Implementation and evaluation of the Landsat Ecosystem Disturbance Adaptive Processing Systems (LEDAPS) model: a case study in the Colombian Andes

Abstract

This paper analyzes the reflectance obtained with a series of Landsat images processed with LEDAPS model in a region of the Colombian Andes. A total of 38 images of TM and ETM sensors were calibrated to surface reflectance using LEDAPS in order to determine difference among bands of the same sensor, difference between sensors and analyze temporal patterns. Exact nonparametric statistics allow to conclude that: a) surface reflectance for band 1–5 and 7 were significantly different and this difference remains among images of different dates; b) there are statistical similarities between the TM and ETM sensors bands; c) temporal variations on surface reflectance from the years 1986 to 2013 with the sensors studied are not statistically significant. These results are supported by the implementation of robust modeling with various methods resistant to unusual observations and other typical problems of the classical least squares modeling.

Keywords

• LEDAPS
• Landsat
• corrección atmosférica
• prueba exacta de Wilcoxon-Mann-Whitney
• modelo de Huber
• regresión por mínimas desviaciones absolutas
• mínimos cuadrados recortados
• Bootstrap.