This section describes what steps are necessary to take when using new inventory data for anthropogenic sources. For information on biogenic sources, see Section 4.4.15, “Use BEIS3”. For more information about the details of each source category (e.g., what is the difference between using MOVES emission factors or precomputed on-road mobile emissions), see Section 2.8.1, “Summary of SMOKE processing categories”.
Because this section is so spread out, we provide a small section guide here. This section includes the following subsections:
Determine the file format that you have (for preformatted data) or want to create (for new data).
Preformatted data considerations
If you have preexisting data in SMOKE input format, then you will most likely want to use that format to input to SMOKE, unless you discover that you need a different format to take advantage of a SMOKE feature not available in a different format.
If you have multiple files in a single format, then you can combine these into a single SMOKE run using a list file.
If you have data in more than one format other than the case just mentioned, you must convert all files to the same format or process them separately in SMOKE, merging the hourly, gridded, speciated emissions files using Mrggrid.
If your area inventory includes nonroad mobile sources, you can process nonroad mobile sources with stationary area sources in most cases (see next bullet for exception).
If you want to combine nonroad mobile criteria and toxics inventories, you need to process nonroad mobile sources separately from area sources (see Section 4.4.2.2, “Nonroad mobile inventories”).
Pregridded data from any source category can be input to SMOKE as an area source, but you must convert the data to an I/O API NetCDF file format first. See Section 4.4.2.7.1, “Pre-gridded inventories from the same modeling domain as an area source without source characterization or attributes” for more information.
Collect or create inventories in correct formats.
Inventory files for area sources are usually named starting with “arinv”, because ARINV is the logical file name (environment variable) that Smkinven uses to read the file. The formats available for area/non-point sources are:
List format. The list format is simply an ASCII file that lists all of the files that you want SMOKE to import in a single run.
List Grid format. The list grid format is simply an ASCII file that lists all of the global gridded inventory files that you want SMOKE to import in a single run. The List Grid format has four columns. Assigned SCC, Pollutant ID, Local Variable Name, Month of inventory and Full inventory fine directory path and filename.
ORL and FF10 formats. These formats are most appropriate for non-point toxics inventories, but also can be used for any inventory data.
Pregridded I/O API format. This format can only be used for pregridded data that has been converted to a time-independent, gridded I/O API file. The variables in the file are the pollutants. No SCCs or other source information can be provided with this format.
Set and/or check the file headers that apply for your file format.
File format header: Make sure that the first line of the list file or individual files start with the #LIST (for list files only),#LIST GRID, #ORL, #FF10 (see Section 8.2.1.2, “File format header”).
File type (#TYPE header): The type is important because it helps SMOKE determine the file format. Make sure that the header has been set according to the directions in Section 8.2.1.4, “Header records”.
Country (#COUNTRY header): Make sure that this head has been used and that the country name used in this header matches the
name in the country section of the COSTCY or GEOCODE_LEVEL[1-4] file (if USE_EXP_GEOCODE Y) file.
Inventory year (#YEAR). Make sure that the inventory year is accurately reflected in the header. This is the only way SMOKE knows what year the data is for. If this year is not a base year but a future-year, make sure to follow the additional instructions in Section 4.4.2.6, “Future-year inventories”.
Description (#DESC header): Although this header is not required, it is good practice to take advantage of it and include as much information about the data as you have. There is no limit to the number of #DESC headers that can be included in the file.
Data header (#POLID or #DATA header): Make sure that this required header is included in the file. It should accurately reflect
the data values provided on each line of the file. If there are data fields at the end of the file that you want to drop from
being imported, you can leave them off of the #POLID or #DATA header to cause them to not be imported by Smkinven (see also the INVTABLE file for dropping emissions during import).
For ORL format, this header does not apply. The pollutants are designated by code number (i.e., CAS number) in the file format and therefore the header is not needed.
Install inventory files in $ARDAT
This directory is $SMKDAT/inventory/$INVID/area
Ensure the Assigns file has the ARINV variable set to the ARINV file name for your area-source inventory file.
Optionally, set up to assign point-source locations to area sources using the ARTOPNT file.
Put the ARTOPNT file in the $INVDIR/other
Ensure the name of the ARTOPNT file is consistent with the ARTOPNT setting in the Assigns file.
