Original Articles

Interpretation of digital soil photographs using spatial analysis: I. Methodology

Published in: South African Journal of Plant and Soil
Volume 23, issue 1, 2006 , pages: 7–13
DOI: 10.1080/02571862.2006.10634723
Author(s): C.W. van HuyssteenDepartment of Soil, Crop and Climate Sciences, South Africa, P., A.L. le RouxDepartment of Soil, Crop and Climate Sciences, South Africa, M. HensleyDepartment of Soil, Crop and Climate Sciences, South Africa

Abstract

Systems for the evaluation of soil wetness use soil colour extensively. The determination of soil colour normally relies on the user's perception of colour and usually employs a colour matching system, e.g. Munsell Soil Color Charts. The South African soil classification system distinguishes between diagnostic grey, yellow-brown and red colours. This paper proposes a computerised methodology for the quantitative determination of diagnostic soil colour using spatial analysis. To achieve this, soil colour definitions had to be converted from Munsell colour notation to the RGB colour notation employed in digital cameras. The conversion also aided in the mathematical manipulation of colour data during spatial analysis. In this study the relationships between photographed and calculated RGB values were as follows: Calculated Red = 0.9238 ¥ Photographed Red −37.24 (R2 = 0.99); Calculated Green = 0.9975 ¥ Photographed Green −38.96 (R2 = 0.99); Calculated Blue = 0.9841 ¥ Photographed Blue −35.74 (R2 = 0.99). The following equations can be used to differentiate between diagnostic grey, yellow and red, as defined in Soil Classification—A Taxonomic System for South Africa, using the RGB values obtained from digital photographs and after correction using the equations given above: Between grey and yellow: Green = 0.88 ¥ Red 5; Between yellow and red: Green = 0.79 ¥-11. The methodology showed a 19% misclassification when tested against photographed Munsell sheets. It was, however, a huge improvement when compared to visual methods. The number of chips classified as grey, yellow-brown or red were equal to the number of chips defined as grey, yellow-brown or red. The misclassification was attributed to the discrete nature of the diagnostic colour definitions in contrast to the continuous nature of the differentiating equations.

Get new issue alerts for South African Journal of Plant and Soil