In estimating the vegetation coverage from NDVI (Normalized Vegetation Index) or SAVI (Soil Adjusted Vegetation Index, Huete, 1988) which are calculated from the satellite data, the vegetation density within a pixel affect the results. To eliminate this undesirable effect, an optimum vegetation index (OPVI) is presented. In this method, the vegetation coverage within each pixel is calculated from the reflectance ratio of radiance in band 4 band to that in band 5 of ASTER (TM5 and TM 7, in case of Landsat TM). Then, if the reflectance ratio is larger or smaller than a critical value, select either NDVI or SAVI for calculating the vegetation index in each pixel. For determining the critical value, the relationship between vegetation coverage and the reflectance ratio are studied from the spectral reflectance of plants and soils as a function of ratio of coverage area of plants to that of soils in the laboratory. Then, the critical value is determined by comparing the results obtained by
NDVI and SAVI.Comparison of results obtained by OPVI method with those by both NDVI and SAVI are made by using the satellite data collected in and around oases in Taklimakan Desert in China since the vegetation is expected to vary over the wide range. Figure shows the false color image and the NDVI, SAVI, and OPVI maps of the object area for this study. Most of the area inside the oasis in the NDVI map shows higher vegetation index than in OPVI.Further, a high vegetation index in the NDVI is shown outside the oasis which has a lower vegetation density. On the other hand, the vegetation index in the SAVI, as a whole, is evaluated to be a low even inside the oasis. Particularly, even in a high vegetation density such as fruit gardens and cotton fields as confirmed by in situ survey, the vegetation obtain from the SAVI is too low.
The results show that OPVI gives a better indicator of actual vegetation cover in an arid area than single NDVI or SAVI alone. Further, OPVI becomes effective to investigate the desertification, which is becoming at issue in many arid and semi-arid areas in the world.