NCEP EMC Para Experimental Run  CMAQ Air Quality Forecast Change Log


 CB05/AERO-6  Developmental/Experimental Run Forecasts 






Fall 2020 AQM 6.0 upgrade
10/19/19  para13
  • FV3 Precip accumulation coupling error fixed.  Retrospectives from June 2019 ongoing.
 07/15/19  para12
  • Error in GBBEPx FRP product found and corrected by NESDIS.
  • Switched to ARL 1.0/.25 scale factor
  • Switched to EPA diurnal smoke emissions scale factor as used by ARL
07/01/19 para12
  • Switched to use NESDIS Global Biomass Burning Emission Product (GBBEPx) on 0.1x0.1 lat-lon grid
    • Plume rise calculated from GBx Fire Rad Power (FRP)
    • BlueSky smoke diurnal emissions profile
    • Scale factor 0.25/0.25
    • Only wild fire emissions where forest fraction  >0.4 on .1 degree grid
  • Full lateral boundary conditions from FV3GFS-Chem (25 km) two day old forecast.
04/15/19 Para11
  • CMAQ V5.0.2 extended to 72 hours
  • Use FV3GFS V15.1 4x/day
  • Operational CONUS NEI 2014V2 anthropogenic emissions
  • Updated point sources with EGU projections
  • Inline smoke and dust emissions
  • ARL HMS fire smoke solution for 06z cycle.  Emissions:
    • daily BlueSky emissions assumed 6 hrs ( divide by 1/6)
    • Use Li Pan  flaming/smoldering scale factors (divide by 1.46)
      • Total emissions scaling: 0.114
    • WF/RX fire distinghuishment using USFS fuel type maps
    • NO Heat Adjustment
    • Briggs plume rise
                                                        2018 AQM V5.1 upgrade
Implementation : 12/18/18
10/15/18 PARA5
  • All AQM V5.1 codes and emissions transferred to NCO production and running 4x/day
08/01/18 Para5
  • Reinstated scaling factor of 4x smoke emissions during pre-analysis portion of run (4X/1X/1X).
07/25/18 Para6 Began
  • Using NEI 2014V2 emissions for area and mobile sources (para6 run)
  • Use of HMS fire activity file for 06 and 12 Z cycle runs.  All fires assumed 24 hour type
  • Included diurnal fire smoke emission profile
  • Used 2x smoke emissions during pre-analysis portion of run.  Maintained emission strength during forecast. (2x/2x/2x)
06/01/18 Para5
  • Using NEI 2014V2 emissions for area and mobile sources (para5 run)
  • Use of HMS fire activity file for 06 and 12 Z cycle runs.  All fires assumed 24 hour type
  • Included diurnal fire smoke emission profile
  • Removed scaling of 4x smoke emissions during pre-analysis portion of run.  Maintained emission strength during forecast.
    • (1x/1x/1x)
 4/27/18 Para5
  • ARL bugzilla 7049 bug fix for dynamically allocated fire points array.  Also initiailized PBOT array (plume base).  Appears fire smoke plume rise is lower and nearer source.
4/17/18  Para5
  • NESDIS SAB HMS fire activity processing now filters out all fire points with 2 hours burn duration or less to try and reduce smoke emissions and number of fire points.
4/11/18  Para5
  • NESDIS SAB HMS fire activity file (_prelim.txt) updated processing to produce automated fire points over the West and manual points over East.  More than 500 fire points were created and PREMAQ failed.  
         Fix implemented to increase number fire points to 10000.
          06/12 UTC cycles rerun.
3/10/18 Para5
  • Changed ending utc hour range label for exp bias corrected  1hr daily max ozone  from 04 to 05 UTC to match operational grib2 files.
March 1, 2018  Para5
  • Update to the  fine Particulate Matter (PM2.5) bias correction system to use:
    • Consistent training model predictions for the unified Kalman Filter Analog (KFAN) bias correction system.
    • Increase the number of observation sites for model bias correction to an average over 900 monitors.
    • Improvements to forecast extreme events by adding the difference between the current raw model forecast and historical analogs’ mean to the KFAN bias-corrected predictions.  The assumption behind this correction is that the CMAQ model, when predicting an extreme event, has some skill beyond that of its historical best analogs.   A measure of the exceptionality of such a rare event is given by the difference between the CMAQ forecast and its previous analogs , which is then added to the final bias-corrected forecast.
  • Implementation of a new unified  ozone bias correction system with the same codes and configuration as for PM2.5 except
    • the use of NOx, NOy and ozone as parameters to identify analogs
    • Improvements to address rare ozone exceedance events
  • Quantitative evaluation of AQM V5.1 CONUS upgrade
  • Update the Alaska and Hawaii domain CMAQ code to the same version 5.0.2 used for CONUS :
    • CB05 gas-phase and aero6 aerosol chemistry (155 species)
    • Improved heterogeneous, aqueous, winter-time reactions
    • Improved SOA and coarse mode PM
June 14, 2017  12 UTC   CMAQ V5.0.2 Implemented into Production
