Modeling the disturbance of vegetation by fire in the boreal forest
Cyril
Crevoisier, AOS, Princeton University, ccrevois@princeton.edu
(Presenting)
Elena
Shevliakova, EEB, Princeton University, elena@princeton.edu
Manuel
Gloor, University of Leeds, E.Gloor@leeds.ac.uk
Christian
Wirth, MPI-Jena, cwirth@bgc-jena.mpg.de
Boreal regions are important for the global carbon cycle because it is the largest forested area on earth and there are large belowground carbon pools (~1000 PgC). It is also a region where largest warming trends on the globe over the last decades have been observed and changes of the land ecosystems have already started. A major factor that determines the structure and carbon dynamics of the boreal forest is fire. As fire frequency depends strongly on climate, increased fire occurrence and related losses to the atmosphere are likely, and have already been reported. In order to predict with more confidence the occurrence and effect of fire on forest ecosystems in the boreal region, we are developing a fire model that takes advantage of the large on-ground, remote sensing and climate data from Canada, Alaska and Siberia.
We have designed a prognostic model to estimate the monthly burned area in a grid cell of 2 by 2.5 degrees, from four climate (air temperature, air relative humidity, precipitation and soil water content) and one human-related (road density) variables. Parameters are estimated using a Markov Chain Monte Carlo method applied to a dataset of observed burned area for Canada.
The model is able to reproduce the seasonality of fire as well as the location of fire events, and to predict the large fire events that have occurred in the last two decades, for both Canada (on which data the model has been designed) and Siberia. The results also compare well with remote sensing observation. The fire model will be implemented in LM3, the new vegetation model of GFDL, in order to make prediction of future fire behavior in boreal regions, and the related disturbance of the vegetation and carbon emissions.