Evaluation of soil calcium carbonate using of satellite images and NIR spectroscopy

Document Type : Original Article


1 Msc of Soil Science, Faculty of Agriculture, Malayer University, Malayer

2 Department of Soil Science, Faculty of Agriculture, Malayer University, Malayer


One of the benefits of remote sensing and Vis–NIR are rapidity, simplicity, and low cost of analysis compared with traditional methods. The aim of this research, is to use Landsat 8 satellite sensors and near infrared spectrum in agricultural and forestry uses in Gyan Nahavand plain, Hamadan province to estimate soil calcium carbonate. 48 soil samples were collected from Gyan randomly and some physico-chemical characteristics of soils were analyzed. The correlation between the value of the main bands, composition of the bands, calcite indices with the amount of soil calcite was done. Spectral analysis of the desired soils was done using a fieldSpec3 with a wavelength range of 350-2500 nm. Laboratory results showed that the average of soil calcium carbonate in agriculture and forest use are 30 and 22.22%, respectively. The results showed that the bands 10 and 11 had a significant correlation with soil calcite in forest use (p<0.05). 12 band compositions at the 5% level and 6 band compositions at the 1% level showed a significant correlation with the amount of soil calcite. Also, R1 index ((Band5/Band4)/(Band5/Band2)) with soil calcite had significant correlation (p<0.05). The correlation between the measured calcite in laboratory and the equation achieved from satellite image was found to be equal (r2= 0.45) for agriculture use. In the spectroscopic method, the highest correlation was observed at the wavelength of 612 nm (r2=0.85**). Among the models fitted by multivariate regression, in satellite images, the Stepwise Multivariate Linear Regression (SMLR) model is suggested as a suitable model for calcite estimation. The partial least square regression (PLSR) model has been an almost suitable for estimating calcite by spectroscopic method. It can be concluded that Vis-NIR spectroscopy method has more accuracy than remote sensing and titration methods, but it requires more.