Remote Sensing of Soil Organic Carbon in Semi-Arid Region of Iran

Abstract

The soil organic carbon (SOC) concentration is a crucial soil property to guide agricultural applications. Researchers have used remotely sensed data to estimate and quantify the SOC content. Our objective is to compare the performance of the existing techniques of simple regression models (SRM), principal component analysis (PCA), and the soil line approach in SOC estimation in a semi-arid environment. Models were developed between dependent variables of SOC and independent variables of digital value of soil reflectance in satellite bands, Euclidian distance from soil line (D), and first principal component (PC1). The SRM technique provided the most accurate SOC predictions (R2 = 0.75) but the accuracy for PCA and soil line techniques were R2 < 0.44. Our result reveals the SRM technique can be used in management decision making when the cost and rate of mapping procedure is more important than SOC measurement accuracy.