The FIM Global Model

image not available
2-m Temperature

image not available
Precipitable Water/500hPa Height

image not available
250hPa Wind

image not available
Total 5-day Accumulated Precip

A unique combination of 3 numerical design features

  • Icosahedral horizontal grid, mostly hexagons except for 12 pentagons ("I" in FIM)
  • Isentropic-sigma hybrid vertical coordinate, adaptive, concentrates around frontal zones, tropopause, similar to RUC model ("F" for Flow-following in FIM)
  • Finite-volume horizontal transport (Also under "F", for "finite-volume" in FIM)

News Items

  • 11 April 2014 - Update:
    A revised spring-2014 version of the FIM, with improvements primarily in numerics, is ready for application to the 2014 hurricane season.
    The 15km version of this updated FIM global model will provide experimental forecast guidance to the National Hurricane Center for the 2014 season, as funded under the NOAA Hurricane Forecast Improvement Project. Key improvements for 2014 are incorporation of a high-resolution MODIS land-use data-set, which refines for land-water contrast and land-use down to 15km or 10km resolution, a refined icosahedral grid (more regular polygons), improved momentum interpolation to polygon sides, improved vertical regridding, and 1/8-degree output.
  • March 2014 - Update:
    FIM9 (15km, on jet) graphical products are now being produced over North America at a 13km resolution for the US subregions (e.g.,NW, SW, NC, SC, NE, SE, Great Lakes) also available for the Rapid Refresh..
  • January 2014 - Update:
    An evaluation of the 2013 version of the FIM is available in this document.
  • 19 June 2013 - 10km real-time FIM runs now available
    (As of 8 July, suspended through October for HFIP runs). FIM is being now run at 10km resolution out to 66h at 00z each day when computer resources are available. Output graphics at full resolution over North America are available here
  • 21 Aug 2012 - Update:
    New version of FIM running since April 2012 with 2nd-order momentum diffusion, ozone physics, improved interpolation near poles. This version is being used for summer-fall 2012 FIM runs for the NOAA Hurricane Forecast Improvement Project, including a 15km version run every 6h. FIM-15km (FIM9) tropical cyclone (hurricane) track forecasts are available here
  • 14 Dec 2011 - Update:
    FIM update today: Upgrade to GFS physics suite: 1) decrease in vertical diffusion in stratosphere (above 200 hPa), 2) change to thermal roughness length. These changes were implemented in the operational GFS in April 2011.
  • 21 Nov 2011 - Update:
    A new version of FIM with significantly improved performance and faster runtime was implemented at 00z 16 Nov 2011 for the FIM real-time runs. Information and verification from a 5-month retrospective test period is available here
  • 22 Aug 2011 - Update:
    Hurricane season:
    FIMY (EnKF init conds) is being used as experimental guidance at the Natl. Hurricane Center as part of the NOAA Hurricane Forecast Improvement Project (HFIP). A 10-member FIM ensemble and a 20-member FIM-GFS multi-model ensemble are also being run as part of HFIP. Experimental track guidance available at http://ruc.noaa.gov/tracks and at http://www.esrl.noaa.gov/psd/forecasts/gfsenkf/fimens .
  • 16 May 2011 - All versions of FIM were updated for the 12z run today to allow solar radiation calculations to vary with the valid date during the simulation (previously was fixed to the earth orbit/tilt corresponding to the initial date of the forecast simulation).
  • 4 Apr 2011 - All versions of FIM were updated for the 12z run today to correct a coding error in the horizontal flux-corrected transport. The effects of this fix are significant, resulting in less diffusion overall and in somewhat tighter and stronger tropical cyclones.
  • 17 Jan 2011 - All versions of FIM updated to correct CO2 value. They had inadvertently been using zero CO2 since 13 Sept 2010.
  • 16 Nov 2010 - The 30km FIMZ is now running with a correction to the shallow convection scheme from the "new" GFS physics, ensuring that there are no negative fluxes and excessively large heating rates. This problem has been an issue for all FIM runs since 14 Sept (see next news item).
