Full text
| |
| Author(s): |
Felipe Coelho de Souza Petean
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Piracicaba. |
| Institution: | Universidade de São Paulo (USP). Escola Superior de Agricultura Luiz de Queiroz (ESALA/BC) |
| Defense date: | 2015-08-28 |
| Examining board members: |
Demóstenes Ferreira da Silva Filho;
Andreia Medinilha Pancher;
Luiz Carlos Estraviz Rodriguez
|
| Advisor: | Demóstenes Ferreira da Silva Filho |
| Abstract | |
Urban forestry is a key element to maintaining the quality of life in urban centers. The existence of a broad network of trees distributed along roads and public spaces acts to promote air quality, water conservation, thermal comfort, acoustic and psychological citizens. Urban forests are able to mitigate the emissions of Greenhouse Gases (GHG) such as CO2, acting as sinks. Since its importance, new applications of remote sensing tools have emerged to assist in planning and implementation of urban forestry. The laser scanning system airborne LiDAR (Light Detection And Ranging), generates a three-dimensional representation of the target object through a cloud of points georeferenced. The crossing with very high resolution sensors provides more in-depth analysis of the object and can be extracted several forest metrics such as height, basal area, and even species. The study aimed to verify the contribution of LiDAR derived points in the identification and classification of six most common tree species in Parque Municipal Americo Renne Giannetti, Belo Horizonte, Minas Gerais, Brazil, in order to assist urban forestry planning and management. Through supervised classification, field survey information, segmented areas, LiDAR treetop points, and a multispectral WordlView-2 image were crossed together. The classification accuracy was measured by analyzing overall accuracy and Kappa index. The LiDAR treetop points contributed to location and classification of tree species\' classes, when compared to the same process without these points. The segmentation of crowns performed by eCognition program facilitated the launch of training and test samples. ECHO classifier showed the best accuracy and Kappa index in comparison to other Multispec program classifiers. The aggregation of LiDAR data showed promise in urban forest multispectral images, increasing supervised classification overall accuracy. (AU) | |
| FAPESP's process: | 13/21338-0 - Use of hyperspectral images and LiDAR technology in the identification of tree species in urban environment in the city of Belo Horizonte, Minas Gerais |
| Grantee: | Felipe Coelho de Souza Petean |
| Support Opportunities: | Scholarships in Brazil - Master |