A Generalised Soil-Adjusted Vegetation Index (GESAVI)

M.A. Gilabert, J. Gonzalez-Piqueras, F.J. García-Haro and J. Melia.

Operational monitoring of plant cover in large areas commonly relies on vegetation indices (VIs) that are determined using remote sensing. In most cases, these indices are functions of the reflectance in the red (R) and near infra-red (NIR) spectral bands.

In drylands with sparse plant cover, the contribution of soil to the total radiometric response is so large that hinders assessing that of vegetation. Such a soil background as a source of variation has received much attention in recent years, and soil adjusted indices (SAVI) have been introduced to address this issue.

As a contribution to solve this problem, a generalised soil-adjusted vegetation index (GESAVI) is being developed in the frame of SURMODES. Let us assume a Cartesian coordinate system generated by the red (R) and near-infrared (NIR) reflectances. The “soil line” is the straight line that best fits the reflectances measured for bare soils with varying amount of wetness, roughness, etc. Due to the presence of green vegetation, NIR reflectance increases and R reflectances decreases, and the corresponding points in the NIR/R plane moves to the upper left part of the soil line.

Lines connecting points corresponding to a similar vegetation amount over different soils are called “vegetation isolines”, which can be represented rather accurately by straight lines. In particular, one of the lines in the family, corresponding to absence of vegetation, is the soil line. The distance between a given point and the soil line (measured as an Euclidean distance or an angular difference) is directly connected with the vegetation amount of the considered canopy.

GESAVI is defined in terms of the soil line parameters (A and B) as:

GESAVI = (NIR-B*R-A)/(R+Z)

where Z is a soil adjustment coefficient related to the red reflectance at the cross point between the soil line and vegetation isolines.

In order to analyze the GESAVI sensitivity to soil brightness and soil color, high resolution reflectance data from laboratory experiments have been used. Different VIs from the SAVI family, including GESAVI were computed and their correlation with LAI for the different soil backgrounds was analyzed. Results confirmed the lower sensitivity of GESAVI to soil background in most of the cases, thus becoming the most efficient index .