  • 14 Sept 2010 - All FIM runs are now updated to "new" (2010) GFS physics suite, including the vanilla 30km FIM run with GSI initial conditions.
  • 1 Sept 2010 - Update: 1. 15km FIM added (using new GFS physics and EnKF initial conditions) - run daily at 00z.
    2. FIMZ started using fully updated new GFS physics on Wed 25 Aug - FIMZ real-time products over the western Atlantic domain.
  • 9 July 2010 - New GFS physical parameterization package to be implemented in operational GFS on 29 July 2010 is now also running in the FIMZ version. FIMZ real-time products over the western Atlantic domain.
  • 27 June 2010 - For best FIM hurricane/tropical cyclone forecasts (per 2009 results), use FIMY with EnKF-derived initial conditions (mean EnKF from Jeff Whitaker's GFS T254 ensemble). -- FIMY real-time products over the western Atlantic domain.
  • 13 May 2010 - FIM-Chem-Ash now includes washout from precipitation/cloud processes. As of 11 May, the FIM-Chem-Ash model started to include scavenging of volcanic ash by large-scale precipitation processes. This results of much more realistic areal coverage in the real-time FIM-Chem-Ash forecasts. In addition. The real-time forecasts now include ash products for 3 different layers: sfc-20km, 20-35km, 35-50km. These are available under the FIMX real-time products.
  • 4 May 2010 - FIM-Chem-Ash now producing experimental real-time volcanic ash forecasts. An inline chemistry of the FIM global model is now running in real-time with 17 aerosol and gas-phase tracer concentrations, including 4 size bins of volcanic ash. The FIM-Chem-Ash real-time runs are now in the FIMX slot. Other key information:
    • 30-km resolution
    • 14 aerosol concentration and 3 gas-phase prognostic variables are cycled and have been since 14 April (start of explosive eruption of Eyjafjallajokull)
    • Forecasts are run out to 5-day duration for each 00z initial time, and a shorter 12-h forecast is run at 12z each day to maintain the aerosol/ash cycling.
    • Forecast products are available at http://fim.noaa.gov/FIMscp/Welcome.cgi?dsKey=fimx
    • GFS analyses are used for atmospheric initial conditions in the FIM-Chem-Ash runs, combined with cycled 3-d aerosol initial fields from previous 12-h forecast
    • These runs are experimental and still under development -- expect further changes. They are not official guidance.
    • Only eruption data from Eyjafjallajokull is currently included.
  • 12 Mar 2010 FIM (primary) now using coordinate pressure smoothing (50 m/s- diffusion velocity) -- a "promotion" of the 4Feb-11Mar FIMX. As of 12 Mar, FIMX (experimental) now is distinguished from FIM only in that it uses still smoother horizontal hybrid (isentropic-sigma) surfaces (125 m/s-diffusion velocity). FIMY (EnKF initial conditions) continues to use the same model configuration as the FIM-control model, so comparisons of FIM and FIMY are solely due to EnKF initial conditions vs. GSI initial conditions.
  • 4 Feb 2010 New FIM version in parallel FIMX slot for real-time runs - now using revised vertical coordinate with significant reduction in noise via smoothing of coordinate surface pressure for hybrid theta-sigma.
  • 27 Nov 2009 - 3 FIM variations now running at 30km
    • FIM - GSI-GFS initial conditions - hybrid isentropic-sigma vertical coordinate
    • FIMX - GSI-GFS initial conditions - hybrid sigma-pressure vertical coordinate
    • FIMY - EnKF initial conditions - hybrid isentropic-sigma vertical coordinate
    Real-time products for each version and FIM-FIMX differences (vertical coordinate) and FIM-FIMY (initial conditions) 14 Oct 2009 - FIM runs at TACC (15km, 10km) are discontinued - end of available TACC computing resources and approximate end of Atlantic hurricane season
  • 17 August 2009 Much progress in FIM development and testing - FIM is now running at 10km resolution once daily at the TACC supercomputer. Also, an ensemble Kalman filter global assimilation cycle is now being used to initialize both 15km runs and 10km FIM forecasts at TACC. Viewers can compare FIM-15km forecasts initialized by GFS vs. those initialized by EnKF. All of these TACC 10km/15km FIM runs include the changes reported for GSD FIM runs on 26 July (bullet below).
