Comparação entre metodologias para mapeamento da arborização de vias públicas do bairro Chapada, Ponta Grossa-PR, com uso de imagens plêiades

Knowing about the arboreal heritage present along the city’s roads is a challenge. Acquiring this information requires personal experience, time, and investing financial resources in the field. It is not always possible to carry out counts in situ, so different techniques are implemented. Pixel-oriented methodologies (supervised, unsupervised classification) are used, along with NDVI segmented classification to gain prior knowledge of the roadside tree heritage, and to map and quantify them. The Chapada neighbourhood in the city of Ponta Grossa-PR (Brasil) has been previously mapped using the visual analysis methodology. The count of 3101 trees on 228 roads was used as a reference to compare with the three aforementioned methodologies. A soil classification was carried out using the three methodologies to find out about the presence of roadside vegetation. In the case of NDVI, it was possible to show 56.00 % similarity with the trees obtained in the visual analysis; the unsupervised classification obtained a map of 91.19 %, this being the largest number of trees counted, and the supervised classification figure was 82.68 %.