Modelagem de variáveis biofísicas em floresta tropical seca por meio de geotecnologias

• Autor(a) principal: MOREIRA, Giselle Lemos
• Data de Publicação: 2021
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UFRPE
• Texto Completo: http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8939
• topic: Sensoriamento remoto; Floresta tropical seca; Caatinga; Biomassa; Volume; Inventário florestal; CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL

The forest inventory is an important instrument used to estimate the productive potential of a forest area, however its conventional techniques require time and, in some cases, present difficulties related to access or the size of the area. Thus, the objective of this study was to evaluate the capacity of LiDAR (Light Detection and Ranging) technology and of different orbital sensors in the prediction of biophysical variables volume and dry aboveground biomass (dry AGB) in areas of shrub-tree vegetation in dry tropical forest. To this end, the present work was divided into two chapters with specific objectives. In the first chapter assessed the potential of the metrics derived from the LiDAR ALS (Airborne Laser Scanner) system in predicting volume and dry AGB in dry tropical forest areas. The study was conducted in two areas of Caatinga at Fazenda Itapemirim, located between the municipalities of Floresta and Betânia, in Pernambuco. To assess the potential of LiDAR technology, the multiple linear regression model was adjusted using the ordinary least squares method, where volume and dry AGB stocks, per sample unit, from the forest inventory data carried out in the study areas were used as response variable and as predictive variables used were height LiDAR metrics (position, dispersion and proportion), extracted, per sample unit, from the normalized LiDAR point cloud. From the results found, it was observed that the best equations for the prediction of biophysical variables reached an adjusted coefficient of determination (Adjusted R-squared) of 0.67, with a percentage standard error (SE%) of 20.22% for volume and an Adjusted R-squared of 0.75, with SE% of 14.71% for dry AGB, in addition, it was observed that the equations generated, both in volume and dry AGB, showed tendencies to overestimate lower values and underestimate higher values. Thus, it is concluded that the predictive equations of volume and dry AGB generated through LiDAR technology generated biased values compared to those obtained through the conventional forest inventory. In the second chapter, the potential of the association between metrics derived from LiDAR ALS technology and spectral data from different orbital sensors in the prediction of volume and dry AGB in areas of dry tropical forest was evaluated. To this end, the study was carried out in two fragments of Caatinga at Fazenda Itapemirim, located between the municipalities of Floresta and Betânia, in Pernambuco and data from LiDAR technology and orbital data from the Landsat 8, ResourceSat-2 and RapidEye satellites were used. For the adjustment of the multiple linear regression model, the response variables were the stocks of volume and dry AGB, per sampling unit, derived from the forest inventory data carried out in the study areas and as predictive variables, the LiDAR metrics of height and average values of spectral bands and vegetation indices. It was observed, from the results, that the best equation for volume prediction reached an Adjusted R-squared of 0.80, with SE% of 16.64% and the best equation for dry AGB obtained an Adjusted R-squared of 0.82, with SE% of 10.84%. It was also possible to observe that the generated equations showed tendencies to overestimate lower values and underestimate higher values. In this way, it is concluded that the predictive equations of volume and dry AGB adjusted from the integration between LiDAR ALS data and orbital data were not able to generate results with accuracy compatible with that obtained through the conventional forest inventory, however they presented good precision. In general, it is concluded that the integration of LiDAR ALS metrics and orbital data enabled the generation of more accurate volume and dry AGB estimates when compared to the use of LiDAR ALS data in isolation.