This section provides some ideas on things to check after SMOKE has been run to a successful completion (and after the other QA steps described above have been completed). The review steps listed here will help ensure that SMOKE is working properly. However, these recommendations will not enable you to identify any missing sources (the source exists but was not included in the inventory) or wrong emissions sources (the source is in the inventory, but the emissions are completely incorrect), so it is up to you to understand your emission inventory and make sure that it is representative of the emissions for the regions and years of interest.
The following checks should be performed to ensure that the area-source and nonroad mobile-source processing was done properly:
Compare state and/or county totals among the following processing steps:
Inventory import
Spatial allocation
Temporal allocation for key days (e.g., weekday, Saturday, and Sunday)
Chemical speciation (e.g., add up NO/NO2 and VOC species using the SELECT DATA instruction in Smkreport with the S- prefix)
Final merge
Make comparisons for multiple grid resolutions, if applicable.
The differences and similarities revealed at each stage should be these:
Import totals (after Smkinven processing) should be the same as inventory totals outside of SMOKE (if available).
Spatial allocation totals should be the same as the import totals for all states/counties except those on a grid boundary. Comparing totals after spatial allocation among multiple grids should reveal that the emissions totals are the same in all grids for states/counties that are completely within the grids.
Temporal allocation should be different for weekdays, Saturday, and Sunday, and different from the average-weekday or average
annual value represented by the import totals. The differences should be in the correct direction: generally, area sources
are higher on weekends because of lawn/garden emissions. If you are using an average-day inventory (not an annual inventory),
confirm that monthly adjustments have not been made. Check that any holiday days have been processed as directed based on
the HOLIDAYS files (e.g., that weekday holidays have been processed as Sundays).
Chemical speciation should match total mass before and after (again, the Smkreport SELECT DATA option with the S- prefix is helpful here).
Final merged files should match between the Smkmerge and Smkreport reports.
Create report of temporal allocation factors used, by SCC. Review profiles assigned to SCCs representing the top 50%-80% of key area/nonroad pollutants for your inventory.
Create report of spatial surrogates used, by SCC. Review spatial surrogates assigned to SCCs representing the top 50%-80% of key area/nonroad pollutants for your inventory.
Create report of speciation profiles used, by SCC. Review speciation profiles assigned to SCCs representing the top 50%-80% of key area/nonroad pollutants for your inventory.
Create reports by state/SCC for key states in your domain (the SELECT REGION statement in Smkreport can be used to isolate certain states for this report). Review these reports to ensure that the totals are consistent with expectations. Compare and contrast emissions by SCC among states. Include per-capita reports as well as total emissions reports, and review these to determine if per-capita emissions differences make sense from state to state; this approach can often identify outliers in inventories (caused by improper units, double counting, or out-of-date inventory creation approach).
Create spatial plots (e.g., using VERDI) of average-day emissions for each day (these daily-total emissions can be created
by running Smkmerge with MRG_TEMPORAL set to N). Daily-total plots can also be created from the model-ready plots using the I/O API utility m3tproc. Do for all grids (if applicable) and compare.
Create temporal profile plots (e.g., using VERDI) of hourly, gridded data over the entire episode. Look for consistent patterns on consistent days (e.g., all non-holiday weekdays in a single month should have the same value). Do for all grids (if applicable) and compare. Check for holidays and ensure these were processed as Sundays.
The following checks should be performed to ensure that the point-sources processing was done properly:
Compare state and/or county totals at the following processing steps for all grids:
Inventory import
Spatial allocation
Temporal allocation for weekday, Saturday, and Sunday
Chemical speciation (e.g., add up NO/NO2 and VOC species using the SELECT DATA instruction in Smkreport with the S- prefix)
Layer fractions
Final merge
The differences and similarities revealed at each stage should be these:
Import totals (after Smkinven processing) should be the same as inventory totals outside of SMOKE (if available)
Spatial allocation should be the same as the import totals for all states/counties except those on a grid boundary. Comparing totals after spatial allocation among multiple grids should reveal that the emissions totals are the same in all grids for states/counties that are completely within the grids.
