Mapping of soil organic carbon using machine learning models: Combination of optical and radar remote sensing data

Abstract

Soil organic C (SOC) plays an important role in soil quality. Thus, it is of great significance to know the spatial distribution characteristics of SOC. Environmental variables, such as natural predictors, remote sensing (RS) variables, and digital soil mapping approaches have been widely used in SOC prediction. However, it is still challenging to determine which methods and variables are effective to map SOC in farmland. In this study, we compare the performance of three machine learning models, including random forest, Cubist, and the support vector machine model for the prediction of SOC contents using RS variables, topographic variables, climate variables, soil property variables, and spatial position variables as environmental covariates. The recursive feature elimination (RFE) technique was used to select optimal variables. Of the 130 sampling sites of the cultivated layer that were collected in Shanggao county, 70% of the data were used for training and 30% for validation. Prediction accuracy and uncertainty was quantified using the bootstrap approach with 100 iterations. The performance of the model was verified by the mean absolute error, RMSE, R2, and Lin's concordance correlation coefficient (CCC); the SD was used as a measure of prediction uncertainty. Results showed that the Cubist model combined with the RFE method (Cubist_RFE) had a mean R2, RMSE, and SD of .84, 2.39 g kg–1, and 0.73 g kg–1, respectively, which showed the highest prediction accuracy and the lowest uncertainty. The soil property variables (soil alkaline hydrolysis N and pH) and RS variables (soil adjusted vegetation index, enhanced vegetation index, and band7 of Sentinel‐2) were the controlling factors of SOC prediction. Thus, the results indicate the potential of the Cubist model in SOC prediction, and the combination of the optical and radar RS data is a viable way for predicting SOC.

Mapping Surface Organic Soil Properties in Arctic Tundra Using C-Band SAR Data

“…One study shows that including SAR data at longer wavelengths such as L-band helps distinguish different peatland classes from non-peatlands, but with less importance than Cband [25]. The different studies also report different conclusions on the relative importance of optical remote sensing versus SAR data for regional SOC characterization [25]- [27], which likely depends on the regional differences in topography, vegetation and surface wetness conditions represented from the various studies. The strong sensitivity of radar backscatter to multiple factors including land cover, surface snow and moisture changes affected by heterogeneous freeze/thaw and precipitation events may add additional challenges for regional mapping.…”

Preliminary Results in Innovative Solutions for Soil Carbon Estimation: Integrating Remote Sensing, Machine Learning, and Proximal Sensing Spectroscopy

“…When combined, remote sensing data enriches machine learning models with comprehensive environmental information, enabling precise analysis and prediction [17]. The multiple, spatially extensive spectral data from remote sensing serve as input to the machine learning models, leading to model the complex relationships for SOC estimations [24]. This approach can reduce the cost of measuring SOC by reducing the number of sampling profiles required for estimation of SOC, therefore providing a more cost-effective and scalable solution for large-scale SOC mapping [17,18,51,54].…”

“…The integration of techniques aims to overcome the limitations of traditional methods and provide cost-effective, accurate, and scalable solutions for SOC measurement [24,25]. Remote sensing allows for the collection of large-scale, spatially explicit data on soil properties, including SOC, by utilizing various sensors mounted on satellites or aircraft [20,25].…”

Research on Evaluation Methods of Black Soil Farmland Productivity Based on Field Block Scale

“…Machine learning models can effectively avoid this evaluation uncertainty caused by empirical knowledge and subjective judgement, and by learning from soil data, assessment models can be built more quickly and accurately. Zhou et al [14] used a combination of optical and radar remote sensing data to apply the SVM algorithm to build a Soil organic C (SOC) prediction model; Zou et al [15] collected historical soil data from southern China and combined multivariate linear model (MLM) and mixed effects regression model (MEM) for soil productivity assessment; Shehu et al [16] obtained 1781 sets of maize farmland data comparison in Northern Nigeria using linear regression models, as well as random forest machine learning to predict maize yields based on nutrient concentrations in spike leaves; Pan Y et al [17] provided an estimate of land productivity in the conterminous United States of America (CONUS) through machine learning algorithms using a data-driven approach to incorporate relationships from the data into the land productivity evaluation. However, challenges remain in terms of applicability and interpretability of machine learning models [18], including the requirement of datasets (e.g., combining large remote sensing datasets and larger historical datasets) and due to the black box nature of machine learning resulting in little insight into agricultural management.…”