Identificación de la especie Polylepis reticulata mediante teledetección en las zonas alto andinas del Ecuador.

Abstract

The present research proposes: to identify the species Polylepis reticulata by remote sensing in the high Andean areas of Ecuador, through the processing of an orthophotograph with a resolution of 0.0938 meters and Sentinel 2 A satellite images with a resolution of 10 meters; where three training sessions were carried out in orthophotography with discriminations of NDVI≤0.01; NDVI ≤0.04; NDV ≤0.07 and two workouts in the Sentinel 2A scenes of the Random Forest algorithm. This process was developed by calculating the vegetation spectral indexes, generating a predictive model of categorization by means of decision trees; the results show that the relevant factors in the categorization for each training are: in the three workouts of the orthophotography, the variables of NDVI, SAVI5, SAVI15, have a level of importance equal to 100%, while in the satellite image they do not have areas of great size or considered of importance did not make an adequate identification of the species under study, in the same way a comparison was made in the effectiveness of identification of Polylepis in 100 hectares through three softwares: ArcGis, Envi, RF, the supervised classification that allowed to identify Polylepis coverage areas, in the first software it is identified 4.68% (1) Polylepis and 95.32% (2) No Polylepis, in the second software it identifies 8.89% (1) Polylepis and 91.11% (2) No Polylepis, finally, in the third software identifies 2.32% (1) Polylepis and 97.68% (2) No Polylepis showing in the raster information a better identification with the software RF and determining the species with greater precision. KEYWORDS: TELEDETECTION - SPECTRAL INDICES - RANDOM FOREST - POLYLEPIS - ALTE ANDINAS VEGETATION.