Soil Profile Organic Carbon Prediction And Genetic Horizon Identification Based On Visible-near-infrared Hyperspectral Imaging Spectroscopy

Complete and coherent soil profile information is important for studying soil morphological characteristics,vertical variation patterns of soil physicochemical properties and clarifying soil types.However,traditional soil survey methods use a limited number of discrete sample points at different depths as a data source for a series of studies.Due to the variability of soil physicochemical properties at vertical depth,there is a strong uncertainty in the conclusions drawn from the limited sample points.With the development of near-earth remote sensing technology,the near-earth hyperspectral imaging system can acquire hyperspectral images of soil profiles and characterize the subtle texture differences and spectral variability of soils with continuous spectral profiles and images,which play an important role in inversion of soil physicochemical properties and identification of soil genetic horizons and soil types.Soil organic carbon(SOC)is one of the important physicochemical properties to identify the genetic horizon,and the genetic horizon type is an important part of soil classification,therefore,this paper explored the ability of visible–near-infrared(Vis-NIR)hyperspectral imaging spectroscopy to predict organic carbon and identify the genetic horizon in soil monoliths.The dataset contained five soil monoliths retrieved from the upper Yellow River,China.To reduce the effort of collecting soil samples at equal intervals,the organic carbon content in the genetic horizon was used to establish a depth function to obtain the continuous content variation with the depth.Considering the soil heterogeneity,the average spectrum within the region of interest(ROI)of the image could not characterize the variation in soil spectra,so the accuracy differences in five spectral statistics of ROI: maximum(max),mean,median,mode,and minimum(min)were compared.Specifically,random forest(RF)models were developed for organic carbon prediction and genetic horizon identification,respectively.In organic carbon prediction,the differences between modeling with and without depth function were compared,and the effect of three equal-spacing sampling schemes(1,5,and 10 cm)on the depth function and impact of the above five spectral statistics on the accuracy were compared;in the genetic horizon identification,the classification accuracy of five spectral statistics at 1-cm equal-spacing sampling scheme was evaluated.The results indicated that hyperspectral imaging spectroscopy could predict organic carbon and genetic horizon identification in soil profiles,(1)the depth function under the 1-cm sampling scheme predicted organic carbon the best;(2)the main genetic horizons and their subcategories could be well identified,and the distribution areas could be identified for the transition horizons;(3)the mode of the spectra within the ROI could yield a comparable or even a higher accuracy than that obtained with the mean,the median model was second only to the mean model,and the maximum and minimum models could not provide accurate predictions.The predicted organic carbon and genetic horizons could be mapped at a high resolution using hyperspectral imaging spectra to characterize fine soil profile changes,thus overcoming the limitations of conventional soil sampling.