A spatially explicit simulation model of deforestation for the Amazon Basin
Britaldo
Silveira
Soares Filho, Universidade Federal de Minas Gerais, britaldo@csr.ufmg.br
(Presenting)
Daniel
Curtis
Nepstad, Woods Hole Research Center, dnepstad@whrc.org
Gustavo
Coutinho
Cerqueira, Universidade Federal de Minas Gerais, cerca@ufmg.br
Ane
Alencar, Instituto de Pesquisa Ambiental da Amazônia, ane@ipam.org.br
Eliane
Voll, Universidade Federal de Minas Gerais, voll@ufmg.br
Paul
A.
Lefebvre, Woods Hole Research Center, paul@whrc.org
We present a model to simulate deforestation through the entire Amazon Basin. Its architecture embodies coupled models structured in two spatial levels: 1) subregions defined from socioeconomic regionalization and 2) raster cells. An upper model projects the deforestation rates for the subregions, based on data provided mainly by PRODES, and passes them to a spatially explicit simulation model. Each subregion has a unique spatial model with its own set of parameters. Cartographic algebra and cellular automata technique are used in order to develop the spatially explicit simulation that comprises a map of 3144x4238 cells at 1 km2 resolution. The model uses cartographic data on infrastructure, administrative units, and biophysical settings and was spatially calibrated for 12 case study areas in the Brazilian Amazon. We ran the model for two scenarios. The “business as usual” scenario considers the deforestation trends across the basin, projecting 1997-2000’s rates linearly by using their variations between 2000 and 2001, and adding to them the effect of paving a set of major roads. In this scenario the paving follows a predefined schedule and its effect on deforestation is empirically estimated using PRODES data analyzed at municipality level. The other scenario also considers the current deforestation trends, but now projecting them logarithmically over time. In this so called “governance scenario”, it is imposed a 50% limit for deforested land within each basin’s subregion and existing and proposed protected areas play a decisive role in hindering deforestation as well. Both simulations encompass 50 annual time steps starting at 2001. Results for the “business-as-usual scenario” show 23% and 37% of total forest decline within 30 years and 50 years, respectively. This expected deforestation is reduced up to 55% in the governance scenario. The simulated spatial patterns of deforestation can be used as inputs to Global Climate Models.
Submetido por Britaldo Soares Filho em 18-MAR-2004
Tema Científico do LBA: LC (Mudanças dos Usos da Terra e da Vegetação)