March 21, 2017  12 UTC
  • NAM V4 scheduled to be promoted to operations
February 17, 2017
December 1, 2016   06 UTC
  • Began KFAN bias correction on NAM-X CMAQ V5.0.2 parallel run.  BC run retroactively from July 1, 2016.
    • Sept. 2016 maps here  
  • Testing  NAM-X V5.0.2 No OMI NOx adjustments for August 7-31, 2016, maps here  
    • Note:  ARL doing similar run but with grid point OMI NOx adjustments
October 24, 2016   06 UTC
  • Fixed dust coupling and PM bullseye with cmaq para runs.  Retrospective runs from Sept. 2016 performed
 September 16, 2016   06 UTC
  •  4x/day cycling with smoke emissions updated with 06 UTC cycle
  • Driven with NAM-X parallel run V4.0 meteorology
  •  Run on WCOSS Cray stablized and improved run time (480 processors)
  • Problem with dust coupling near the lateral boundaries noted for both 1x/day and 4x/day cycling
July 1, 2016  12 UTC   CMAQ V5.0.2  CONUS Experimental Runs began
  • CMAQ V5.0.2 system with increased PM speciation 
  • CB05 gas-phase and aero6 aerosol chemistry (155 species)
  • Improved hererogenous, aqueous, winter-time reactions
  • Improved SOA and coarse mode PM
  • Lateral Boundary Conditions : 
    • Static GEOS-Chem 2006 climatology
    • Dynamic dust boundary conditions from NEMS Global Aerosol Capability (NGAC V2)
  • Point sources: 2011 NEI projected to 2017 (with DOE estimates and CEM data)
  • Area Sources : 2011 NEI (Canada 2012)
  • Mobile Sources: 2005 projected to 2011 with CSAPR inventories
  • Biogenics: BEIS 3.14
  • Updated smoke emissions
  • Dust:  ARL Fengsha emissions model
  • 1x/day cycle at 12 UTC
  • Available one day after real-time
February 4, 2016 12 UTC     CMAQ V4.7.2 implementated into operations (EPA Version number 4.6)
 April 21, 2015  12 UTC    CMAQ V4.7.2 Experimental Run began:
  • Updated emissions using EPA NEI 2011 base year values.
  • Increased number of vertical levels from 22 to 35.
  • Using Daily NEMS Global Aerosol Capability (NGAC) dust predictions at lateral boundaries.
  • Separate bias corrected PM 2.5 output using NOAA/ESRL bias correction system
    • See Djalalova, et al. (2015): Atm. Envir., 108, May,  pgs 76-87
              2014  RUNS
January 29, 2015  12 UTC  CMAQ experimental changes implemented into operations as V4.6.5
August 12, 2014  12 UTC
July 16, 2014  12 UTC
  • V4.6.3: Turned off gas emissions from wild fires so ozone predictions are not degraded.
  •  Using Parallel NAM to drive all CMAQ experimental predictions.
July 1, 2014  12 UTC
  • Turned on O-CONUS V4.6.3 AK/HI CMAQ runs 4x/day
June 27, 2014  12 UTC
  • V4.6.3: Correction to dust emissions
    • Mixed dust through multiple layers rather than only at the surface
    • Change from 20 to 5% of dust included into PM2.5 concentration.
June 20, 2014  00 UTC
June 13, 2014 12 UTC
  • V4.6.3 CB05/AERO-4 CMAQ EMC para run begun:
    • NTR, organic nitrate, biases high and influenced ozone production. It is now photolyzed and removed quicker (Dickerson et al., 2014) by shortening NTR lifetime by a factor of 10.
    • Modulate fugitive dust emission with a binary switch: whenever there is ice/snow suppress emission.
    • Incorporate NESIDS Hazard Mapping System “observed” wild fire  near real-time smoke emissions  (24-48 hrs old fire ponts)
    • Layer specific time step was added to speed up code.
  • 4x/day w/ 6/12 Z runs thru 48 hours
April 28, 2014 12 UTC
 April 24, 2014 12 UTC
  • V4.6.2 CB05/AERO-4 CMAQ run transferred to EMC  development real-time parallel
    •  51 species with  156 reactions, Sarwar et al., 2006
    • AERO-IV: Represent size distribution by log normal distributions of φ, geometric diameter of the particles: Aitken (φ <0.1 μm), accumulation (0.1<φ <2.5μm) and coarse (2.5<φ <10μm). New particle formation: gas conversion and nucleation. 
    • Heterogeneous hydrolysis reaction of N2O5 ( key linkage between gas and aerosol phase reactions, includes temperature and humidity impact).
    • ISORROPIA V1.7  gas-particle partitioning:  Partitioning between inorganic gas and particulate species due to thermodynamic equilibrium (v1.7 increased stability).
January 15, 2014 12 UTC
  •   CMAQ V4.6.2 CB05/AERO-4 run restarted on development machines by NOAA/ARL
    • Deposition velocity calculation over trees  were adjusted as VOCs and ozone were overpredicted
    • Increased  the minimum PBL to 50 m where high pollutants were noted esp. over marine boundary layers
    • Implemented  a hybrid lateral boundary condition where GEOS-Chem 2002 climatological LBCs were used on the Northern and Western boundaries.  Static boundaries still used on the southern and eastern boundaries since very high ozone plumes were noted when GEOS-Chem LBCs were used there.
  • Using NCEP NAM  Operational run meteorology


