Metodología para la identificación de especies vegetales invasoras Ulex europaeus (retamo espinoso) y Genista monspessulana (retamo liso) en zonas piloto de Bogotá D.C usando imágenes multiespectrales de alta resolución.

Abstract

Invasive species are the second leading cause of biodiversity loss in terrestrial ecosystems, many of which are considered transformative species that for itself can drastically change the conditions and characteristics of terrestrial habitats as in the case of bushes Ulex europaeus (spiny gorse) and Genista monspessulana (smooth retamo), although the latter less aggressive, the spiny gorse is regarded worldwide as one of the 100 most aggressive invasive species on the planet and in Colombia it is among the top 10 more invasive species. One of the main challenges for ecological restoration is the restoration of areas with invasive alien species and to fulfill it is necessary to identify the areas affected by the invasion. However, identifying the location of this species through field trips is a task that would take a long time. For this reason the overall objective of the study was to develop a methodology from processing digital multispectral high-resolution satellite images for identifying the spatial distribution of invasive plant species in two areas of Bogotá D.C. Although he could not develop the methodology for plant smooth spice retamo, since it is mainly in the undergrowth which did not allow to identify the species by remote sensing.The general methodology of research includes satellite images for the pre-processing (atmospheric correction and generation of vegetation indices); processing classification using two environments: oriented pixels using unconventional object-oriented algorithms LDA, SVM and DT, and using the maximum likelihood algorithm. These two areas were tested to determine which one offers the best classification of satellite images and thus identify the spatial distribution of invasive plant species. After a thorough analysis, the best result was obtained using the object-oriented environment.