Mapeamento da arborização de calçadas nas vias públicas de Ponta Grossa - PR com uso de sensoriamento remoto

Abstract

Urban afforestation on city sidewalks is associated, worldwide, with the quality of life of the inhabitants, as it is the main factor that helps to break the concrete landscape, in addition to contributing to air circulation, regulation of the microclimate and encouraging leisure. However, the planting of trees on the sidewalks must be carried out in compliance with certain rules established by the control bodies. For this reason, it is important to know the distribution and state of the trees in relation to the urban environment, with the inventory being the most used tool to know the state of the trees. One of the ways to carry out inventories is using tree censuses, in which quantitative and qualitative data can be included. One of the ways to carry out the census is through work in the field, which requires professionals with specific knowledge and has a high cost, especially for cities with large extensions. Alternatives are being sought to optimize this process. Within this context, this research carried out a mapping and quantification of trees present on the sidewalks of Chapada, Boa Vista, Oficinas and Central Area, in the urban area of Ponta Grossa - PR. Three methodologies oriented to pixel supervised classification, unsupervised classification and classification by segmentation of the NDVI were used using the Pléiades image (2013), with spatial resolution of 2m in the multispectral bands and 0.50cm in the panchromatic band. The tree mapping obtained by the visual analysis methodology, provided by the Laboratory of Socioenvironmental Studies-LAESA of the State University of Ponta Grossa, was used as a reference for points of truth. Compared the trees mapped with the three methodologies, having as reference the mapping made with the methodology of visual analysis, the hit rate for Oficinas was 97.78%, Boa Vista 94.74%, Central Area 81.01% and Chapada 76.02%. Making it clear that the NDVI methodology is considered in the comparison, although the conditions for its spatial resolution is 2m.