Close Window

Projecting Future Amazonian Landscapes: An Econometric Approach

Robert Walker, Michigan State University, rwalker@msu.edu (Presenting)
Alex Pfaff, Duke University, ap196@columbia.edu
Stephen Perz, University of Florida, sperz@ufl.edu
Eustaquio Reis, IPEA, ejreis@ipea.gov.br
Marcellus Caldas, Michigan State University, caldasma@msu.edu
Eugenio Arima, Hobart and Smith Colleges, arimaeug@msu.edu
Claudio Bohrer, Universidade Federal Fluminense, bohrer@vm.uff.br
Juan Robalino, Columbia University, jar101@columbia.edu
Stephen Aldrich, Michigan State University, aldric30@msu.edu

This presentation gives details of a methodology for projecting future Amazonian landscapes, based an econometric estimation of the factors associated with deforestation. Regression models were fit using an extensive data base with information on natural resources, including rainfall and soils, the spatial distribution of human populations, the historical development of the transportation system, and time series of deforestation for three intervals (1976-87, 1987-92, and 1992-2000). Taking deforestation at census tract level over the entire basin as the dependent variable, equations were estimated to identify relationships between magnitude of forest loss and variables in the data base. These equations also controlled for so-called, fixed spatial effects at municipal level, since the actual observations implemented consisted of census tracts, for which there are multiple cases per municipio. The use of census tract observations allowed for a large number of statistical observations which, together with the fixed effects specification, enabled an estimation achieving high explanatory power. To undertake the projections, two possible development scenarios were defined based on infrastructure developments, possible trends in population growth, and status of protected areas. The scenarios were defined in terms of the variables used in the estimation of the econometric model, so that projections could be made by using anticipated variable values associated with the two scenarios. One scenario, referred to as the worst case scenario, assumed an early completion of Avanša Brasil projects, continued population growth with little out-migration, and negligible enforcement of protected areas. The other scenario, the best case, assumed no more infrastructure investments in the region, reduced population growth with increased out-migration, and effective protection of the areas so designated. The presentation gives results for both cases out to 2050, and contrasts the impacts of the two development scenarios. In addition, it gives an update on the status of Avanša Brasil projects in the Amazon region.

Science Theme:  LC (Land Use and Land Cover Change)

Session:  3B: Modeling LCLUC

Presentation Type:  Oral

Abstract ID: 3

View presentation

Close Window