Soil color modeling for the visible and near-infrared bands of Landsat sensors using laboratory spectral measurements

Abstract

Color is a key feature used in the identification and classification of soils. Soil reflectance has a direct relationship with soil color, as well as to other parameters such as texture, soil moisture, and organic matter. Earlier studies have quantified the relationship between reflectance and soil color using simple linear regression equations having correlation coefficients of about 0.5. In the present study, a method called optimal rotational transformation technique has been employed to maximize the correlation between soil color and transformed reflectance. Seventy-six soil samples having a wide range of soil color were employed in the laboratory measurements. Multispectral reflectances were measured using an Exotech radiometer in the visible and near-infrared bands of the Landsat MSS sensor, and soil color was measured using the Munsell color charts. The soil color data were quantified using the RGB color coordinate system since colors recorded in the Munsell system are not readily usable in numerical analysis. Simple linear regression models using the raw reflectance data yielded correlations of about 0.5. Optimal rotational transformation of the reflectance data and multiple linear regressions showed an improved correlation of r more than 0.8, and the models predicted the main color component (viz. hue) with acceptable accuracy. Hue images can be used for identification and discrimination of soils and lithological formations, since hue is independent of illumination and therefore capable of suppressing shadows. Optimal models can be developed to predict soil reflectance using color data since metamerism is rarely observed in the case of soils. Results of this study have a potential application for identification and mapping of soil and geologic materials of nonvegetated or sparsely vegetated regions using data received from airborne and space-borne remote sensors operating in the visible and near-infrared bands.