Make sure that the ARTOPNT file data are consistent with the year of your inventory (e.g., use 1999 coordinates for a 1999 inventory), or that you can
accept any discrepancies.
See also steps for all source categories in Section 4.4.2.8, “Files and settings for all inventories”.
Determine the file format that you have (for preformatted data) or want to create (for new data).
See Section 4.4.2.1, “Area/non-point inventories” for issues relating to stationary area / non-point sources. These are the same issues as for nonroad mobile sources.
If you need to process nonroad mobile separately from stationary area/non-point sources because of recommendations in Section 4.4.2.1, “Area/non-point inventories”, you must separate the stationary area/non-point and nonroad mobile sources to two separate files if they are not already separated. There is no SMOKE utility to do this, but the UNIX grep command can be used to search for all SCCs in the nonroad mobile inventory; for example, all SCCs starting with “22”.
New data considerations
See Section 4.4.2.1, “Area/non-point inventories” for issues relating to stationary area and non-point sources. These are the same issues as for nonroad mobile sources.
Decide whether area and nonroad mobile sources will be processed together, which may depend on whether the available inventory data include both area and nonroad mobile or whether they are already separate. One reason for separate processing is if you anticipate control strategies that affect only stationary area or only nonroad mobile sources, though it is not required that the processing be separate to apply such controls. A second reason for separate files if you are also using toxics inventories.
If processing nonroad mobile sources separately, continue with these steps. Otherwise, the steps for area/non-point sources listed in Section 4.4.2.1, “Area/non-point inventories” will be sufficient.
Collect or create inventories in correct formats.
Inventory files for nonroad sources are usually named starting with “nrinv”, because the Assigns file uses the NRINV environment variable to set the ARINV logical file name (environment variable) that Smkinven uses to read the file. The list of available formats for nonroad mobile sources is:
List format. The list format is simply an ASCII file that lists all of the files that you want SMOKE to import in a single run.
ORL format. This format is most appropriate for criteria and toxics inventories, but also can be used for any inventory data.
FF10 format. This format is most appropriate for criteria and toxics inventories, but also can be used for any inventory data.
Set and/or check the file headers that apply for your file format.
File format header: Make sure that the first line of the list file or individual files start with the #LIST (for list files only),#LIST GRID, #ORL, #FF10.
For other headers, please refer to the documentation in step (3) of the area/non-point steps in Section 4.4.2.1, “Area/non-point inventories”.
Install inventory files in $NRDAT
This directory is $SMKDAT/inventory/$INVID/nonroad
Ensure the Assigns file has the NRINV variable set to the NRINV file name for your new nonroad-mobile-source inventory file.
Optionally, set up to assign point-source locations to nonroad mobile sources using the ARTOPNT file.
Please refer to the documentation in step (6) of the area/non-point steps in Section 4.4.2.1, “Area/non-point inventories”.
See also steps for all source categories in Section 4.4.2.8, “Files and settings for all inventories”.
Determine the file format that you have (for preformatted data) or want to create (for new data).
If using MOVES emission factors through SMOKE, create a VMT file in FF10 on-road mobile activity format (or use in combination with list format).
If using both VMT and emissions, you must use a list-format to import the two separate inventories together.
If using both criteria and toxics data and wish to have the full functionality of SMOKE for both inventories, you must use a list file to list both inventory files (criteria and toxics) for SMOKE to combine.
Collect or create inventories in correct formats.
Inventory files for on-road mobile sources are usually named starting with “mbinv”, because Smkinven uses the MBINV logical file name (environment variable) to read the file. The list of available formats for on-road mobile sources is:
List format. The list format is simply an ASCII file that lists all of the files that you want SMOKE to import in a single run.
ORL and FF10 format. These formats are most appropriate for toxics inventories, but also can be used for any inventory data.
Set and/or check the file headers that apply for your file format.
Please refer to the documentation in step (3) of the area/non-point steps in Section 4.4.2.1, “Area/non-point inventories”.
Install inventory files in $MBDAT
This directory is $SMKDAT/inventory/$INVID/mobile
Install other mobile-specific files in inventory directories.
