ABSTRACT
ABSTRACT: The spatial distribution of soil inorganic carbon was investigated with remote sensing methods. The relationship between laboratory spectral measurements and inorganic carbon content was accomplished with respect to the characteristic absorption features of carbonate and C.I.E. color coordinates. An empirical model for the spectral detection of carbonate content was generated which allows the prediction of soil inorganic carbon content with a cross-validated r2 of 0.957. In a second step the established model was modified to allow its application to Landsat images. Since C.I.E. color coordinates were found to be well suitable parameters for predicting the inorganic carbon content of soils under laboratory conditions, the reflectance values of the Landsat-TM bands were transformed into C.I.E. color coordinates. Subsequently, the C.I.E. based model approachwas adapted to a Landsat imagewith low vegetation cover from July 1998 to predict spatial distribution of the soils inorganic carbon content. Transferring the regression model to the satellite image allowed the prediction of the inorganic carbon content in the spatial dimension. The concentrations predicted from satellite data correspond well with the concentration range of the chemical analysis. They reflect the geographic conditions and show a dependence on the annual rainfall amount. A general trend to increasing concentrations of inorganic carbon can be stated with increasing aridity. Furthermore, local conditions are well reflected by the predicted concentrations.
1 INTRODUCTION
