Variación del índice de vegetación de diferencia normalizada (NDVI) en relación con la gradiente altitudinal en las lomas de Atocongo (Lima – Perú)

• Autor: Cuya Matos, Oscar Alejandro
• Formato: tesis de maestría
• Fecha de Publicación: 2016
• Institución: Universidad Ricardo Palma
• Repositorio: URP-Tesis
• Lenguaje: español
• OAI Identifier: oai:repositorio.urp.edu.pe:20.500.14138/972
• Enlace del recurso: https://hdl.handle.net/20.500.14138/972
• Nivel de acceso: acceso abierto
• Materia: índice de vegetación de diferencia normalizada (NDVI); ecosistema de lomas; gradiente altitudinal; lomas de Atocongo

The fog oases, locally called hills, are ecological patches that are distributed between 6 ° to 30 ° South Latitude, on the desert coast of Peru and Chile, and constitute fragile ecosystems. The area of study was located between the districts of Villa María del Triunfo and Pachacamac (Lima - Peru), in the sector of Atocongo, between the coordinates UTM 293372 and 294615 Easting; 8650536 and 8649292 Northing, which include the hills of Puquio, Lúcumo, Quebrada Verde and Guayabo. A direct analysis of gradients was made, following the original approach of Robert Whittaker. The research was retrospective. Vegetation inventories were used of 2000, 2001, 2002, 2007 and 2011 and SPOT images of 20 m of spatial resolution of 30 November 1998 were used; IKONOS, dated 10 March 2002, of 1 m of spatial resolution; ASTER, dated 03 March 2001, of 15 m resolution; WORLD VIEW, dated March 10, 2011, with a 0.5 m spatial resolution. In the exploratory analysis it was verified that the data were not normal, they presented anisotropy and in some cases spatial autocorrelation. We chose to apply generalized linear models and generalized additive models for the direct analysis of gradients. The study verified the existence of a positive linear relationship between vegetation cover as an explanatory variable versus normalized difference vegetation index (NDVI) as a dependent variable. It also verified the existence of linear and unimodal relationships between altitude as an explanatory variable versus vegetation cover as a dependent variable; and between the altitude as an explanatory variable versus the NDVI as a dependent variable. The values of deviance, AIC (Akaike's information criterion) and p-value were highly variable. Not all analyzes were close to the ideal. For the case of the influence of altitude, a high deviation was expected, p-value less than 0.05, low AIC, quadratic, with gamma distribution, logarithmic link function and unimodal form. The seasonality of the climate and the orientation of each basin to the winds largely explain the variability of the models obtained. The NDVI isolines showed nuclei of higher values, on which the isolines are concentrically organized. Whenever the set of samples intercepts a core of high values will have a unimodal behavior, otherwise the behavior will tend to be linear.