The following mobile-specific files must also be installed in the MBDAT directory. These files can usually be copied from the example files provided with the SMOKE installation. The formats and
additional discussion about the contents of these files can be found in Section 8.9.2, “Mobile”. These files are:
MCXREF (MOVES usage only)
MFMREF (MOVES usage only)
MEPROC (MOVES usage only)
For all inputs needed for MOVES usage only, refer to Section 4.4.12, “Use MOVES for on-road mobile sources”.
See also steps for all source categories in Section 4.4.2.8, “Files and settings for all inventories”.
Determine the file format that you have (for preformatted data) or want to create (for new data).
Preformatted data considerations
If you have preexisting data in SMOKE input format, then you will most likely want to use that format to input to SMOKE, unless you discover that you need a different format to take advantage of a SMOKE feature not available in a different format.
Pregridded data must be input as an area source.
New data considerations
Hour-specific or day-specific data (including precomputed plume-rise data) are covered in Section 4.4.2.5, “Day-specific and hour-specific point and fire inventories”.
For wildfires and prescribed fires emission modeling without precomputed plume rise, you could use ORL FIRE format for both criteria and toxics inventory together. Using a master (annual) inventory file (PTINV) for fires described in Section 8.2.8.3, “ORL FIRE Format”, SMOKE requires day-specific inventory file(s) (PTDAY) described in Section 8.2.6.2, “ORL FIREEMIS Emission Format” to provide both criteria and toxics inventories for those matched sources. See more infomation about how-to-do at Section 4.4.17, “Plume Rise Calculation for Fires”.
Collect or create inventories in correct formats.
List format. The list format is simply an ASCII file that lists all of the files that you want SMOKE to import in a single run.
ORL format. This format is most appropriate for point toxics inventories, but also can be used for any inventory data.
Set and/or check the file headers that apply for your file format.
Please refer to the documentation in step (3) of the area/non-point steps in Section 4.4.2.1, “Area/non-point inventories”.
Install inventory files in $PTDAT
This directory is $SMKDAT/inventory/$INVID/point
Ensure the Assigns file has the PTINV variable set to the PTINV file name for your point-source inventory file.
For hour-specific and day-specific data, including fire data, see Section 4.4.2.5, “Day-specific and hour-specific point and fire inventories”.
See also steps for all source categories in Section 4.4.2.8, “Files and settings for all inventories”.
Point sources in SMOKE can include day-specific or hour-specific data. This can include fire emissions and plume rise data. The following steps must be performed to use day-specific and/or hour-specific data in SMOKE.
Determine the files that need to be created and their formats
For day-specific emissions data, create the list file PTDAY and one or more files with day-specific emissions in the format provided in Section 8.2.6, “PTDAY: Point source day-specific emissions”. The format requires that a list file be used to point to the actual data files, even if a single data file with emissions
data is used.
For day-specific emissions data, the daily total should ideally be based on the same time zone as is being modeled by SMOKE.
SMOKE needs to use the same output time zone (set with the OUTZONE setting) as is used in the meteorology data that will be input to the air quality model. For example, if the time zone being
modeled by SMOKE were GMT, then ideally the daily total emissions would be from 0 GMT to 23 GMT for each day.
For hour-specific emissions data that is not CEM data, create the list file PTHOUR and one or more files with hour-specific emissions in the format provided in Section 8.2.7, “PTHOUR: Point source hour-specific emissions”. The format requires that a list file be used to point to the actual data files, even if a single data file with emissions
data is used.
If using CEM data, the data files for 1995 through 2000 are available with SMOKE. A list file PTHOUR must be created to list the one or more CEM data files that will be input. To speed processing, the DATERANGE setting can
be used in the list file PTHOUR to cause SMOKE to skip the dates in the CEM files that are not needed. Detail information is described in Section 2.9.19, “Processing hour-specific CEM data”
If using hourly precomputed plume rise (e.g., for fire data), the PTHOUR file should be created as described above for emissions data, but specific variable names must be provided. The allowable
data includes the fraction of the plume in layer 1, the plume bottom height, and the plume top height.
If using daily fire inventories without precomputed plume rise, the PTDAY file should be created as described above for emissions data and specific variables ( size of acres burned, fuel loading,
and duration of fire ) must be provided for plume rise calculation and re-normalization of temporal profiles, as described
in the file format documentation in Section 8.2.8.3, “ORL FIRE Format” and Section 8.2.6.2, “ORL FIREEMIS Emission Format”.
