Coupling socioeconomic and demographic dimensions to a spatial simulation model of deforestation for the Brazilian Amazon
Soares-Filho, Universidade Federal de Minas Gerais, firstname.lastname@example.org
Garcia, Universidade Federal de Minas Gerais, email@example.com
Rodrigues, Universidade Federal de Minas Gerais, firstname.lastname@example.org
Moro, Universidade Federal de Minas Gerais, email@example.com
Nepstad, Woods Hole Research Center, firstname.lastname@example.org
The future of the Amazon forest is at a crossroads. At the same time that the State increasingly moves to curb unrestricted forest destruction, growing national and international agricultural markets encourages the advance of the deforestation frontier towards inner Amazon regions. To assess these opposing trends, we have developed a model that simulates future Amazon deforestation under a set of plausible scenarios, encompassing a range of socioeconomic and demographic contexts, conservation strategies, and public policies. In order to parameterize the simulation, we developed an econometric model that analyzes the influence of a series of socioeconomic and demographic variables - selected from 1996 and 2000 IBGE censuses and other economic and social surveys - on the recent deforestation trend. The model consists of a spatial lag regression that assesses the effect from changes in the socioeconomic context on the 1997-2001 deforestation rates of 399 counties. After spatial autocorrelation removal and heteroskedastic control, the model achieved R2 of 0.64. Proximity to paved roads, increase in cattle herd, cropland expansion, net migration rates, and percent of protected areas were the most important variables to explain the deforestation rates, with only the latter showing a negative effect. We employed the obtained spatial regression equation to project deforestation rates at the county level under scenarios of agricultural and economic growths, expansion of protected area network, and infrastructure improvement. Because this equation incorporates a neighborhood matrix, it can be used to infer the potential for future deforestation from changes in the socioeconomic context, not only within a specific Amazon county, but also from its neighboring counties. The projected rates are passed to a spatially explicit model that integrates the influence of a set of spatial determinants on the location of deforestation. Yearly deforestation rates for the entire Brazilian Amazon from 2002 to 2006 were used to validate the model, which showed a maximum deviation of only 10%. As a result, modeled scenarios show that a large expansion of the protected area network has an immediate effect on lowering regional deforestation rates. However, the effect of this measure alone tends to weaken over time, becoming even less important in a scenario of rapid growth of the national agricultural sector.
Science Theme: LC (Land Use and Land Cover Change)