Análise de tipologias florestais por meio da resposta espectral de uma imagem hiperespectral (Hyperion/eo-1) no município de Manaus, Reserva Ducke

Abstract

Hyperspectral remote sensing allows for radiometric measurements of a target in a large number of narrow spectral bands. The data collected by these sensors can be transformed into information about different vegetation covers that are related to biophysical aspects of vegetation. Vegetation being an important element of ecosystems, their studies gain emphasis, especially for seeking knowledge about its variations, distributional patterns, cycles, physiological and morphological changes. With this data was used Hyperion that enables the acquisition of data sufficient to reconstruct absorption bands in the spectra of the pixels associated with the chlorophyll content, water content in leaves and features of lignin and cellulose spectral resolution, which may be important parameters in differentiating vegetation types . On the other hand the poor signal to noise ratio ( SNR) of the sensor, especially in the SWIR, relationship is an obstacle to proper measurement of these features without interference from noise. This research aimed to conduct an analysis of the spectral response of the sensor with the Hyperion plant communities present in the Ducke Reserve (Forest plateau, slope, and lowland campinarana) and quantified by means of unsupervised classification of these communities. Using ENVI Flaash application, atmospheric correction was performed based on the radiative transfer model MODTRAN - 4. Floristic characterization of the study area was raised regarding esttudos related flora of the Reserve. It was possible to perform a summary analysis of the spectral characteristics of the sensor as present an interaction with vegetation between the near infrared (0,7 - 2,5μm) and middle infrared (3 - 6μm), where we can distinguish stage of phenology, canopy structure and water content on leaf responses in four plant communities.