August 1, 2013 12 UTC
  •   Experimental and devlopmental  CONUS runs turned off at NCO with switch to WCOSS Supercomputerr

April 1, 2013

12 UTC

June 1, 2012

 12 UTC

June 1, 2011

 12 UTC

  • Updated CEM point source data from 2008 to 2009.
  • Updated DOE projection factors from 2010 to 2011. 
  • Changed the number of eletricity marketing module (EMMs) from 13 to 22, consistent with the changes made in DOE models.
  •  Removed average fire emissions from area sources in the experimental domain.



July 6, 2010

12 UTC

May 26, 2010

 18 UTC

  • Post-processing upgraded for efficiency.  Now using cmaq2grib codes.  Ozone day 1 Daily Maximum predictions added to NDGD files and display.

May 1, 2010


  • Point, area and mobile emissions were upgraded based upon recent EPA National Emissions Inventory (NEI,  2005) and then projected for the current year. EGU sources use 2008 CEM data projected for 2010.  These emissions are also consistent with the CB05 chemical mechanism.


August 12, 2009


  • NAM-CMAQ V4.6 production AQF model capability declared operational on NCEP IBM Power 6 Computing platform.
  • A correction to Biogenic emissions cycling from 24 to 6 hour cycling was implemented with the machine switch.