If using hourly stack exit temperature, velocity, or flow rates, the PTHOUR file should be created as described above for emissions data, but specific variable names must be provided, as described
in the file format documentation in Section 8.2.7, “PTHOUR: Point source hour-specific emissions”.
Install inventory files in $PTDAT
This directory is $SMKDAT/inventory/$INVID/point
Ensure the Assigns file has the PTHOUR and PTDAY variables set to the PTHOUR and PTDAY file names for your inventory files.
Ensure script settings will use the day-specific and hour-specific data.
For day-specific data import and usage, set DAY_SPECIFIC_YN setting to Y.
For hour-specific data import and usage, set HOUR_SPECIFIC_YN setting to Y. This applies to using either hourly emissions or hourly-precomputed plume rise.
If using hourly data, make sure the OUTZONE setting is consistent with what will be used for other SMOKE processing steps.
The example SMOKE scripts make it difficult to not use the same OUTZONE setting for all processing steps. Nevertheless, the user is responsible for ensuring that the same setting is used for all
SMOKE processing steps, including import of hourly inventories. The OUTZONE setting will affect the hourly inventory import because the hourly emissions will be adjusted by Smkinven from the local time zones in the PTHOUR input files to the output time zone set by OUTZONE to create the PHOUR intermediate file.
If using precomputed plume rise, ensure script settings will cause SMOKE to use the plume rise data.
Set HOUR_PLUMEDATA_YN setting to Y.
For UAM and CAMX modeling, set:
RUN_LAYPOINT setting to Y (this program is not usually needed for these models)
EXPLICIT_PLUMES_YN setting to Y
For fires emission modeling with internal plume rise calculation, SMOKE compute the plume rise data with internally computed heat flux from fuel loading and burned area. See detail at Section 4.4.17, “Plume Rise Calculation for Fires”
If providing future-year hour-specific data, ensure that the dates used in the file are consistent with the base year dates. The dates must match the dates of the meteorology data that will be used by the air quality model.
For future-year modeling, you can choose to input future-year inventories to SMOKE. This is a different approach from importing a base-year inventory and using SMOKE growth to create a future-year inventory. The future-year inventory could be created outside of SMOKE and would need to be in one of the SMOKE input formats available for the source categories of interest.
A future-year inventory could also be created by SMOKE, since the Grwinven program can output emissions in ORL format. In this section, we do not cover importing base year inventories and using SMOKE to grow them to a future year; that topic is covered in Section 4.4.11, “Setup projection and control scenarios”. Following the instructions in that section, it is feasible to create a future-year inventory in SMOKE raw input format and use it as we describe in this section.
Set the #YEAR setting in the inventory input files to the future inventory year.
Set the SMK_BASEYR_OVERRIDE setting in the run script to the value of the base year.
Ordinarily, this setting would simply be set to zero, which indicates that it is not to be used and the inventory year is the same as the base year. In this case, the base year cannot be set by the inventory year, so this setting provides the actual year to use. The base year is most often the same year as the year of meteorology data you are using to model your episode. When this setting is defined by a script, Smkinven will set the base year with that setting and set the future year as the year of the inventory data.
Set the YEAR setting in the Assigns file to the base year.
Set the FYEAR setting in the run scripts to the future year.
Set the SMK_FUTURE_YN setting in the run scripts to Y.
See Section 4.2.4, “Script settings” for more information about how this setting affects the output directory structure and file names.
Smoke has two ways of processing the pregridded inventory data. If the inventory data is in native netCDF format then see Section 4.4.2.7.1, “Pre-gridded inventories from the same modeling domain as an area source without source characterization or attributes”. If the inventory is in IO-API/netCDF format then see Section 4.4.2.7.2, “Pre-gridded inventories processing for converting between different modeling domains with source characterization and attributes”
SMOKE has a limited capability to import pre-gridded data from the same modeling domain. It can import the data if it is time-independent and if the grid cell and pollutant names are the only attributes needed to identify the emissions. If SCC or other attributes are available, SMOKE cannot use these directly; though in the case of SCCs, each SCC could conceivably be tediously input as a separate dataset and modeled as a separate inventory.
