1. AIRPACT-5 Fire Emissions Processing Methodology
Farren Herron-Thorpe
WF Smoke and AQ Forecasting Workshop
Nov. 28, 2017 1

2. Overview: Summer 2017 AIRPACT-5 Fire Methodology
Recent/Upcoming Changes 2

3. AIRPACT-5 Fire Emissions
Summer 2017 Methodology
SMARTFIRE-2 is queried at 10 pm
Fire locations and sizes are obtained
NOAA HMS fire-detects
Infrared fire perimeters from GEOMAC
Emissions spin-up (24-hrs) and “persistence” is used
This means that AIRPACT is modeling fire locations that are from 2 days prior to the forecast date.
FCCS fuel type is obtained using BlueSky 3.5.1
Fire emissions and heat flux are obtained using lookup tables that were derived from a sensitivity analysis
Allows fire emissions to be processed in less than 5 minutes, even when there are 1000s of fire locations
Fuel-type specific
Emissions scale linearly
Heat flux scales non-linearly 3

4. AIRPACT-5 Fire Emissions
Summer 2017 Methodology

5. AIRPACT-5 Fire Emissions
Summer 2017 Methodology
VOCs are multiplied by 3
Addresses issue that FEPS VOC EF is possibly low
Implemented to allow more SOA production
May be leading to ozone over-estimates
1-inch of duff added to select fuels
(unknown, sagebrush, grasslands)
Canadian Fire emissions obtained from BlueSky Canada
SMOKE processor does speciation and plume rise
Divide heat flux by 3 to deal with SMOKE bug
Boundary Conditions derived from MOZART (NCAR)
underestimated PM2.5 (switched SAPRC to CB05)
bug fixed in late August 2017 5

6. AIRPACT-5 Fire Emissions
Recent Changes (Oct. 2017)
New scheme addresses SMARTFIRE fire-type problems
All HMS locations not within a GeoMac perimeter are set to “Rx (Broadcast)” or “WF” by month
“WF season” is June-Aug for WA/ID/OR/MT
Implemented new fire-type icon scheme (w/ season var.)
Wildfires are red
Prescribed fires are blue
Possible Agricultural Burns are green (unknown fuel)
Possible Pile Burns are green/red (county list file)
Rx pile burns are often detected by HMS but get treated as broadcast burns in BlueSky.
Emissions are sometimes overestimated due to large size and fuel loading assumptions
Emissions are now recalculated for pile burns 6

7. AIRPACT-5 Fire Emissions
Recent Changes (Oct. 2017)

8. AIRPACT-5 Fire Emissions
Upcoming Changes (Dec. 2017)
Modified Plume Rise Approach
Based on Wei Zhang’s (IDEQ) approach
Addresses two problems in the default BlueSky/AIRPACT approach
Smoldering fraction is too low:
Change implements new smoldering fraction by plume class based on the literature (WRAP/ DEASCO3)
SMOKE calculates smoldering fraction as a function of fire area.
New “virtual area” provided to SMOKE
Plume rise is too high for large fires:
Change assumes multiple fronts for big fires and reduces the heat flux which is used for plume rise
SMOKE calculates plume rise as a function of heat
New “virtual heat” provided to SMOKE 8

9. Most Updated AIRPACT Modeling
Updated profiles for PM/VOC
HFLUX corrected for Plume Rise(IDEQ)
Improved anthro SOA
In-line photolysis calculation
SAPRC07 mechanism
Highlights in this modeling setup:
Anthro emissions from NEI-2014
WRF has full 37 layers
CMAQ, version 5.2
- Five new SOA species from anthropogenic VOCs were added to AERO6 in v5.1+ that were not present in v5.0.2
- In-line photolysis rate calculation to support feedback between meteorology and aerosol
Modified EFs
for PM/VOC
Fire-SOA project: Tsengel Nergui, Yunha Lee, Brian Lamb (Washington State University), Serena Chung (US. EPA), Robert Yokelson (University of Montana) and Kelley Barsanti (University of California Riverside)

10. Wildfire locations in PNW, August 2013
C:/Users/Ts/My_R_Workplace/Fire_SOA_eval/Rscripts/BBOP2013_ R (Search string: FINAL PLOT 2 FIRE LOCATION XY plot using ggmap() with google imagery background map)

11. August 2013 simulation scenarios
HEAT Flux/3 (a bug in SMOKEv3.5.1)
Adopt a new approach for HEAT Flux (IDEQ)
VOC
PM2.5
Emissions
Profile
Base
Yes No
VOC x 3.0
Default
Sensitivity 1
VOC x 1.54
Updated
Sensitivity 2
PM2.5 x 2.23
Sensitivity 3
VOC x1.54
Sensitivity 4
PM2.5 x 2.23

12. Mean Fractional Biases for O3
BASE Case
SENS3 Case
C:/Users/Ts/My_R_Workplace/ModelTempSeparation/ScaleScript_part2/Scale_part2_model-obs_FireSOA-2013_ R (search string: FINAL3 X-Y map PLOT for MFBs or MFEs at sites)
Sensitivity Runs
Obs. Mean
(ppbv)
Mod. Mean
Mean Bias
Mean Error
RMSE
Correl.
MFB, %
MFE, %
BASE 29 35 6 10 12
0.66
22%
41%
SENS3 34 5 9
20%
40%

13. Mean Fractional Biases for PM2.5
BASE Case
SENS3 Case
C:/Users/Ts/My_R_Workplace/ModelTempSeparation/ScaleScript_part2/Scale_part2_model-obs_FireSOA-2013_ R (search string: FINAL3 X-Y map PLOT for MFBs or MFEs at sites)
Sensitivity Runs
Obs. Mean
(ug/m3)
Mod. Mean
Mean Bias
Mean Error
RMSE
Correl.
MFB, %
MFE, %
BASE 8 10 2 14
0.33
-12%
68%
SENS3 18 29
0.32 9%
74%

14. MFB for PM2.5 at AQS sites, August 2013
BASE Case
SENS3 Case
C:/Users/Ts/My_R_Workplace/ModelTempSeparation/ScaleScript_part2/Scale_part2_model-obs_FireSOA-2013_ R (search string: FINAL Bugle PLOT: Fractional bias and Fractional Errors)