Uso de imagens de alta resolução, características espectrais e classificação orientada ao objeto no mapeamento de vegetação de manguezal

Abstract

The use of remote sensing is one of the most efficient techniques for evaluating biomass and distribution of mangroves. However, there still’s a big problem for identifying the species composition within a same vegetal unit. This difficulty is linked to the mixture of spectral patterns, resulting from the coexistence of a diverse number of vegetal species and therefore similar spectral responses in the same unit of area. Working towards the individualization and quantification of the spatial occupancy of Rhizophora and Avicennia mangrove trees in the peninsula of Bragança, this study analyzed the subtle differences in the mangrove reflectance levels from a set of pan sharpened IKONOS imagery and field data collection of the reflectance signature obtained directly on leaves of the mangrove vegetation. Object oriented classification was used, within a combined approach of samples and knowledge (fuzzy logic) classification, supported by the following image properties: mean of the spectral bands, normalized difference vegetation index (NDVI), soil brightness index (SBI) and brightness index (BI). Thus, on September 22 of 2003, date of image acquisition, the mangrove forests occupied an area of approximately 141,39 km2 of the Bragança Peninsula and is represented mainly by Rhizophora (104.91 ± 1,59 km2), Avicennia (18.76 ± 3,61 km2), and shaded mangrove vegetation (17.72 km2). The classes Terra Firme, Water and Others occupy 5.85, 49.09 and 96.55 km2, respectively. The results obtained allow a better location and quantification of mangrove species in the peninsula of Bragança-PA, individualization of Avicennia and Rhizophora trees, confirms the use of orbital sensors of high spatial resolution as appropriate for the effective mapping of coastal vegetation and enables the subsidize of future studies on relative density of mangrove forests in a regional scale, estimates of biomass, development dynamics of mangrove forests, environmental analysis and resilience of mangrove forests the Bragança Peninsula.