干旱区生态环境知识资源中心

Mapping of soil organic carbon (SOC) was accomplished with remote sensing methods to assess its spatial variability. The relationship between bi-directional reflectance measurements and SOC was investigated with respect to C.I.E. color coordinates. Empirical relationships were generated for the spectral detection of SOC of two semi-arid to arid study sites. These regression models allowed the Prediction of SOC with a cross-validated r(2) = 0.910 (RMSE(cv) = 2.825) and r(2) of 0.795 (RMSE(cv) = 2.113), respectively. Because C.I.E. color coordinates notes were found to be appropriate parameters for predicting the SOC, reflectance values of Landsat TM bonds were transformed into C.I.E. color coordinates, The C.I.E. based regression models were applied to a Landsat image to estimate SOC in the spatial domain. Concentrations predicted from satellite data corresponded well with concentration ranges derived from chemical analysis. Estimated concentrations reflect the geographic conditions and depend on annual rainfall, with a general trend to decreasing SOC with increasing aridity.