COMPARISON OF CLASSIFICATION METHODS FOR THE MAPPING OF LAND COVERAGE IN THE NORTHERN SECTOR OF MARANHÃO ISLAND
Classification methods comparison for the mapping of the land cover in the North sector of the Maranhão Island
DOI:
https://doi.org/10.5016/geociencias.v39i04.14128Abstract
The coastal extends from the coastal plains to the coastal plains to the outer rim of the continental shelves. The GEOBIA (Geographic Object-Based Image Analysis) is an effective method for analyzing high-resolution images, including photointerpretation and classification of landscape features. Thus, the present study has as main objective to compare the method of pixel to pixel supervised classification with the GEOBIA method for the mapping of the land cover in the North sector of the Island of Maranhão. The supervised classification method consists of prior knowledge of some areas in which one wishes to work, which allows for reliable training samples. The classifier algorithm used was the Maximum Likelihood Classification. The Supervised Classification method showed a Kappa value of 0.8469, Global Accuracy of 0.8950 and Kappa Variance of 0.0009295. For GEOBIA it presented a Kappa value of 0.9410, Global Accuracy of 0.9600 and Kappa Variance of 0.0004075. Remote sensing proves extremely efficient in land use and occupation surveys. The results of the Supervised Classification method, despite their limitations, showed satisfactory results when compared to the results of the GEOBIA method, which makes interesting its application for mapping initiatives in several Brazilian municipalities.