# Suppression

ELMFIRE includes experimental initial attack and extended attack
suppression models. These models, both disabled by default, can be
configured from the `&SUPPRESSION`

namelist group.

## Initial Attack

The approach used here to quantify initial attack probability of
containment is based on the analysis of `Hirsch et al.
<https://academic.oup.com/forestscience/article/44/4/539/4627517`_ who
leveraged expert judgment to quantify initial attack effectiveness as a
function of fire size and head fire fireline intensity, *i.e.* intensity
at the main advancing fire front, at the time of initial attack
commencement. The authors developed an expression for probability of
containment (*POC*) as a function of fire size (*A*) and fireline
intensity (*I*) which is given in the equation below:

In the equation above, \({A}\) is in hectares and \({I}\) is in
kW/m. Since trends in probability of containment are not immediately
apparent upon inspection of the equation, probability of containment
calculated from the equation is tabulated in the following table as a
function of fire size and head fire fireline intensity at the time of
initial attack. Although the qualitative trends in the table are
logical, *i.e.* containment probability increases with smaller fires,
lower intensity, or both, the `Hirsch et al.
<https://academic.oup.com/forestscience/article/44/4/539/4627517`_ study
was based on expert opinion from Canadian firefighters so differences in
suppression tactics between Canadian and U.S. agencies are not reflected
in the table.

In order to use the equation, the time of initial attack commencement relative to the time of fire ignition (\({t_{ign}}\)) must be established for each ignition location. This can be viewed as the sum of fire detection time (\({t_d}\)), report time (\({t_r}\)), and travel time (\({t_t}\)):

Initial attack suppression modeling, disabled by default, can be enabled
by setting `ENABLE_INITIAL_ATTACK=.TRUE.`

. The time of initial attack
commencement, in seconds, must also be specified using the
`INITIAL_ATTACK_TIME`

keyword.

## Extended Attack

Change in containment (\({\Delta C}\)) during time interval \({\Delta t}\) (in days) is:

where \({Chi}\) is suppression effectiveness coefficient (-), \({A_d}\) is areal growth on a daily basis (acres/day), and \({A_{d0}}\) is areal growth on a daily basis at which containment change becomes negative (acres/day).

The function \({f(SDI_{\Delta t})}\) is calculated from the mean suppression difficulty index during interval \({\Delta t (\bar{SDI_{\Delta t}})}\) as:

where \({B}\) is the calibration constant, \({i}\) is the pixel index, \({n}\) is the number of pixels during time interval \({\Delta t}\), and \({SDI_i}\) is the Suppression Difficulty Index of pixel i.

The extended attack model can ben enabled by setting
`ENABLE_EXTENDED_ATTACK=.TRUE.`

. Additional parameters that can be
used to fine tune the extended attack model are:

```
B_SDI = 1.0
DT_EXTENDED_ATTACK = 3600.
AREA_NO_CONTAINMENT_CHANGE = 10000.0
MAX_CONTAINMENT_PER_DAY = 100.0
SDI_FACTOR = 1.0
USE_SDI = .FALSE.
USE_SDI_LOG_FUNCTION = .FALSE.
```