Influência dos dados e métodos no mapeamento do uso e da cobertura da terra

Abstract

The purpose of this study is to analyze the influence of different data and classification methods on land-use and land cover mapping, according to the following objectives: 1) to quantify land cover and land-use change from 2011 to 2014 and 2) to map and compare land-use and land cover in between data from different orbital platforms. The chosen study area was Joao Leite Watershed (762 km²), which has both great socioeconomic and environmental importance to the Goiania Metropolitan Area (Goias State, Brazil). The database was composed by satellite images from the sensors of Landsat-8 OLI for 2014 (30 meter spatial resolution) and Theos for 2011 (2 meter spatial resolution), as well as a land-use and land cover vector file for 2011 (Theos/2011). The mappings of Landsat/2014 and Theos/Atualizado-2014 were originated from image classification, visual interpretation and vector editing procedures. From the period of 2011-2014 there was a change of 18.2 km² in Joao Leite Watershed, primarily related to agricultural and forestry expansion into pastureland, and pasture expansion into dense vegetation. Regarding the Theos/Atualizado-2014 mapping, Landsat/2014 overestimated the dense vegetation class (by 58.7 km²) and underestimated the pasture (by 50.4 km²), agriculture (by 6.4 km²) and forestry classes (by 3.8 km²). Besides the variation from data choices, inappropriate use of image classification methods and procedures may also influence the results, which can generate incorrect information maps.