Caracterização e mapeamento biofísico ambiental da bacia hidrográfica do rio vermelho a partir de dados remotamente sensoriados

• Autor(a) principal: Santos, Pablo Santana
• Data de Publicação: 2014
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Repositório Institucional da UFG
• dARK ID: ark:/38995/0013000006w3f
• Texto Completo: http://repositorio.bc.ufg.br/tede/handle/tede/4084
• topic: Sensoriamento remoto; Pastagem; Análise biofísica; Remote sensing; Pasture; Biophysical analysis; CONSERVACAO DA NATUREZA::CONSERVACAO DE BACIAS HIDROGRAFICAS

The agricultural expansion in the Brazilian Cerrado represents nowadays more than half of the total biome area, where conversion of the natural vegetation to pasturelands is the main dynamics imposed by the productive sector along the relatively recent historical occupation process. Such transformations tend to create distinct spatial patterns regarding the biophysical behavior of the main existent vegetation formations, which can be identified and discriminated through a variety of available remote sensing products. This study, focused on the rio Vermelho watershed, a representative unit of the Cerrado biome, and considering a variety of satellite data for the 2001 – 2011 period, pursued a detailed analysis of the biophysical responses in face of the historical processes that determined and shaped the landscape transformations. Specifically, the following datasets were utilized: MODIS (vegetation index, evapotranspiration and primary productivity) and TRMM (Tropical Rainfall Measurement Mission) precipitation. Our results indicated that climatic factors, such as seasonality, have a direct influence on biomass levels of key vegetation formations, particularly those with shallow root systems, for which reductions in radiometric values, as high as 50%, could be observed during the dry season. In fact, natural vegetation areas, comparatively to areas dominated by cultivated pastures, show very distinct and significant biophysical behavior (p < 0.05). Likewise, two major landscapes could be identified, i.e. “Region 1” and “Region 2”, where converted and remnant vegetation areas predominate, respectively. The distribution of natural and cultivated pastures seems to be related to local edafoclimatic factors, such as soil (depth/lithic content) and water availability, which explain the occurrence of most grass species. Based on field biomass, it is possible to say: 1) cultivated pastures found in “Region 1” are slightly more productive comparatily to those from “Region 2” (for both dry- and wet-seasons) and 2) there is a correspondence between “natural” and “cultivated” green biomass during the dry season, as opposed to the wet season. However, and based on the seasonal moderate spatial resolution MODIS images, it is not possible to separate between natural and cultivated pastures. In fact, such separation can be accomplished via the supervised classification (Support Vector Machine) of medium spatial resolution imagery (Landsat). Based on high spatial resolution (5m) RapidEye/BlackBridge images, it was produced a land cover / land use map for the entire basin at the 1:50.000 scale, in which it was possible to discriminate the major converted and natural physiognomies, and distinct grass species, in particular (Overall Accuracy ~ 90% and Kappa = 0.83).