Fragment of native forest using remote sensing

Author(s)

André Gustavo Nave

• Document type: Master's Dissertation
• Press: Piracicaba.
• Institution: Universidade de São Paulo (USP). Escola Superior de Agricultura Luiz de Queiroz (ESALA/BC)
• Defense date: 1999-09-03
• Advisor: Ricardo Ribeiro Rodrigues

Abstract

Nowadays, there is a great necessity of developing work methodologies that allow the identification and checking of the forest mosaic. The analyses made by aerial photographs or orbital images can make possible, with low cost, the identification, quantification and evolution studies of the ecological units in the forest mosaic, allowing adaptation of the characterization proposals and management of the forest fragments. The objective of this work was to test some methodologies for the zoning and the identification of the different forest ecological units (natural and sucessional) of Santa Genebras Municipal Reserve in Campinas, São Paulo. Methods of interpretation of aerial photographs and processing of Landsat TM and SPOT XS orbital images were used. The confirmation of the existence of the ecological units identified in the image analyses was obtained in field expedite survey, by physiognomic, floristic and structure characterization of those units. For the evaluation of the similarity among the communities of each identified ecological unit, similarity analyses based on the data obtained in the floristic and structural characterization of each unit were carried out, confirming the vegetacional particularities of these units. The comparison of the maps produced by the analyses of the aerial photographs and orbital images was made according to the number of identified ecological units, detailed ground truth, time spent in image analysis, necessary technical resources, image availability, image temporal resolution, use frequency in vegetation works and acquisition cost, among others. The accuracy of the produced maps was obtained generating error matrices, calculating the global and individual accuracy (commission and omission errors), medium performance and the global and individual Kappa concordance coefficients. The results of this work showed that the analogical method of analysis of aerial photographs using stereoscopy was more appropriate and that the identification of the ecological units and that the larger was the scale used, better was the level of map detailing. ln the analyses of orbital images, the unsupervised classification, that it is one of the easiest techniques in terms of image processing, obtained the best result in the identification of the forest mosaic, mainly the variation of the sucessional gradient. The results obtained by the textural classifiers in the classifications in aerial picture and in orbital images were unsatisfactory in the recognition of the units. The use of the grouping and ordering analyses were efficient in the determination of the similarity among the studied communities, confirming the existence of the units in the field. The analysis of images of different ages was an efficient low cost instrument of environmental monitoring, subsidizing the use of differentiated practices of conservation and management of forest remainders. The identification of the environmental mosaic also provides a better definition of sampling areas in fragments. (AU)