May 1, 2009

12 Z

  • Point, area and mobile emissions were upgraded based upon recent EPA National Emissions Inventory (NEI,  2005) and then projected for the current year. EGU sources use 2007 CEM data projected for 2009.  These emissions are also consistent with the CB05 chemical mechanism. 
  • For mobile sources, the  EPA Office of  Transportation and Air Quality (OTAQ ) was used in addition to 2005 NEI v1 emission data sets.  
  • A ten day test showing the minor impact of these upgrades are shown here.


December 16, 2008


Changes to the NAM  driving meteorological model forecast system:

  1. The background for the first (tm12) analysis in each NDAS run is now from the GDAS instead of the previous NDAS run (so-called "partial cycling"). Land states are still fully cycled from the previous NDAS cycle.
  2. WRF-NMM Model changes (also implemented into the DGEX):
    1. The PBL/turbulance schemes were modified to mix each hydrometeor species in the vertical.
    2. To apply vertical diffusion for separate water species, the model was changed so that (a) it can apply vertical diffusion to an arbitrary number of species, (b) the counter gradient option can be applied to some or all of the species if desired, and (c) option to set to zero some or all of the surface fluxes is also made available.
    3. In the radiation parameterization, the absorbtion coefficients for water and ice have been doubled to 1600 and 1000, respectively
    4. Changes to land-sfc physics:
      1. Let the potential evaporation decrease linearly with Bulk Richardson number under stable condition, and weighted by snow coverage.
      2. Let the slope of saturated humidity function wrt temperature decrease linearly with snow coverage.
  3. Changes to GSI analysis
    1. Use latest (1Q 2008) version of the GSI analysis code
    2. Assimilate METOP radiance data
    3. Assimilate TAMDAR/AMDAR aircraft data
    4. New version of Communitity Radiative Transfer Model
  4. Use AFWA 1/16 bedient snow depth analysis
  5. Use WPS (instead of WRF-SI) codes to process GDAS first guess input files, which are used as a first guess to the first (tm12) GSI analysis in the NDAS

         August 19, 2008

    12Z Cycle

The  biogenic emission files for the experimental run CB05 chemical mechanism were incorrect and did not include some secondary organic aerosols.  The point source emission files were also incorrect.  These files were corrected for the experimental run mechanism.  Retrospective testing showed that ozone forecasts were decreased by 3-4 ppb overall while developmental PM 2.5 forecast products were increased slightly.

June 12, 2008

   06Z Cycle

 PM2.5 developmental product predicted from the experimental run was computed incorrectly with updated CB05 emissions.  The primary organic aerosol emission was given in carbonaceous mass instead of aerosol mass.  The error resulted in an underestimate of PM2.5 since the upgraded emissions were implemented on June 10, 2008.  

June 10, 2008

   06Z Cycle

Specific experimental run system improvements include:

  • The CB05 chemical mechanism with more detailed Volatile Organic Compound (VOC) chemistry was implemented. 
  • A harmonized version of the AERO-4 aerosol chemistry mechanism was turned on. The harmonized NOx night time chemistry is configured with heterogeneous pathways turned off but gas-phase N205 hydrolysis turned on.  This was done to minimize the impact of aerosol chemistry on the ozone foreceast.
  • Sea salt emissions were included in CMAQ.
  • Point, area and mobile emissions were upgraded based upon recent EPA National Emissions Inventory (NEI,  2005) and then projected for the current year. EGU sources use 2006 CEM data projected for 2008.  These emissions are also consistent with the CB05 chemical mechanism. 
  • For mobile sources, the  EPA Office of  Transportation and Air Quality (OTAQ ) was used in addition to 2005 NEI v1 emission data sets.  Use of OTAQ on-road emission estimates is a departure from the temperature dependent regression approach used in previous years.
  • Experimental and Developmental products from these runs are now cycled every 6 hours with forecasts to 48 hours at both 06 and 12 Z cycles.
  • Impacts of these potential changes are shown here

Additional Developmental Products:
These additional products are  restricted to designated scientists and  state forecasters through the EMC  restricted access AQF web page.

