Classification of the most important spectral factors extracted from Landsat-8 images in explaining the topsoil organic carbon in semi-steppe rangelands using exploratory factor analysis (EFA)

Background and Objective:

Soil organic carbon in rangeland ecosystems has a variety of functions such as increasing soil fertility, controlling erosion, increasing soil water permeability and, reducing the effects of greenhouse gases. Therefore, it is a key indicator in determining soil health that affects all physical, chemical, and biological properties of soil. The large area of the country's rangelands causes a serious challenge to the use of traditional methods in estimating soil organic carbon. In such situations, the use of remote sensing capabilities can be considered as a suitable option for monitoring the organic carbon of the country's rangeland soils.  The aim of this study was to determine the most important spectral factors affecting topsoil organic carbon in two summer rangelands.

Materials and Methods:

 This research was carried out in two summer rangelands of Lazour and Asaran. The first rangeland (Lazour) with an area of 8150 hectares and an average height of 2875 meters is located in the range of eastern longitudes 52.514 to 52.694 degrees and northern latitudes 35.855 to 35.934 degrees in Tehran province. The second Rangeland (Asaran) with an area of 5642 hectares and an average height of 2465 meters is located in the range of eastern longitudes 53.265 to 53.392 degrees and northern latitudes 35.804 to 35.882 degrees in Semnan province. In this research, the data of the OLI sensor of the Landsat 8 satellite were used. After pre-processing satellite imagery of the studied areas, Top of Atmosphere (TOA) reflectance layers of bands 2 to 7 along with the variables of surface albedo, Clay index, Carbonate index, Grain Size index, NDVI, brightness, greenness, and wetness index of Tasseled cap transformation were calculated. In each of the target areas, using Digital Elevation Model (DEM) maps, the slope, aspect, and hypsometric maps were prepared and by combining the last three layers with each other, a map of homogeneous sampling units was obtained. Soil sampling was performed using the stratified-random sampling pattern. In this way, in each of the homogeneous units, according to its area, several soil samples were randomly taken from a depth of zero to 20 cm and the amount of organic carbon of the samples was measured using the Walkley-Black method.

Results and Discussion:

 The results of this study showed that the spectral variables of Top of Atmosphere (TOA) reflectance layers of bands 2 to 7 along with the variables of surface albedo, Clay index, NDVI, brightness, greenness, and wetness index of Tasseled cap transformation have a significant correlation with topsoil organic carbon (p < 0.01). Also, the results of factor analysis by principal component analysis (PCA) with eigenvalues greater than one showed that the total cumulative variance explained by the 12 variables is 91.74%, which was explained by two factors. The first factor (soil color) explained 76.6% of the variance and the second factor (vegetation and soil texture) explained 15.14% of the variance.

Conclusion:

 The results of this study confirm the existence of a significant relationship between topsoil organic carbon and spectral factors extracted from Landsat 8 OLI sensor data in semi-steppe rangelands. Because of the large area of rangelands in Iran, the use of traditional methods in estimating soil organic carbon is not possible due to the need to spend a lot of time and money. And in such situations, the use of Remote sensing (RS) capabilities can be considered as a suitable option for monitoring the topsoil organic carbon in the rangelands.