. Atmospheric component of the HWRF model was coupled to the Princeton Ocean Model (POM) developed by GFDL/URI using a sophisticated coupler developed at NCEP for providing accurate representation of air-sea interactions. An advanced vortex initialization scheme and NCEP GSI based HWRF Data Assimilation System (HDAS) provide means to represent the initial location, intensity, size and structure of the inner core of a hurricane and it's large-scale environment. The NCEP Global Forecast System (GFS) analysis and forecasts provide initial and boundary conditions for the HWRF model. POM uses a feature based initialization procedure for representing oceanic features such as the loop current, warm/cold core rings and the cold wake generated by the storm.

HWRF team at EMC has also been providing experimental real-time forecast guidance for all tropical cyclone basins in the world (including Western North Pacific, Southern Pacific, North Indian and South Indian Ocean regions) to the Joint Typhoon Warning Center (JTWC) and National Weather Service (NWS) Pacific Region (PR) with support from NOAA's Hurricane Forecast Improvement Project (HFIP) and using HFIP Research&Development computational resources on Jet supercomputers. Based on the demonstration of superior performance from HWRF compared to other regional models, JTWC has included HWRF model guidance in their operational consensus forecasts.

The objective of the HWRF team is to implement planned scientific and product enhancements to the operational HWRF annually, with an aim towards improved forecast performance using state-of-the-art numerical techniques. The HWRF project is an NCEP Annual Operating Plan (AOP) milestone which maps to NCEP's strategic goal to produce and deliver the best products and services, and prepare for a Weather Ready Nation.

Highlights for HWRF's FY2020 implementation:

  1. HWRF Science and Technical Enhancements: 
    • Use high-resolution land-sea masks for the moving nests
    • Upgrade dynamic core from WRF3.9.1 to WRF4.0a
    • Obtain three-hourly (instead of six-hourly) lateral boundary conditions from GFS
    • Optimize and unify the domain sizes for the ghost domains for initialization and data assimilation (DA)
    • Adjust the horizontal mixing length scale parameter (coac)for D03
    • Use the exponential random cloud overlap method(cldovrlp=5) with a constant decorrelation length (idcor=0) in the RRTMG radiation scheme and include a recent shortwave radiation related bug fix from Atmospheric Environment Research (AER) through Developmental Testbed Center (DTC)
    • Sync the scale-aware SAS convection scheme with a recent GFS version, but keep using the HWRF's detrainment rate
    • Use the unified Ferrier-Aligo microphysics scheme (consistent with the versions used in Hurricanes in a Multi-scale Ocean-coupled Non-hydrostatic, HMON, and North American Meso-scale, NAM) with bug fixes (collection efficiency calculation, calculation of the saturation mixing ratio with respect to water at 0 degC, limit the ice/rain fraction within 0 to 1, etc.)
    • Adopt new settings for the Data Assimilation and Gridpoint Statistical Interpolation, DA/GSI, increment blending
    • Turn off smoothing in Vortex Initialization (VI) and turn off intensity correction when model mean sea level pressure is shallower than observed and wind speed is stronger than observed wind speed
    • Skip the effects of VI but keep DA/GSI for weak storms/cycles with vmax <= 25 kts, and fall back to use Global Forecast System (GFS) analysis as initial conditions (ICs) for weak storms/cycles with vmax < 20 kts
    • Implement a new domain merging method and procedure to better handle the transition from the HWRF analysis to GFS analysis
    • Update and fix issues related to the preprocessing of the temperature dropsonde (TEMPDROP) data
    • Assimilate some additional satellite observations used in GFS
    • Unflag and turn off the thinning method for the Advanced Scatterometer (ASCAT) data
    • Assimilate the Next Generation Weather Radar (NEXRAD) radial wind data from coastal radar sites together with Hurricane Research Division, and Oklahoma University, collaborators)
    • Switch to use the Global Real-Time Ocean Forecast System. Use RTOFS data (instead of the Generalized Digital Environmental Model, GDEM climatology) to initialize the ocean model for the national basin (same as the east Pacific basin)
    • Improve the regridding of initial data from RTOFS to Princeton Ocean Model (POM) grid especially over shallow layers to fix the cold spots of Sea Surface Temperature (SST) appeared in the previous forecasts
    • Unify the POM related scripts to support both the current operational HWRF or basin-scale HWRF configurations
    • Use a newer version HYbrid Coordinate Ocean Model (HYCOM) for ocean coupling for JTWC basins and update the ocean lateral boundary condition specification
    • Use upgraded unified post-processor code.
    • Use the latest version of Geophysical Fluid Dynamics Laboratory (GFDL) storm tracker (from Tim Marchok, GFDL)

Highlights for HWRF's FY2018 implementation:

  1. HWRF Infrastructure Enhancements: 
    • Upgrade dynamic core from WRF3.8.1a to WRF3.9.1
    • Test and evaluation with 2017 4D-Hybrid GDAS/GFS initial and boundary conditions
    • Increase horizontal resolution from (18/6/2-km) to (13.5/4.5/1.5-km), with slightly reduced domain sizes for the two nested domains
    • Unify the vertical level configuration for the JTWC basins (WPAC, NIO, and SH) to be the same as the NHC and CPHC basins (NATL, EPAC, and CPAC), which has 75 vertical levels with a model top of 10 hPa
  2. HWRF Vortex Initialization and Data Assimilation Improvements: 
    • GSI code upgrades and changes to disable assimilating SSMI channel 2 data
    • Stochastic physics for self-cycled DA ensemble members
    • Admit new data sets (GOES-16 AMVs, NOAA-20, SFMR, TDR from G-IV)
    • Considering dropsonde drifting
  3. HWRF Physics Advancements: 
    • Upgrade the RRTMG scheme with a modified cloud overlap method
    • Adjust the horizontal diffusion and convergence damping coefficients
  4. HWRF Air-Sea Interaction and Coupling Upgrades: 
    • Unified HWRF/HMON coupler with double precision coordinates from the HWRF component
    • Add a POM ocean domain for the CPAC basin
    • Enable ocean coupling (with HYCOM) for Southern Hemisphere basins
    • Sea surface wave initial condition from global wave model
    • Add ocean coupling (HYCOM) for Southern Hemisphere basins
  5. HWRF Post-Processing and Product Upgrades: 
    • File name and resolution changes in HWRF storm and core GRIB2 files
    • Add the instantaneous precipitation rate variable (PRATE) in HWRF GRIB2 files
    • Change maximum number of storms running in operational from 8 to 7 storms