Transformación del bosque tropical seco en la región del alto magdalena (Tolima- Colombia): valor predictivo de variables ambientales.

The knowledge of the transformation of biodiversity from remote sensing and the estimation and analysis of indices spectral can become a practical way to evaluate the territory and its resources, in addition to being a technique that can provide information base to guide decision making in the identification of priority areas for conservation. In this research I determined the predictive value of environmental variables (topographical, hydrological, anthropogenic and biomass) through statistical procedures with the purpose to analyse and establish if possible the transformation of space-time for subsequent years of the coverage of Tropical Dry Forest in the region of alto Magdalena (Colombia). Is processed by a series composed of 112 images of the sensor Landsat 4-5 TM, Landsat 7 ETM+ and Landsat 8, corresponding to the periods of dry season and wet in the temporalities 1987, 2000 and 2014; next to a series subsequent to the years 1990, 1995 and 2010 are used as years of control for the values of the variables of Biomass (Indices NDVI and NDII). To improve the level of interpretation of the changes that had the coverage I performed a correction of the values by means of the TVI (Vegetation Index Transformed) and the ranges established by Kalacska et al. (2004) for NDVI in BTs. For the variables of anthropogenic disturbance and watersheds, was applied and modified the methodology suggested by Quijas (2011) where they evaluated a series of distances euclidean from sampling sites with respect to hedges closest and with the greatest impact on the plant communities, which in this case were Grasses, Crops and bare ground and degraded. Considering, however, as the spatial scale can affect the ability of different predictor variables of biomass, was calculated from the values of the indices of plant biomass (NDVI, NDII) three spatial scales: 50, 150 and 300 m. These data were added as the fifth group of predictor variable and is called “Donuts”. Finished the processing, we obtained a total of 28 predictive variables, which were grouped and processed according to its attribute by means of the statistical programmes SPSS and JMP to obtain the 15 best models of testing for each year, giving as result a mathematical algorithm of prediction with best variables to set area.The data obtained are presented below.

• Autores: Merchán Garzón, Jairo Andrés
• Tipo de recurso: Trabajo de grado de pregrado
• Fecha de publicación: 2017
• Institución: Universidad de Ciencias Aplicadas y Ambientales U.D.C.A
• Repositorio: Repositorio Institucional UDCA
• Idioma: spa
• OAI Identifier: oai:repository.udca.edu.co:11158/826
• Acceso en línea: https://repository.udca.edu.co/handle/11158/826
• Palabra clave: Teledetección; NDVI; Variables Predictivas; Conservación; Ordenamiento Ambiental; Modelos Predictivos; Bosques secos; Ecología forestal; Ordenamiento territorial; Ingeniería Geográfica y Ambiental
• Rights: closedAccess
• License: Derechos Reservados - Universidad de Ciencias Aplicadas y Ambientales