Summer 2007 Experimental Run Changes

September 18, 2007

  12 Z run

 Experimental CONUS Run moved to operational Status

        July 18, 2007

 12 Z run

1. An error in the dry deposition velocity calculation was identified and corrected for selected species (NO, NO2, and CO). The error is associated with the mesophyll resistance which was erroneously set to zero. The impact is a reduction in deposition velocity and a resultant increase in predicted concentrations of these species. Higher NOx concentrations then result in higher predicted O3 concentrations. Test simulations confirm this trend with widespread enhancements in predicted O3 in regions with vegetation (primarily in the East U.S.). The max 8-hr O3 increased by 2-5ppb for cases during May and June 2007. The direction of the predicted O3 change was in the correct direction, i.e., reduction in under-estimation in the current version.

2. An error was corrected in the plume-rise calculations to use the correct pressure fields estimated from the NAM Hybrid vertical coordinate. Test cases during June 2007 showed minimal impact.

June 12, 2007

 12 Z run

  • BL mixing scheme in CMAQ has been change from RADM version 2.6 (Byun et al. 1999) using the NAM TKE PBL height to Asymmetric Convective Model-2 (ACM-2) using a Richardson calculation in PREMAQ (Jon Pleim, JAM in press).
  • Minimum eddy diffusivity used in ACM-2 reduced from 0.5 to 0.1 m^2/s

May 29, 2007

 18 Z run

  • Two Summer 2006 changes were not implemented on 5/25/07. These were the ACM convective and NAM clear sky radiation photolysis attenuation and  were implemented today.

May 25, 2007

  12 Z run

  • Ozone has been  underpredicted in the Conus experimental run since CMAQ was upgraded.  Therefore, Summer 2006 CMAQ configuration with RADM-2 PBL parameterization and NAM PBL height has been temporarily reinstated. Vertical diffusivity (KZL) minimum has been set to 0.1 m^2/s.

May 16, 2007

   00 Z run

  • Mobile emissions were missing from experimental run since May 1. Therefore, mobile emissions not included in CMAQ predictions from May 1-15, 2007.  Reverted back to using the 2006 Mobile emissions for now.

May 4, 2007

  12 Z run

May 1, 2007

  12 Z run

  • Area source emissions updated for 2007 projection. Updated California Non-Road NEI 2002 inventories

Summer 2006 Experimental Changes

September 5, 2006

 18 Z run

PREMAQ met processing calculations of vertical eddy diffusivities were mistakenly set from the 1st 22 layers from the NAM 60 layer predictions rather than for the actual  22 CMAQ layers.  This adversely impacted the vertical advection in the PREMAQ Jacobian calcluations by increasing vertical diffusion in CMAQ.  This was corrected and impact can be evaluated from Lee, Slides 7-17

September 5, 2006

 12 Z run

CMAQ now driven by updates to NAM model (additional NAM-Y changes implemented) that include increase to horizontal diffusion and corrections to SST initial conditions.

August 15, 2006

 12 Z run

CMAQ now driven by NAM-Y meteorology  which was implemented into NAM slot.  NAM-Y had corrections to vertical and horizontal diffusion and lake SST initialization which resulted in more realistic warmer,dryer boundary layers that increase production of ozone.  Impacts can be evaluated from Lee, Slides 7-17.

Augus 04, 2006

  18 Z cycle

CRISIS Fix:  Some Point  Source Emissions were still configured for 2005 .  These emissions files were updated for 2006 projections.

