A numerical technique for delineation of soil mapping units using multi-spectral remote sensing data

A numerical technique for transformation of ground based soil spectral information into soil mapping - unit information, in terms of the total information content index has been proposed. The study earned out on 14 surface soil samples, widely differing in their physical appearance of colour and collected from different parts of India, revealed that total information content index could distinctly discriminate between the contrasting soil physiographic units with black cotton, red and sandy soil types. A comparison of the proposed index with the conventionally used two or three waveband specific indices (e.g. NIR/Red and NIR-Red/Red-Green) showed that the proposed index was more characteristic of the various soil types studied, further, unlike the conventional 2-D indices, the proposed numerical technique lead to the complete compression of the information contained in the entire reflectance spectrum (irrespective of the number of wavebands) to a single characteristic value in 1-D space and a simplified 1-D clustering analysis.