Evaluation of Landsat TM5 Multispectral Data for Automated Mapping of Surface Soil Texture and Organic Matter in GIS

Abstract

Mapping fine scale spatial variations of soil properties is important for site specific agriculture. The current study explores the potentials of remote sensing (RS) and geographical information system (GIS) techniques in studying the spatial variability of surface soil attributes. Around 170 surface (0–30 cm) soil samples collected from the soils of Shorkot Tehsil, Punjab, Pakistan were analyzed for surface soil texture and organic matter (O.M.). A multivariate linear regression (MLR) analysis technique was employed to relate surface soil variables with the spectral data from Landsat TM5 satellite. The MLR analysis showed significant (p<0.05) relationship of band 4 and band 6 with silt% (R2 = 0.724) and clay% (R2 = 0.509) while soil O.M. was best modeled using data from band 1, 6 and 7 (R2 = 0.545). The resulting MLR equations were then used for the spatial modeling of these attributes for the entire study area. For developing surface soil texture map, the USDA textural triangle limits for clay% and silt% were used to develop a code in Visual Basic Language in ArcGIS environment. The results showed that ‘sandy clay loam’ was the most abundant textural class in the area followed by ‘sandy loam’ and ‘clay loam’ classes. Moreover, the status of O.M. in the entire study area soils was very poor (<1%). The results indicate that RS and GIS techniques could be successfully used for fine scale mapping of soil texture and O.M. of a larger area.