Estimating Amazonian forest fire probability based on edaphic and climatic factors and proximity to land-use and infrastructure.
Paul
A.
Lefebvre, Woods Hole Research Center, paul@whrc.org
(Presenting)
Daniel
Curtis
Nepstad, Woods Hole Research Center, dnepstad@whrc.org
Ane
A.
Alencar, Instutudo de Pesquisa Ambiental da Amazônia, ane@ipam.org.br
We have developed a spatial model of forest fire probability in the Amazon basin by combining our prior work in estimating forest flammability, with a spatial analysis of ignition likelihood. Recent fire scars were marked by analysis of satellite images of three locations in the eastern and southern Amazon basin, in the zone of intense landuse transition where human activities frequently include the use of fire to manage and transform the landscape. We tabulated the proximity of the fire scars to several landcover types as well as key infrastructure components. The more common indicators derived from this analysis were then used as predictors of ignition probability.
Forest flammability was estimated from our RisQue soil moisture balance model. This model uses edaphic factors together with climatic information (Pennman-Monteith PET and actual precipitation) to track soil moisture to 10m depth at monthly intervals. Our fieldwork has shown that forest flammability is linked to LAI, and LAI is linked to soil moisture variations. We calculated minimum plant-available soil water (PAW) during a 1-year period prior to the date of the imagery of our fire scars, to determine what range of soil moisture had been present on these sites when these fires occurred. By combining proximity to a number of land use types with soil moisture parameters from actual fires, we generate a probability surface for fire for the Amazonian landscape on a seasonal basis.
Submetido por Paul A. Lefebvre em 18-MAR-2004
Tema Científico do LBA: LC (Mudanças dos Usos da Terra e da Vegetação)