If any source category is available as pre-gridded data, the following steps must be taken.
Plan to import the data as an area source.
Only SMOKE area-source processing can use pre-gridded data, no matter that actual source category of origin for the data. Essentially, once emissions data no longer has attributes such as SCCs, stack information for point sources, or country/state/county codes, it all appears to be the same type of data to SMOKE and can be processed as an area source.
This is not to say that all pre-gridded inventories can be processed correctly by SMOKE. For example, if a pre-gridded point source inventory including stack identifiers and stack parameters were provided with grid cell numbers instead of lat/lon coordinates, this could not be appropriately processed by SMOKE area processing as pre-gridded data.
If any SMOKE processing steps such as chemical speciation or temporal allocation need to be done based on SCC, this is not possible. Only a single default profile for speciation or temporal allocation can be applied to each pre-gridded inventory.
Convert data to I/O API format.
Users must convert the pre-gridded data to I/O API format. The format is like the AG file that can be output from Smkmerge. This format has each pollutant stored as a variable and is time independent. Some sort of ASCII-to-I/O API converter must
be created and used to perform this step. At this time, there is no general-purpose tool to make such a conversion released
with the SMOKE system.
Set the IMPORT_GRDIOAPI_YN setting to Y.
This setting will cause Smkinven to expect the I/O API gridded input and will request the AG logical file name instead of the ARINV file.
Ensure that the AG logical file name is set by the Assigns file or by the run script.
This logical file name must be set to the I/O API-formatted pre-gridded inventory file.
SMOKE will regrid the data from the native NetCDF-formatted pregridded inventory data (e.g. lat-lon for global inventories) to the CMAQ modeling grid, and apply a unique source ID to each grid cell to allow SMOKE to perform chemical speciation, temporal allocation from annual to hourly timesteps, and elevated source selection and vertical allocation. The pre-gridded inventory files each contain data for a different sector, the sector and pollutant contained in each pre-gridded inventory file is specified in the ARINV file using the #LIST GRID format.
To process pre-gridded data from different modeling domains, the following steps must be taken.
Set the IMPORT_GRDNETCDF_YN to Y to read the native netCDF annual inventory files.
Grdmat is used to convert the data from the input data grid (e.g. lat-lon for EDGAR/RCP/HTAP) to the output modeling grid.
Set the IOAPI_GRIDNAME_1 to specify the output grid.
SMOKE reads each grid cell in the gridded inventory file as a unique source that will have a unique source ID. The lower left corner of inventory grid will be source 1, cell (2,1) will be source 2, cell (3,1) source 3, etc
For each gridded inventory file, a Geographical Code Mask Input file (GRIDMASK) is provided by UNC for EDGAR inventory files that assigns a GEOGRID_LEVEL2 value to each grid cell.
The ARINV file is used to read in a list of gridded inventory files and to specify the SCC associated with each gridded inventory file
Set the NETCDF_POL_UNIT to kg m-2 -s
Set the NETCDF_INV_YEAR to the year of the inventory (e.g. 2010)
To apply control factors to adjust emissions by GEOCODE_LEVEL and SCC use the CONTROL packet in the GCNTL file
To apply vertical profiles to gridded inventory data using the SMOKE Utility Program Layalloc. The EDGAR inventory includes fixed vertical profiles by sector that define the percentages of emissions by altitude band. Layalloc will create 3-d output files for the gridded emissions sectors that emit above the surface layer (e.g. energy and industrial sources).
For importing all inventories, the following steps should be taken to review settings and change them as appropriate for your specific modeling case.
Assigns file settings:
Change the Assigns files inventory settings.
Following the settings in Section 4.4.1, “Change Assigns file to set scenario names, grid names, and other case-specific configuration information”, you should have created a new Assigns file for your case and set the INVID, INVEN, and INVOP settings described in Table 4.1, “Variable in Assigns file for naming files and directories”. Since the inventory input directories depend on the INVID setting, you should make sure that the value set for INVID is consistent with the location of the inventory files. For example, the ARINV files should be installed in $SMKDAT/inventory/$INVID/area$INVID label in the paths for the files installed in prior subsections is the same as the $INVID label in the Assigns file that is included in the Assigns file you will use.
