A study of the performance of the MSR vegetation index, using probabilistic and geostatistical methods

Abstract

In the present paper is studied the effect of the MSR (Modified Soil Ratio) vegetation index on multispectral digital images, with the aid of probability theory and geostatistics. Using proper distributions to describe the histograms of the image at the red and infrared band zones, an analytical expression of the distribution g of the MSR values is deduced. The study of the behaviour of g shows that the ratio of the standard deviation to the mean value of the MSR image is higher than that of the NDVI vegetation index, which is quite often used. This means that the MSR vegetation index produces images with a good contrast. It was also observed that the MSR image has a better signal to noise ratio than that of the NDVI image. Finally, the autocorrelograms of the MSR and NDVI images showed that the tonality differences between adjacent pixels of the MSR image are slightly stronger than those of the NDVI image. The general conclusion is that the MSR vegetation index produces images with a good contrast and a high signal to noise ratio, which could aid in making a reliable mapping of the vegetation cover of the area under study