Integrated Use of Hyperspectral Remote Sensing and Geostatistics in Spatial Prediction of Soil Organic Carbon Content

Abstract

Estimation of spatial variability of soil organic carbon (SOC) content is important for agricultural management and environmental studies. In this study, geo-spatial prediction of SOC content was conducted to evaluate and compare geostatistical techniques of ordinary kriging (OK) and cokriging (CK) with hyperspectral satellite data (Hyperion) as an auxiliary variable. The study area located in western Uttar Pradesh, India. Hyperspectral satellite-derived spectral colour indices and spectral band reflectance used as auxiliary variables in CK. Spectral colour indices, viz. brightness index, coloration index, hue index and saturation index, were computed using atmospherically corrected Hyperion spectral bands reflectance data. Results of correlation analysis revealed that the spectral reflectance of Hyperion band 57 (1033.88 nm wavelength) has the highest correlation coefficient (r) with the SOC content (-0.86) followed by brightness index (-0.79), coloration index (-0.78), hue index (-0.77) and saturation index (-0.66). Both spherical and exponential models best fitted the semivariograms with SOC and spectral reflectance of Hyperion band 57 and also with SOC and soil coloration index as auxiliary variables. Cross-validation results indicated that among the kriging methods, CK performed best using exponential semivariogram model and soil coloration index as auxiliary variable for spatial prediction of SOC content with highest R 2 (0.653) and lowest RMSE (0.136%) values. In general, CK methods were superior to OK in spatial prediction of SOC content. A combination of cokriging geostatistical technique and remotely sensed hyperspectral data as auxiliary variables is very useful for improved and reliable spatial prediction of SOC content.

Synergistic Use of Multi-Temporal Radar and Optical Remote Sensing for Soil Organic Carbon Prediction

“…CI and BI showed a significant contribution to SOC prediction due to their ability to capture variations in soil color, which are often indicative of SOM content and other soil properties [63,64]. The correlation between SOC and CI and BI was already highlighted in previous studies, such as Saha et al [65], which demonstrated that different spectral color indices, especially CI, are important for SOC prediction and mapping.…”