CRISIS FIX:   Major errors found in the tightly coupled PREMAQ emissions processor.  Isoprene biogenic emissions were never set, causing strong ozone underpredictions in the tightly coupled 5X CMAQ runs.  Also, some arrays were improperly dimensioned causing array bound errors.  For impact, see Mathur, slides 20-25

 July 12, 2006

    00Z cycle

  Most Emissions updated with 2006 Projections 

 June 8, 2006

    06Z cycle

WRF-CMAQ tightly coupled runs were available late for this case only at 18:40 UTC.  Removal of GFS ozone interpolation scripts(since GFS ozone no longer used for CMAQ) and optimizing job procedures are being tested for quicker turn around.

 May 8, 2006

    18Z cycle

WRF-CMAQ tightly coupled system transferred to NCO 4x/day  parallel testing and replaces the parallel NAM-Eta CMAQ system.
6 hour cycling turned on for CMAQ initialization.
06 and 12 UTC cycles are run to 48 hours.

 April 12, 2006

   12Z Cycle

WRF-CMAQ tight vertical coupling

Testing CMAQ driven by NAM-WRF parallel run with tight vertical coupling (using the same Sigma-p Pressure hybrid vertical coordinate).  WRF NMM output on all 60 sigma-pressure hybrid layers now read in by PREMAQ and collapsed to 22 CMAQ model layers.  These 22 layers are a subset of the WRF 60 level system.

  • CMAQ ACM convective parameterization
  • Constant Static Lateral Boundary Conditions (O3=100 ppb)
  •  Emissions for 2005
  • Optimized CMAQ advection scheme
  • addition of diagnostic tracer outputs

SUMMER 2005 Developmental Run Changes

 Aug. 20, 2005

   12Z Cycle

Cold start. 
Aug. 19 warm start rerun but not in time to initialize Aug. 20 cold start.
Aug. 17 run GRIB output not created.

  July 26, 2005

   12Z Cycle

CMAQ CONUS Run  now using ACM  approach to parameterize deep convection and  photolysis attenuation from NAM clear sky radiation predictions and  static lateral boundary conditions (ozone at lbcs=100ppb).
This was done to correct overmixing of ozone downward in high terrain. 

  July 16, 2005

   12Z Cycle

CONUS Run  w/S3 convection modified to use static lateral boundary conditions (ozone at lbcs=100ppb).
This was done to correct overmixing of ozone downward in high terrain.  Problem may be related to CMAQ mass adjustment scheme and NAM interpolated winds that cause spurious vertical motions near model top.

  July 9, 2005

   12Z Cycle

CONUS Run modified to replace ACM convective mixing scheme with Standard CMAQ mixing scheme except with downward convective mixing turned off (S3)
This was done to correct overmixing of ozone downward in high terrain.

  June 29, 2005

   12Z Cycle

 CONUS run was restarted.  This first run was initialized as a cold start (no atm. chemistry init fields).  However, updated emissions are used.  Successive runs(6/30/05 and on) will use 24 hour cycling to initialize CMAQ atm. chemistry with a previous runs 24 hour forecast.

  June 24, 2005

  12Z Cycle

 NCEP Develop machine crashed.  No CONUS runs available due to this hardware problem. We expect CONUS runs to be available again soon with a cold start initialization (no cycling).

  June 17, 2005

  12Z  Cycle

GFS Ozone predictions are used to specify the CMAQ lateral boundaries at only the CMAQ model top level (100 mb, level 22).  This appears to help overprediction over the Rocky Mountain areas.

  June 01, 2005

  12 Z  Cycle

The CONUS AQF (5X) 48 hour Forecasts were begun with the following options:

  • CONUS Emissions, with Special California Inventories
  • Testing the Asymmetric Convective Model (ACM) in CMAQ for pollutant mixing in deep convection (S4 configuration)
  • Use of GFS Ozone at lateral boundaries above 6 km
  • Use of NAM clear-sky radiation to drive CMAQ photolysis rates
  • Maximum cloud fraction is recomputed to be the global maximum cloud fraction from all possible cloud types
  • 24 hour cycling