Model specifications and input fields for biomass burning project


The meteorological model, WRF

The WRF and CMAQ domains have sizes of 470×310 and 459×299 for 12-km domain, respectively. Both WRF and CMAQ share the same vertical structure (total 15 vertical layers) since no layer collapsing has been employed in MCIP.

Input analysis data: We have evaluated existing analysis datasets and decided to use NCEP’s (National Centers for Environmental Prediction) NARR (North American Regional Reanalysis) as input. The NARR data are based on an NCEP Eta 221 regional North American grid (Lambert Conformal) (see: http://www.nco.ncep.noaa.gov/pmb/docs/on388/tableb.html) at 29 pressure levels. Its horizontal resolution is 32-km and the frequency is 3-hourly. An alternative to NARR is the Eta-NAM analysis data. However, data frequency is reduced from every three hours to every six hours starting in 2013. Our validation tests showed it is not as good as NARR for WRF input, probably because of lower temporal resolution.

Major WRF configurations: Implemented WRF options are shown in the table below. First guess and boundary conditions will be from NCEP NARR analyses. Grid nudging is turned on with the same NARR analysis data.

WRF Version V3.6.1
Microphysics Lin et al. Scheme
Long-wave Radiation RRTMG
Short-wave Radiation New Goddard scheme
Surface Layer Option Monin-Obukhov with CB viscous sublayer scheme
Land-Surface Option Unified Noah LSM
Urban Physics None
Boundary Layer YSU
Cumulus Cloud Option Kain-Fritsch
FDDA Grid and 1-hr observation-nudging

The chemical transport model, CMAQ

Major CMAQ configurations are shown in the table below. All of these options have been tested by the UH modeling group.


CMAQ version V5.0.1
Chemical Mechanism CB05 gas-phase mechanism with active chlorine chemistry, updated toluene mechanism, fifth-generation CMAQ aerosol mechanism with sea salt, aqueous/cloud chemistry
Lightning NOx emission Included by using inline code
Horizontal advection YAMO
Vertical advection WRF omega formula
Horizontal mixing/diffusion Multiscale (multiscale)
Vertical mixing/diffusion Asymmetric Convective Model version 2
Chemistry solver EBI optimized for the Carbon Bond-05 mechanism
Aerosol AERO 5 for sea salt and thermodynamics
Cloud Option ACM cloud processor for AERO5
IC/BC source Default static profiles

Dynamic chemical boundary conditions: One of the drawbacks of the standard CMAQ model is the fact that it uses temporally static boundary conditions, implying that the concentrations of model species over the model grid cells do not exhibit any diurnal variation. This could potentially bias model-measurement comparisons, especially at grid cells near the model lateral and upper boundary. In order to mitigate this potential source of model-measurement error, we used input boundary conditions simulated by a global Chemical Transport Model (CTM). In this space, we ran the GEOS-Chem model (Bey et al., 2001) for the years of 2011 to 2014 to generate 3-D gridded species concentrations over a lateral grid (2°×2.5° GEOS-Chem spatial resolution regridded for the new CMAQ chemical boundary conditions) over the Continental United States. An additional advantage of using GEOS-Chem is that it has 47 vertical profiles stretching from the surface to about 80 km, which makes it possible to provide vertical boundary conditions over a larger height. This could help simulate the long-range transport of ozone from wildfires and capture stratospheric impact on the surface ozone.


Emission processing using SMOKE

Emission modeling was performed with the Sparse Matrix Operator Kernel Emissions (SMOKE) model. The 2011 National Emission Inventory (2011 NEI) generated by the Environmental Protection Agency (EPA) was used to estimate hourly emission rates from anthropogenic sources for the continental U.S. Emissions from natural sources were estimated with BEIS3 (Biogenic Emission Inventory System version 3; additional details at http://www.epa.gov/ttn/chief/conference/ei11/modeling/vukovich.pdf). Mobile emissions were processed with MOVES. Various surface NOx emissions were prepared for inverse modeling. Again, in this project, we used the latest 2011 NEI emissions “as is”, that is, without adjusting for possible emission changes. A brief summary of the emissions data used in this emissions modeling platform follows:

        2011 platform v6.1 represents all platform sectors (area, nonroad, and so on) other than onroad mobile sources;
        For onroad mobile source emissions, the latest 2011 platform v6.2 based on the latest Motor Vehicle Emissions Simulator (MOVES) 2014 was used.

More details on SMOKE modeling are available here .


Conversion of the FINN inventory to CMAQ format

The fire inventory from NCAR (FINN) version 1.5 provides global daily emissions of trace gases and particle from biomass burnings (Wiedinmyer et al., 2011) (available at http://bai.acom.ucar.edu/Data/fire/). The data has high spatial resolution of 1 km owing to the MODIS Thermal Anomalies Product used for fire detection. The land cover/land use (LULC) of the spotted fire is initially classified by the MODIS Collection 5 Land Cover Type (LCT) product for 2005. The vegetation density in each fire pixel is assigned based on the MODIS Vegetation Continuous Fields (VCF) product. Emissions are speciated for MOZART-4, SAPRC-99 and GEOS-Chem chemical mechanisms for global and regional model applications.

The steps involved in the conversion were (a) re-gridding the FINN inventory to the model grid cell resolution (b) mapping the speciation lumping of VOC and PM2.5 emissions, since FINN was speciated for MOZART while CMAQ needed CB05 (c) vertical allocation of fire emissions. Each of these is described in detail below.

Re-gridding of biomass burning emissions: The FINN emissions were re-gridded to the CMAQ domain using the utility provided by the data developer. After re-gridding, species fire emissions were classified by fire sizes and area fractions of four vegetation types, tropical forest, extra tropical forest, savanna, and grassland.

Mapping from MOZART-4 to CB05: Since CMAQ v5.0, the primary PM2.5 emissions were split into 18 species: organic carbon (OC), elemental carbon (EC), sulfate (SO4-2), nitrate (NO3-), water (H2O), sodium (Na+), chloride (Cl-), ammonium (NH4+), selected trace elements (Al, Ca2+, Fe, Si, Ti, Mg2+, K+, Mn), non-carbon organic matter (NCOM) and un-speciated fine PM (PMOTHR). Primary unspeciated coarse particulate matter (PM) is named PMC (Simon, 2015).
In the FINN inventory, PM2.5 includes OC and EC. PM2.5 particles are a subset of PM10.

Vertical allocation of fire emissions: The vertical fraction of FINN emissions is calculated on both pressure and smoke smoldering effect.

More details on mapping from MOZART-4 to CB05 and calculating the vertical fraction of FINN emissions are available here .