Multispectral georeferenced information acquisition of vegetable species from Bogotá Botanic Garden using an unmanned aerial system

Abstract

In this paper  presents the spectral characterization of six plant species of interest to the urban trees of Bogotá by a low-cost unmanned aerial system (UAS). For this purpose a hexacopter FH-H800 equipped with PixHawk driver and autonomous flight capacity was assembled. Two standard cameras Canon A2300, one manually modified to allow data capture in the near infrared range was used. The photogrammetric flight planning was made for an altitude of 120 m above the ground in order to cover the area of the Botanical Garden by two flights of ten minutes each one. There were obtained approx. three hundred images with pixel size of 0.10 m covering the three channels of the visible spectrum (B, G, R) channel and the near infrared (NIR). The images were corrected by a photogrammetric process based on structure from motion (SFM) and stereo correspondence (MVS) using open source software that yielded a digital surface model (DSM) and develop multispectral orthofothomosaics (4 bands) with accurately horizontal position of 1.20 m for 90 % of the data. From the orthomosaics, identification and characterization of the spectral response of the following plant species was conducted: magnolia (Magnolia grandiflora), carbonero (Calliandra trinervia), sietecueros (Tibouchina lepidota), roble (Quercus humboldtii), pino colombiano (Podocarpus oleifolius) y sangregao (Croton funckianus). The results indicated that the near-infrared band is essential for the differentiation of these plant species. This work demonstrates the potential of UAS for fast and economical characterization and monitoring of natural or cultivated vegetation.