  • 26 July 2009 FIM-30km-GSD runs now include random variations in cloud-top in the convective parameterization, a further correction to initial soil moisture, and a restructuring of the reference potential temperatures in the upper stratosphere to lower mesosphere to improve model stability. (r671) These changes will be added to TACC FIM runs in the near future, along with others.
  • 24 July 2009 Products from new ensemble-Kalman-filter initialized 15km FIM runs (be sure to click on FIM-ENKF on left) are now available to compare with counterpart 15km FIM runs initialized with GFS analysis . Both of these are run on the NSF TACC Ranger supercomputer. This is the first time that the FIM model has been run by any options other than the GFS analysis, and these FIM-EnKF runs will be compared with the FIM-GFS-analysis runs as part of NOAA's HFIP.
  • 10 July 2009 15km FIM runs have returned, initialized once daily (00z), NOAA-NSF agreement renewed for use of TACC for NOAA Hurricane Forecast Improvement Project (HFIP) (previous agreement expired in March09). Also, changes coming soon on cumulus parameterization cloud top and higher-resolution topography.
  • 20 Mar 2009 Correction to initialization and evolution of liquid (and frozen) soil moisture in GFS/Noah LSM currently used in FIM. (r577)
  • 26 Feb 2009 Correction to land-use table used for 13-class land-use in GFS/Noah LSM in FIM. (r559)
  • 30 Dec 2008 Introduced momentum damping for winds > 100 m/s above 5 hPa, found to be necessary after seasonal changes began to develop polar-night jets. (r487)
  • 25 Nov 2008 Snow accumulation at surface added, less noise (vertical adjustment) (FIM r456)
  • 17 Nov 2008 Correction to atmospheric radiation time error (significantly improves 12z init FIM forecasts). Shallow convection added on 10 Nov.
  • 5 Nov 2008 64-level version introduced with pressure-top at 0.1 hPa (previous - 50-level with ptop at 20 hPa). Also added accurate use of sea-ice fraction. (FIM r392).
  • 17 Oct 2008 Important FIM changes for GSD-FIM30(G8) and TACC-FIM15(G9) runs as of today: Fix to initial height problem, now using initial cloud/condensate fields from GFS, physics called now every 3 min (less frequently but matching GFS), run time reduced 20-25%. (FIM r383).
  • 7-9 Oct 2008 Isentropic-Hybrid Modeling Workshop - NOAA/ESRL
  • 22 Sept 2008 15km FIM global model runs out to 10-day duration since 28 Aug, output every 3h - Real-time products here. - Status here. Twice daily forecasts run on TACC (Texas Area Computing Center) supercomputer as part of NOAA-TACC demonstration for improved hurricane forecasts.
  • 30 Aug 2008 Two key problems fixed for FIM over last 9 days, one concerning interpolation of GFS initial conditions to the FIM grid (fixed 22 Aug), and one concerning initializing soil moisture (fixed 28 Aug). Both of these bugs were significant. (FIM r317) Also, regional output is now available for the western Pacific . .
  • 17 July 2008 Regional products added for Arctic and western N. Atlantic . Also, sensible and latent heat flux products added for all global and regional plot domains.
  • 2 July 2008 Regional products now available for Africa and CONUS . Other regions under development.
  • 2 July 2008 Output redesign for FIM model for isobaric grids on icosahedral grid
  • 4 June 2008 Error in land-surface specification corrected.
    • Drying trend in forecasts removed
    • Improved overall forecast skill
  • 8 May 2008 Twice-daily forecasts now out to 168 h (7 days) since late April.
  • 17 April 2008 Virtual temperature effect added to calculation of pressure gradient. Prognostic temperature variable is now virtual potential temperature. Average global precipitation and mean zonal wind at jet level increased by about 10% in spring 2008 cases. Change made on 15 April.
  • 8 April 2008 Real-data FIM forecasts started in Feb 2008. Graphics here.
    • GFS initial conditions, interpolated from GFS spectral data for analysis
    • 30km horizontal resolution for FIM runs
    • 50 vertical levels (changed to 64 levels with ptop = 0.1 hPa - 5 Nov 2008)
    • Use of GFS physical parameterizations (other options to be added including WRF physics options and WRF-chem as a further option)