Improving the representation of fire disturbance in dynamic vegetation models by assimilating satellite data: a case study over the Arctic
Fire provides an impulsive and stochastic pathway for carbon from the terrestrial biosphere to enter the atmosphere. Despite fire emissions being of similar magnitude to net ecosystem exchange in many biomes, even the most complex dynamic vegetation models (DVMs) embedded in general circulation mode...
| Published in: | Geoscientific Model Development |
|---|---|
| Main Authors: | , , |
| Format: | Article in Journal/Newspaper |
| Language: | English |
| Published: |
Copernicus Publications
2015
|
| Subjects: | |
| Online Access: | https://doi.org/10.5194/gmd-8-2597-2015 https://doaj.org/article/db014cd27d014cd59b31d43842d8111b |
| Summary: | Fire provides an impulsive and stochastic pathway for carbon from the terrestrial biosphere to enter the atmosphere. Despite fire emissions being of similar magnitude to net ecosystem exchange in many biomes, even the most complex dynamic vegetation models (DVMs) embedded in general circulation models contain poor representations of fire behaviour and dynamics, such as propagation and distribution of fire sizes. A model-independent methodology is developed which addresses this issue. Its focus is on the Arctic where fire is linked to permafrost dynamics and on occasion can release great amounts of carbon from carbon-rich organic soils. Connected-component labelling is used to identify individual fire events across Canada and Russia from daily, low-resolution burned area satellite products, and the obtained fire size probability distributions are validated against historical data. This allows the creation of a fire database holding information on area burned and temporal evolution of fires in space and time. A method of assimilating the statistical distribution of fire area into a DVM whilst maintaining its fire return interval is then described. The algorithm imposes a regional scale spatially dependent fire regime on a sub-scale spatially independent model; the fire regime is described by large-scale statistical distributions of fire intensity and spatial extent, and the temporal dynamics (fire return intervals) are determined locally. This permits DVMs to estimate many aspects of post-fire dynamics that cannot occur under their current representations of fire, as is illustrated by considering the modelled evolution of land cover, biomass and net ecosystem exchange after a fire. |
|---|