Vegetation indices for remote sensing of beans (Phaseolus vulgaris L.)

Abstract

Vegetation indices (VIs) are an important tool in remote sensing of vegetation. These VIs are composed of two or more single bands and they are supposed to minimize some intervening effects such as soil background, view and illumination angles and atmospheric effects. However, there is a lack of knowledge on how different VIs behave on a developing crop over different soils and submitted to different viewing conditions; the exact role the bands play in the VIs under those conditions is to be understood. This paper describes an experiment in which bean (Phaseolus vulgaris L.) was planted on the same soil, but in a changing three soil backgrounds, under two view angles and four times of the day, in four dates along the crop cycle, and in randomly blocks design. It is shown that the analysis of individual bands is very important for understanding the behavior of VIs. Soils were statistically significant in all stages analyzed for the three VIs; however, the view angle was significant only for NDVI and for low ground cover. The illumination angle was significant for all VIs and for all ground covers, except for 100%. Moreover, there was an expected NDVI saturation as the ground cover increased while the SAVI and PVI keep on responding to treatments, mainly in function of their sensitivities to near infrared.