Caracterização espectral-temporal de florestas subtropicais integrando dados multiespectrais, fitossociologicos e geomorfologicos

Abstract

The use of remote sensing data for forest monitoring has advanced in recent years, with data acquired by different platforms. Remote sensing has the potential to provide information and solutions at different spatial and temporal scales and spectral resolutions, to characterize and monitor vegetation in space and time. The present work aims to evaluate the spectro-temporal characterization of Parque Estadual do Turvo, integrating multispectral data from satellites and UAVs, combined with geomorphological and phytosociological data. The study area is a Forest Reserve, with full protection, located in the Municipality of Derrubadas-RS, South Brazil. For this study, data from orbital and aerial imaging remote sensors were used as well, and meteorological data, field phytosociological data, and geomorphological data were obtained. The UAV sub-metric images were used as the field truth, due to the high spatial resolution. Using the UAV scenes the tree crowns were manually mapped, following the forest inventory. The multispectral images of the PlanetScope sensors were used for the evaluations through a time series of five years. For the time series tendency analysis, the NDVI and EVI vegetation indices were used. After the separate analyses, the data were integrated to provide more accurate data on the phytosociological dynamics of Parque Estadual do Turvo. The study area can be considered flat to undulating, with a maximum slope of 43.8º. The PET altimetry showed an amplitude of 339 meters, with a minimum elevation of 125 meters and a maximum elevation of 464 meters. A total of 118 crowns of native trees were vectorized, of 36 different species, distributed in the seven sample plots analyzed, from the sub-metric images of UAV. The withincrowns spectral analysis showed a directional effect dependence. North and East orientations tend to present higher reflectance rather than the South and West. As for the analysis of the spectral dynamics, two dynamic scenarios were identified based on the fit lines (Loess) applied to the average data of the vegetation indices per plot. In plot one, close to the entrance to the Park, a downward trend in vegetation indices from 2016 to 2021 was identified. In the other plots, there was a tendency to increase vegetation indices. The NDVI and EVI data for the canopies analyzed showed Intra and interannual differences considering the three species. This study demonstrated that the forest species present in the park are dependent on the summer and winter seasons, regardless of whether they are deciduous or evergreen. In this study, it was possible to verify that data integration of data is a proxy for the analysis of the dynamics of subtropical forests.