Application of Remote Sensing Techniques in the Estimation of Forest Biomass of a Recreation Area by UAV and RADAR Images in Ecuador

In recent years, the estimation of biomass through the combination of field and remote sensing data has been gaining ground in several countries, as it becomes an option that reduces costs in addition to obtaining information in inaccessible areas. Despite the large amount of accessible spatial information, the subject of biomass estimation by remote sensing, especially radar remains a new and interesting topic in the application of forestry studies. We have used radar images and UAV in order to analyze the estimation of biomass. The study area has been performed in the Metropolitan Park “La Armenia” of Quito, central Ecuador, in which 13 plots have been established (1ha). The Forest Aboveground Biomass (FAB) has been estimated and therefore also its fixation of aerial carbon. The applied methodology included to develop the semi-empirical algorithms proposed for the estimation of biomass, in order to correlate the field data including estimation by allometry, with the digital levels of the radar image. Thus, we reached a coefficient of determination of 0.74, which provided a clear relationship between field data and the backscatter coefficients. Within the methodology used with the UAV, in the estimation of FAB by CMS (model 2) it has been adjusted to a logarithmic regression with a coefficient of determination r2 = 0.67, while the obtaining of model 3 by green normalized vegetation index (GNDVI) has been adjusted to a quadratic regression, with a coefficient of determination r2 = 0.57. The adjustment of the three models has been defined as a goodness of positive adjustment, concluding that model 1 is the better one. However, it has been necessary to indicate that model 1 does not conform to reality, since there is a great difference between the dates of taking the radar image versus the data acquired in the field. Therefore, model 2 represents reality better, since it has been able to best characterize abrupt changes in the tree height variable, due to anthropic or natural factors, such as felling, burning, lightning in the tree, among others, that model 3 lacked to perform. The estimation of biomass and carbon fixation based on direct observations of the territory through remote sensing techniques, may provide more reliable information for decision making in the development of policies related to sustainable forest management.