Set the ABASE, MBASE, PBASE, and EBASE settings described in Table 4.1, “Variable in Assigns file for naming files and directories”. This is usually done for a new inventory, and can also be done for new scenarios with an existing inventory.
Script settings:
Double check that the correct Assigns file (e.g., the one with the correct $INVID setting) is used by the run scripts for your case.
Set the RAW_DUP_CHECK setting to Y if you want Smkinven to give an error if any duplicate entries are found in the inventory. See Section 6.13, “Smkinven” for more information.
Set the SMKINVEN_FORMULA setting to compute one or more pollutant values from teh values of other pollutants.
This is most commonly used to compute coarse particulates (PMC) from PM10 and PM2.5 using the setting “PMC = PM10 - PM2_5”.
Set the WEST_HSPHERE setting to reset any positive longitude values that are intended for the Western hemisphere to a negative value.
Set the WKDAY_NORMALIZE setting to account for the desired weekday normalization approach to be used for hourly emissions. See Section 6.13, “Smkinven” for more information.
Set the SMK_PROCESS_HAPS to read the NHAPEXCLUDE file.
This setting is only relevant when importing both criteria and toxic inventories. In this case, if there are VOCs in both
inventories and the INVTABLE has indicate which toxics are VOCs, Smkinven will attempt to integrate VOC emissions for all sources from the criteria and toxics inventories. The NHAPEXCLUDE file provides a list of SCCs for which Smkinven should not integrate the inventories and should instead drop the toxic VOC emissions to prevent double counting. Toxic emissions
of explicit model species are never dropped, however. More information is available in Section 2.9.5, “Combine toxics and criteria inventories”.
Set the SMK_ARTOPNT_YN setting to Y when Smkinven should use the ARTOPNT file to assign point source locations to area-source emissions.
Set the FILL_ANNUAL setting to Y to fill annual emissions values with average-day values, when average-day values are available but annual values
are not.
See Section 4.4.2.9, “Using average-day and annual data in the same run” for more information on why one might want to use this option.
The ORL and FF10 formats provide fields for both annual and average-day emissions values. SMOKE will import both values when
either of these formats are used and store all emissions in the SMOKE intermediate inventory files. To configure the Temporal allocation step, you have the option of using the SMK_AVEDAY_YN setting to select using either the annual or average-day data when both are available. This option and its typical usage
will be explained with the Temporal settings.
This section describes what to do when both annual data and average-day data need to be used in the same SMOKE run. This can occur when part of an inventory has been provided using annual inventories only (the average-day values are missing or incorrect) and a different part with average-day inventories only (the annual values are missing). This is only a problem when you want to run all of the sources in a single run; so, if nonroad and stationary area sources are provided with the two different temporal resolutions, these can just be run with separate SMOKE runs. If two parts of the stationary area source inventory are provided with two different temporal resolutions, then this section can help to setup the modeling correctly.
Two approaches are available in SMOKE to resolve this situation. In the first approach, one can set SMOKE to run using the average-day data, and in the second approach, set to use the annual data. These lists describe these two approaches:
Run all sources using “annual” data
Set the FILL_ANNUAL setting to Y. This will cause the missing annual data to be filled in with the available average-day data by multiplying
the number of days in the year, accounting for leap years.
Ensure that the monthly temporal profile assignments made by the ATREF, MTREF, or PTREF files assign uniform monthly profiles to any all sources that have the annual numbers filled in with seasonal values.
During temporal processing, set SMK_AVEDAY_YN to N. This will ensure that both the annual values that were available in the inventory and the annual values that were computed
will be used by the Temporal program. Since the monthly temporal profiles for the filled-in annual emissions have been set to uniform, the average-day
values will be calculated from the filled-in annual values.
Run all sources using average-day data
Set the FILL_ANNUAL setting to N. By default, Smkinven computes an average-day value when none is provided by dividing the annual value by the number of days in the year, accounting
for leap years.
During temporal processing, set SMK_AVEDAY_YN to Y. This will ensure that both the average-day values that were available in the inventory and the average-day values that
were computed will be used by the Temporal program.
Note that no monthly profiles are applied to average-day emissions, so if monthly variation is needed, it will need to be calculated outside of SMOKE. In this case, it might be easier to run all sources as annual, as described above.