Temporal allocation may or may not be different for weekdays, Saturday, and Sunday, and different from the average-weekday
or average annual value represented by the import totals. The differences should be in the correct direction: generally, many
of the electric generating facilities (major emissions sources) are the same for all days of the week because the power plants
never shut down and their emissions do not depend much on the day of the week. Emissions for major manufacturing plants will
often be closed completely on weekends and have no emissions. If using an average-day inventory (not an annual inventory),
confirm that monthly adjustments have not been made. Check that any holiday days have been processed as directed based on
the HOLIDAYS files (e.g., that weekday holidays have been processed as Sundays).
If using day-specific or hour-specific inventory data, then it will be harder to compare the inventory totals before input to SMOKE to the emissions totals after temporal processing. This is because part of a state/county total when using hour-day- or hour-specific data include these data and part do not, but the inventory totals are often available only for the annual and/or average-day emissions values. This limitation can be rectified by calculating the emissions totals for the day- and hour-specific data outside of SMOKE, and analyzing these sources by SCC instead of only by state/county to separate the analysis of the annual and/or average-day emissions from the analysis of the day- and/or hour-specific emissions.
Chemical speciation should match total mass before and after (again, the Smkreport SELECT DATA option with the S- prefix is helpful here).
Emissions totals should match before and after layer fractions for all grids. Note that the layer fractions will be different across grids for sources that are in multiple grids because the meteorology data are different.
Final merged files should match between the Smkmerge and Smkreport reports.
Create report of temporal allocation factors used, by SCC. Review profiles assigned to SCCs representing the top 50%-80% of key point-source pollutants for your inventory.
Create report of speciation profiles used, by SCC. Review speciation profiles assigned to SCCs representing the top 50%-80% of key point-source pollutants for your inventory.
Create reports by state/plant for key states. These should include the grid cell in which the plant is located, which should be reviewed for correctness. A GIS analysis approach could be helpful for this step, to match the grid cells in the report with an independent assessment of the state/county. Major sources should be reviewed for correct emissions and correct locations.
Create reports with layer fractions for all major power plants in key states. Include grid cell in which the plant is located. Perform a reality-check on the plume rise; large power plants should have high plume rise, but the plumes should still be well distributed throughout the model layers.
Create the spatial and temporal profile plots discussed in items 6 and 7 above for area/nonroad processing, except for the spatial plots you need to first condense the I/O API model-ready files to 2-D instead of 3-D files. The I/O API utility vertot can be used to sum layers from 3-D files to 2-D files.
The following checks should be performed to ensure that the on-road mobile-source processing was done properly:
Compare state and/or county totals at the following processing steps for all grids (if applicable):
Inventory import (VMT only)
Spatial allocation (VMT only)
Compare emissions among all days by states. Compare sample weekday emissions daily-totals with average-day VMT to see if sources with high VMT have high emissions. Calculate ratios between VMT and emissions and look for outliers.
Compare temporalized daily-totals with speciation/temporalized daily-totals of mass emissions
Final merge
The differences and similarities revealed at each stage should be these:
Import totals (after Smkinven processing) should be the same as inventory totals outside of SMOKE (if available)
Spatial allocation should be the same as the import totals for all states/counties except those on a grid boundary. Comparing totals after spatial allocation among multiple grids should reveal that the emissions totals are the same in all grids for states/counties that are completely within the grids.
Temporal allocation should be different for weekdays, Saturday, and Sunday (note: no average-day values are available for on-road mobile-source processing using MOBILE6). The differences should be in the correct direction: generally, mobile sources are lower on weekends. Check any holidays to ensure that Sunday emissions were used.
Chemical speciation should match total mass before and after (the SELECT DATA statement in Smkreport with the S- prefix can be helpful here).
Final merged files emissions totals from Smkmerge should match the temporal/speciated emissions reports from Smkreport.
Create report of temporal allocation factors used, by SCC; review assignments.
Create report of speciation profiles used, by SCC; review assignments.
Create reports by state/SCC for key states. Review these for major sources (e.g., gasoline fueled automobiles on interstates, heavy duty diesel vehicles) and to compare contributions among states to see if these make sense.
Create the spatial and temporal profile plots discussed in items 6 and 7 above for area/nonroad processing (Section 7.5.4.1, “Area, nonroad mobile, and precomputed emissions for on-road mobile sources”).