Uso de imagens do satélite ALOS para estimativa de parâmetros dendrométricos de plantios de eucalipto

• Autor(a) principal: Oliveira, Fernando Soares de
• Data de Publicação: 2011
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: LOCUS Repositório Institucional da UFV
• Texto Completo: http://locus.ufv.br/handle/123456789/580
• topic: Radar; Reflorestamento; Geoprocessamento; Prism; AVNIR; Radar; Reforestation; Geoprocessing; Prism; AVNIR; CNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL::MANEJO FLORESTAL

The general objective of the present study was to develop a methodology utilizing images from satellite ALOS (AVNIR, PRISM and PALSAR) to estimate dendrometric parameters of Eucalyptus grandis. plantations. The study was conducted in the municipalities of Ipaba, Belo Oriente, Santana do Paraíso and Caratinga, all within the state of Minas Gerais, Brazil, in projects belonging to the company CENIBRA S/A. The methodological development was performed at fist with the acquisition of data from the forest inventory, given by the company, which contained information on the average height, DBH, basal area, age, volume and biomass, as well as the coordinates location of the plots. Subsequently, adjustment and selection of equations were carried out to estimate the dendrometric parameters (dependent variables) in function of variables generated from satellite images (independent variables). Variables derived from sensor AVNIR were bands 1, 2, 3 and 4, Band Ratios, NDVI and SAVI. Sensor PRISM provided a panchromatic image as variable, while those generated by sensor PALSAR were polarizations LHH, LHV, LVV and LVH, incoherent attributes Parallel Ratio, Cross-Ratio and Total Power and the vegetation indices VSI, CSI and BMI. With the application of the Stepwise regression model, the equations that best explained the parameters in the populations of Eucalyptus grandis. were selected. The main results found were: the estimates of variables volume and biomass from the forest inventory were those which presented the highest correlations with the radargrammetric and optical data, with coefficient of variation ranging from 69.47 to 71.05%, indicating a great potentiality of the utilization of these data as support in inventories of large areas. For variables Age and DBH, the correlations with the data originated from the orbital sensors were very low, so they were impracticable to estimate these variables. It was not possible to generate models to estimate the number of stems. Sensor PRISM supplied data which were not capable to estimate the parameters studied individually; however, analyzing the models generated by the integration of all the sensors, it participates in the estimation of age, basal area, volume and biomass. It is worth stressing the importance of utilizing techniques of remote sensing and SIG at the estimations, once they enable the performance of quantifications by utilizing non-destructive methods, and also boost the spatialization of information, so they are useful tools for mapping and monitoring the cover and use of the land.