Relationship between TM/Landsat-5 spectral response and coffee agronomic variables

Abstract

Perennial crops have complex spectral behavior. Factors such as row-spacing, shadowing, crop seasonal characteristics, and substrate impinge noises that are difficult to interpret at satellite level. This study describes the influence of some coffee culture factors on spectral reflectance at TM/Landsat-5 level. At farm level, 145 coffee fields were selected. Agronomic variables like plant density and age, plant diameter and height, vegetative vigour and volume of green leaves, row-spacing and direction, substrate and soil spectral characteristics, and topographic features were collected. A correlation analysis was performed for agronomic variables only, and then the correlation between TM bands and agronomic variables was analysed. Some crop variables are well correlated such as coffee height and diameter, percent soil cover by coffee tree and harvest occurrence, pruning year and height. Pruning year and plant height were the most correlated parameters with TM1, TM2, TM3, TM5, and TM7 bands. Those parameters seem to aggregate information related to coffee phytomass and shadowing. A few agronomic variables correlated with TM4, except vegetative vigour. Parameters related to crop characteristics are more correlated with TM bands than parameters related to substrate and terrain geometry.