WRF-Chem Tracers for FIREX-AQ

Description of the Regional NCAR Air Quality Forecasting System

Air quality forecasts provided on this website are based on version 3.9.1 of the Weather Research and Forecasting model coupled with Chemistry (WRF-Chem). The model domain is defined on a lambert conformal project with a horizontal grid spacing of 12 km and (390, 230, 43) grids points in (x,y,z) directions. The model top is located at 50 hPa. We use a meteorological time step of 15 sec and a chemistry time step of 2 minutes. Physical and chemical parameterizations used in the WRF-Chem set-up are listed in Table 1.

Table 1: Physical and chemical parameterizations used in the regional NCAR air quality forecasting system.

Atmospheric Process


Cloud microphysics

Thompson microphysics

Short- and Long-wave Radiation

Rapid Radiative Transfer Model for GCMs

Surface Layer

Eta Similarity

Land Surface Model

Unified Noah Land Surface Model

Planetary Boundary Layer

Yonsei University Scheme (YSU)


Grell–Freitas Ensemble Scheme

Gas-phase Chemistry

Model for Ozone and Related Tracers (MOZART)

Aerosol Processes

Goddard Chemistry Aerosol Radiation and Transport (GOCART)

Dry Deposition


Wet Deposition

Neu and Prather



Anthropogenic Emissions

National Emissions Inventory (NEI) 2014 (Monthly averaged)

Biomass Burning Emissions

Near-Real Time FINN emissions

Plumerise of fire emissions

Freitas et al. (2007)

Biogenic Emissions

MEGAN (online)

Dust Emissions

GOCART (online)

Sea-Salt Emissions

Gong et al. (1997) (online)

Meteorological IC and BC

Global Forecast System (GFS) 00Z cycle

Chemical IC and BC

Whole Atmosphere Community Climate Model (WACCM)

CO Tracers: Our model set-up includes five CO source tracers that keep track of CO emitted from anthropogenic and biomass burning emission sources located inside the domain, photochemical production of CO from non-methane volatile organic compounds (NMVOCs) emitted within the domain, and background CO flowing into the domain produced by all non-CONUS sources including non-CONUS fires. These CO tracers are subjected to the same physical and chemical processing in the atmosphere as the standard model CO but they do not affect the standard model physics or chemistry.

Inert Tracers: Eight inert tracers have been added to forecast the age of anthropogenic and biomass burning emission plumes, the effect of plumerise on vertical distribution of biomass burning emissions, the mixing of anthropogenic and biomass burning plumes, and to determine the influence of non-CONUS fires. The tracer emissions are based on CO emissions from four different sources. For each source, two types of tracers are used. A non-decaying tracer (tr17_1, tr17_3, tr17_5, and tr17_7) that experiences only transport in the model, while its twin tracer (tr17_2, tr17_4, tr17_6, and tr17_8) decays with a time scale of one-day. The logarithm of the ratio of the decaying tracer (e.g., tr17_2) to the passive tracer (e.g., tr17_1) provides an estimation of age of air for the tracer. The four emission sources these tracers track are listed below:

  1. Biomass burning emissions in CONUS are limited to the model surface layer (tr17_1 and tr17_2). (“Fire Tracer 2D”)
  2. Biomass burning emissions in CONUS are distributed vertically through plumerise parameterization of Freitas et al., (2007). (tr17_3 and tr17_4). (“Fire Tracer 3D”)
  3. Anthropogenic emissions from CONUS. (tr17_5 and tr17_6). (“CO Anthro”)
  4. Inflow of CO emitted by fires outside the CONUS from the domain boundaries. (tr17_7 and tr17_8).  (“CO Boundary Fires”)

The wind barbs on the WRF-Chem plots follow the meteorological convention: Circles are calm. Barbs are worth 10 knots, half-barbs are 5 knots. Filled flags are 50 knots.

Model Output: To reduce the storage requirements, we output the full standard three-dimensional WRF-Chem output every 6 hours but key air quality and meteorological variables at the surface every hour. Table 2 provides a list of the variables included in the hourly output along with the dimension information.

Table 2: Hourly WRF-Chem output description.

Variable Type


Variable Names

Air Quality


o3_sfc, no_sfc, no2_sfc, so2_sfc , ho_sfc, ho2_sfc , co_sfc,  hcho_sfc,  c2h4_sfc,  ch3oh_sfc,  ch3cho_sfc,  isopr_sfc,  tol_sfc,  ch4_sfc,  acet_sfc,  c2h6_sfc,  c3h8_sfc,  c3h6_sfc,  bigene_sfc,  bigalk_sfc,  pan_sfc,  ald_sfc,  pm10_sfc, pm2_5_dry_sfc, BC1_sfc, BC2_sfc, OC1_sfc, OC2_sfc, DUST_1_sfc, DUST_2_sfc, DUST_3_sfc, DUST_4_sfc, DUST_5_sfc, SEAS_1_sfc, SEAS_2_sfc, SEAS_3_sfc, SEAS_4_sfc


O3, co, co_anth, co_fire, co_chem, co_bdry, co_bdry_fire



Tr17_, tr17_2, tr17_3, tr17_4, tr17_5, tr17_6, tr17_7, and tr17_8

Aerosol Optical Properties


AOD300, AOD400, AOD550, AOD600, AOD999,

SSA300, SSA400, SSA600, SSA99



T2, Q2, PBLH, SWDOWN, PSFC, U10, and V10






Evaluation of Forecasts: A near-real time evaluation system has been developed for continuous verification of the WRF-Chem forecasts. This system collocates  WRF-Chem forecasts with the near-real time near surface ozone and PM2.5 measurements from the EPA AirNOW network and generates maps of observed and forecasted ozone and PM2.5 for each forecast hour. The system also generates time-series plots of observed and forecasted ozone and PM2.5 averaged over all the sites as well as separately for each of the ten EPA regions and Colorado. Forecast performance is also quantified for each site, EPA regions, and Colorado in terms of correlation coefficient, mean bias, and root mean square error. The evaluation plots for every day’s forecast since 1 Jun 2019 can be seen on our website. All the information is saved in netCDF files for easy access by the users and further processing.

O3_d01 and O3_d02 are results from a WRF-Chem simulation using different emissions (trend adjusted NEI EPA 2014) and chemistry (T1-MOZCART).


Kumar R, Bhardwaj P, Pfister G, Drews C, Honomichl S, D’Attilo G. Description and Evaluation of the Fine Particulate Matter Forecasts in the NCAR Regional Air Quality Forecasting System. Atmosphere. 2021; 12(3):302. https://doi.org/10.3390/atmos12030302

Return to FIREX-AQ Forecasts.