36 An Air Quality Modeling System Providing Smoke Impact …
233
36.2 Extended Forecasting System: HiRes-X
The HiRes-2 system is operational since 2015, forecasting air quality in Southeastern
USA over a domain centered on Georgia. The system uses WRF (version 3.6) and
CMAQ (version 5.0.2) models for meteorology and air quality computations. In
addition, using the DDM-3D feature of CMAQ, it provides daily forecasts of the
potential contribution of prescribed fires to air quality. In HiRes-X, the burn activity
forecast coverage, formerly limited with Georgia, has been expanded to other states
in the Southeastern USA. A classification and regression tree (CART) was built using
meteorological data and either burn permit data in Georgia or satellite-based burn
location and area data in other states, for recent years. The satellites cannot always
detect the prescribed fires in Southeastern USA due to their small size and low heat
intensity, overlaying canopy in understory burns, and frequent cloud cover. On a
given day, if more than a threshold amount of land (e.g., 40 hectares), which varies
by state, was permitted or detected to burn in any county, that day is defined as a
“burn day”. The CART branches out at specific meteorological parameter values
with each branch leading to a “burn day” or a “no-burn day”.
If a burn day is forecasted, burns of sizes typical for lands owned by institutional,
large commercial and small private burners totaling the daily average burn area in
the region are randomly distributed to the respective lands. The Fuel Characteristic
Classification System (FCCS) maps are used to determine the fuel loads at each burn
location. Then, the fraction of the fuel that will be combusted is calculated using the
CONSUME model together with forecasted meteorological parameters such as fuel
moisture. Next, fire emissions are calculated by multiplying the amount of fuel mass
consumed with the emission factors (EFs) for different pollutants. The EFs used here
are specific to the fuels in the Southeastern USA and were derived from prior field
measurements. Finally, the plume height is calculated from the heat released and
the emissions are vertically distributed and injected into the corresponding layers of
CMAQ.
Other than the above forecasting products generated in-house, the system is also
fetching datasets from third parties. These datasets are ozone and PM 2.5 observations
from the national air quality monitoring network (https://www.airnow.gov) and fire
locations and burned areas from Hazard Mapping System (HMS) Fire and Smoke
Product (http://www.ospo.noaa.gov/Products/land/hms.html), both downloaded on
a daily basis, and burn locations and areas from Florida and Georgia’s burn permit
records, updated on an annual basis.
Air quality observational data are ingested into a MySQL database tables immediately after being downloaded onto the local server. HMS detection datasets are
transformed to a simple text format before archival. Permit data undergo quality
checks, during which missing items are filled, if possible. For example, latitude and
longitude of the burns, if not provided, are derived from provided addresses by utilizing Google Earth services. Finally, the checked records are ingested into the MySQL
database while inaccurate or incomplete records are discarded.
233
36.2 Extended Forecasting System: HiRes-X
The HiRes-2 system is operational since 2015, forecasting air quality in Southeastern
USA over a domain centered on Georgia. The system uses WRF (version 3.6) and
CMAQ (version 5.0.2) models for meteorology and air quality computations. In
addition, using the DDM-3D feature of CMAQ, it provides daily forecasts of the
potential contribution of prescribed fires to air quality. In HiRes-X, the burn activity
forecast coverage, formerly limited with Georgia, has been expanded to other states
in the Southeastern USA. A classification and regression tree (CART) was built using
meteorological data and either burn permit data in Georgia or satellite-based burn
location and area data in other states, for recent years. The satellites cannot always
detect the prescribed fires in Southeastern USA due to their small size and low heat
intensity, overlaying canopy in understory burns, and frequent cloud cover. On a
given day, if more than a threshold amount of land (e.g., 40 hectares), which varies
by state, was permitted or detected to burn in any county, that day is defined as a
“burn day”. The CART branches out at specific meteorological parameter values
with each branch leading to a “burn day” or a “no-burn day”.
If a burn day is forecasted, burns of sizes typical for lands owned by institutional,
large commercial and small private burners totaling the daily average burn area in
the region are randomly distributed to the respective lands. The Fuel Characteristic
Classification System (FCCS) maps are used to determine the fuel loads at each burn
location. Then, the fraction of the fuel that will be combusted is calculated using the
CONSUME model together with forecasted meteorological parameters such as fuel
moisture. Next, fire emissions are calculated by multiplying the amount of fuel mass
consumed with the emission factors (EFs) for different pollutants. The EFs used here
are specific to the fuels in the Southeastern USA and were derived from prior field
measurements. Finally, the plume height is calculated from the heat released and
the emissions are vertically distributed and injected into the corresponding layers of
CMAQ.
Other than the above forecasting products generated in-house, the system is also
fetching datasets from third parties. These datasets are ozone and PM 2.5 observations
from the national air quality monitoring network (https://www.airnow.gov) and fire
locations and burned areas from Hazard Mapping System (HMS) Fire and Smoke
Product (http://www.ospo.noaa.gov/Products/land/hms.html), both downloaded on
a daily basis, and burn locations and areas from Florida and Georgia’s burn permit
records, updated on an annual basis.
Air quality observational data are ingested into a MySQL database tables immediately after being downloaded onto the local server. HMS detection datasets are
transformed to a simple text format before archival. Permit data undergo quality
checks, during which missing items are filled, if possible. For example, latitude and
longitude of the burns, if not provided, are derived from provided addresses by utilizing Google Earth services. Finally, the checked records are ingested into the MySQL
database while inaccurate or incomplete records are discarded.