Fire
Fire plays a key role in both vegetation dynamics and the global carbon cycle. The FIRE module (i.e. the Glob-FIRM module described in Thonicke et al. in 2001) derives fractional area burnt in a grid cell from the simulated length of fire season and minimal fuel load annually.
It does not specify ignition sources (it does not distinguish between lightnings- and human-caused- fires), not considering the population density effect (as SPITFIRE does). This fire module can be applied, where ignition sources cannot be reliably quantified. It is ideal for paleo-climate applications, for example.
Fire is modelled as a combination of fire occurrence and its effects. The length of the fire season is calculated as the annual sum of the daily fire probabilities, based on the dead-fuel moisture content (taken as the daily moisture status in the upper soil layer). Fire size and spread are related to temporal burning conditions, and therefore the length of fire season can be translated into an area burnt (expressed as a fraction of the grid cell).
The influence of vegetation is modelled through two PFT-parameters: moisture of extinction and fire resistance. While for the first field measurements can be found, for the latter measurements are difficult to find as the parameter is a combination of fire intensity and fire severity.
1.Fire occurrence is taken to be dependent only upon fuel load (acting as a threshold of 200 gC/m²) and litter moisture (i.e. the amount of dry combustible material available). This concept combines both the influence of climate and vegetation but ignition sources are implicitly assumed to be present whenever there is a risk of fire.
2.Plant (or PFT) influence on fire ignition is depending on the moisture of extinction. This is a threshold value of fuel moisture content which is defined as the fuel moisture content above which a fire would not spread. An exponential function is used to approximate the probability of the occurrence of at least one fire in a day:
where w{l,d} is the daily moisture status in the upper soil layer, used as a surrogate for litter moisture, and me is the threshold fuel moisture. The length of the fire season is estimated by adding these probabilities over the whole year:
It is assumed that a given length of fire season results in the same annual area burnt, irrespective of the dry-day distribution. By fitting a function to observed data (Thonicke et al., 2001) related annual fractional area burnt A to fire season s:
This formulation is based on the hypothesis that the annual fractional area burnt increases at first slowly, when the length of the fire season s is relatively short, but increases more rapidly, when s approaches the entire year.
3.Fire effects are assumed to be driven by the length of the fire season and PFT-dependent plant resistances only. The fraction of individuals killed within the fractional area burnt depends upon the prescribed, PFT-specific fire resistance. Tropical raingreen and temperate broadleaved evergreen woody PFTs have high resistances, reflecting widespread adaptations to fire conditions seen in plants from dry subtropical and tropical environments. This situation contrasts with the typical fire response of trees in the boreal zone, where “stand replacing” fires are the norm and the trees are adapted for efficient post-fire regeneration, rather than fire resistance. Grass species with their meristems below-ground are able to survive fire. Herbaceous leaf foliage, above-ground litter, and dead woody tissue are fully consumed in the fractional area burnt, and released to the atmosphere as CO2.
When the SPITFIRE fire module is not activated, the former fire module, called FIRE, is working (when fire disturbance is enabled).
To run the model with the FIRE module the flag SPITFIRE is
turned off (#define NO_SPITFIRE) and the FIRE module is enabled:
…
#ifdef WITH_SPITFIRE
SPITFIRE
#else
FIRE
…
For src/soil/fire_sum.c have a look at the function src/lpj/update_daily.c look for setting.fire==FIRE.
For src/soil/fire.c have a look at the function src/lpj/annual_natural.c look for setting.fire==FIRE.
Running LPJmL with FIRE the following outputs can be obtained:
Carbon released by fires (firec) in gC/m²
Fire Danger Index (firedi) ranging from 0 to 1
Number of fires (nfire)
Burnt area (burntarea) in ha
Fire return interval (firef) as 1/fraction of grid cell burned
Check in source:trunk/par/outputvars.par par/outputvars.par all the possible outputs.
Sitch S., 2000. The role of vegetation dynamics in the control of atmospheric CO2 content. Doctoral Dissertation. Institute of Ecology , Lund University, Sweden.
Thonicke K., Venevsky S., Sitch S. and Cramer W., 2001. The role of fire disturbance for global vegetation dynamics: coupling fire into a Dynamic Global Vegetation Model. Global Ecology & Biogeography 10, 661-677.