Close Window

An Econometric Approach to Amazon Landscapes and Assessing Their Hydroclimatological Impacts

Nathan Moore, Michigan State University, moorena@msu.edu (Presenting)
Eugenio Arima, Michigan State University, arimaeug@msu.edu
Robert Walker, Michigan State University, rwalker@msu.edu
Alex Pfaff, Columbia University, ap196@columbia.edu
Eustaquio Reis, IPEA, ejreis@ipea.gov.br
Marcellus Caldas, Michigan State University, caldasma@msu.edu
Claudio Bohrer, Universidade Federal Fluminense, bohrer@vm.uff.br
Juan Andrés Robalino, Columbia University, jar101@columbia.edu

This paper presents initial findings from the implementation of a linked land-climate model that can predict Amazonian climate as a function of possible development scenarios. The model uses two sub-components, an econometric land cover model capable of linking land cover changes to spatial and temporal variation in factors that drive human behavior, and a regional climate model (RCM) capable of assessing the effects of land cover change (and external climate forcing) on region-scale climate. As is well known, land cover in the Amazon Basin is undergoing rapid transformation due to agricultural expansion, logging, and road building effects. This transformation is accompanied by alterations in albedo, leaf area index, and fractional cover. Shifts in such biophysical characteristics affect the partitioning of energy in the surface energy budget, and therefore surface temperature, convective precipitation, and latent and sensible heat fluxes. The methodology implemented provides a way to causally link these various processes. It exploits the probabilistic nature of an econometric approach by generating multiple stochastic landscapes associated with possible development scenarios. By serving as inputs to the RCM, these landscapes then enable the prediction of probabilistic outcomes for climate, with a natural representation of uncertainty by the computational production probability distribution functions for key climate variables - e.g. precipitation, temperature, wind, etc. In the simulation presented, we used the econometric model to generate sets of land covers for three cases: (1) a best-case scenario, or BCS, with reduced forest removal due to minimal infrastructure investment associated with Avança Brasil, strong enforcement of protected areas, and high rates of out-migration from the region; a worst base scenario, or WCS, with high forest removal due to the completion of Avança Brasil projects, lax enforcement of protected areas, and low rates of out-migration; and (3) the bracketing case of complete forest removal, with replacement by agriculture. We adapted these land covers for use in RCM simulations, using the Regional Atmospheric Modeling System (RAMS) for 4 years, 1998-2001. For each year we simulated the impacts of each land cover scenario under a variety of atmospheric conditions. The results presented show that WCS simulations produce less rainfall within the basin than BCS simulations, and that these reductions in rainfall occur in close proximity to deforested areas. Total deforestation simulations yield lower domain-averaged rainfall particularly during the rainy season.

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

Session:  1A: Land Surface, Climate, and Hydrology

Presentation Type:  Oral

Abstract ID: 16

View presentation

Close Window