Quantification of soil organic carbon (SOC) and pH, and their spatial variations at regional scales, is a foundation to adequately assess agriculture, pollution control, or environmental health and ecosystem functioning, so as to establish better practices for land use and land management. In this study, we used the random forest (RF) model to map the distribution of SOC and pH in the topsoil (0–20 cm) and estimate SOC and pH changes from 1982 to 2012 in Liaoning Province, Northeast China. A total of 10 covariates (elevation, slope gradient, topographic wetness index (TWI), mean annual temperature (MAT), mean annual precipitation (MAP), visible-red band 3 (B3), near-infrared band 4 (B4), short-wave infrared band 5 (B5), normalized difference vegetation index (NDVI), and land-use data) and a set of 806 (in 1982) and 973 (in 2012) soil samples were selected. Cross-validation technology was used to test the performance and uncertainty of the RF model. We found that the prediction R2 of SOC and pH was 0.69 and 0.54 for 1982, and 0.63 and 0.48 for 2012, respectively. Elevation, NDVI, and land use are the main environmental variables affecting the spatial variability of SOC in both periods. Correspondingly, the topographic wetness index and mean annual precipitation were the two most critical environmental variables affecting the spatial variation of pH. The mean SOC and pH decreased from 18.6 to 16.9 kg−1 and 6.9 to 6.6, respectively, over a 30-year period. SOC distribution generated using the RF model showed a decreasing SOC trend from east to west across the city in the two periods. In contrast, the spatial distribution of pH showed an opposite trend in both periods. This study provided important information of spatial variations in SOC and pH to agencies and communities in this region, to evaluate soil quality and make decisions on remediation and prevention of soil acidification and salinization.

Forest ecosystems play an important role in regional carbon and nitrogen cycling. Accurate and effective monitoring of their soil organic carbon (SOC) and soil total nitrogen (STN) stocks provides important information for soil quality assessment, sustainable forestry management and climate change policy making. In this study, a geographical weighted regression (GWR) model, a multiple stepwise regression (MLSR) model, and a boosted regression trees (BRT) model were compared to obtain the best prediction of SOC and STN stocks of the forest ecosystems in northeastern China. Five-hundred and thirteen topsoil (0–30 cm) samples (10.32 kg m−2 (±0.53) for SOC, 1.21 kg m−2 (±0.32) for STN), and 9 remotely-sensed environmental variables were collected and used for the model development and verification. By comparing with independent verification data, the best model (BRT) achieved R2 = 0.56 and root mean square error (RMSE) = 00.85 kg m−2 for SOC stocks, R2 = 0.51 and RMSE = 0.22 kg m−2 for STN stocks. Of all the remotely-sensed environment variables, soil adjusted vegetation index (SAVI) and normalized difference vegetation index (NDVI) are of the highest relative importance in predicting SOC and STN stocks. The spatial distribution of the predicted SOC and STN stocks gradually decreased from northeast to southwest. This study provides an attempt to rapidly predict SOC and STN stocks in the dense vegetation covered area. The results can help evaluate soil quality and facilitate land policy and regulation making by the government in the region.

Forest soil organic carbon (SOC) accounts for a large portion of global soil carbon stocks. Accurately mapping forest SOC stocks is a necessity for quantifying forest carbon cycling and forest soil sustainable management. In this study, we used a boosted regression trees (BRT) model to predict the spatial distribution of SOC stocks during two time periods (1990 and 2015) and calculated their spatiotemporal changes during 25 years in Liaoning Province, China. A total of 367 (1990) and 539 (2015) sampling sites and 9 environmental variables (climate, topography, remote sensing) were used in the BRT model. The ten-fold cross-validation technique was used to evaluate the prediction performance and uncertainty of the BRT model in two periods. It was found that the BRT model could account for 65% and 59% of SOC stocks, respectively for the two periods. MAP and NDVI were the main environmental variables controlling the spatial variability of SOC stocks. Over the 25-year period, the average SOC stocks increased from 5.66 to 6.61 kg m −2 . In the whole study area, the SOC stocks were the highest in the northeast, followed by the southwest, and the lowest in the middle of the spatial distribution pattern in the two periods. Our accurate mapping of SOC stocks, their spatial distribution characteristics, influencing factors, and main controlling factors in forest areas will assist soil management and help assess environmental changes in the region. soil properties are affected by many environmental factors, it is difficult to predict soil properties accurately and efficiently at the regional scale [8]. As an efficient and low-cost method, digital soil mapping (DSM) technology is able to accurately describe the spatial variability of soil attributes in a region by using a limited amount of sampling data and environmental variables [9]. In order to accurately predict the SOC stocks in a region and analyze its key environmental factors, scholars have carried out a lot of researches [10][11][12]. In fact, the direct relationship between soil properties and environmental factors is non-linear and complex [13,14]. Therefore, machine learning algorithms which can effectively avoid this kind of problem are widely used in DSM mapping [11,15]. For example, on Barro Colorado Island, Panama, Grimm et al. [15] used a random forest model to simulate SOC stocks in 0-10 cm, 10-20 cm, 20-30 cm, and 30-50 cm layers, respectively. Giasson et al. [16] applied a multiple logistic regression model to predict the occurrence of soil types in southern Brazil. Dorji et al. [17] selected regression kriging and equal-area spline profile function to predict the SOC stocks at depths 0-5, 5-15, 15-30, 30-60, and 60-100 cm in montane ecosystems, Eastern Himalayas. In Denmark, Adhikari et al. [18] used regression kriging and 12 environmental variables to simulate the spatial distribution of SOC stocks at five soil layers. Were et al. [19] compared the performance of support vector regression, artificial neural network, and random forest models in predicti...

Soil organic carbon (SOC) and soil total nitrogen (STN) are major soil indicators for soil quality and fertility. Accurate mapping SOC and STN in soils would help both managed and natural soils and ecosystem management. This study developed an improved similarity-based approach (ISA) to predicting and mapping topsoil (0–20 cm soil depth) SOC and STN in a coastal region of northeastern China. Six environmental variables including elevation, slope gradient, topographic wetness index, the mean annual temperature, the mean annual temperature, and normalized difference vegetation index were used as predictors. Soil survey data in 2012 was designed based on the clustering of the study area into six climatic vegetation landscape units. In each landscape unit, 20–25 sampling points were determined at different landform positions considering local climate, soil type, elevation and other environmental factors, and finally 126 sampling points were obtained. Soil sampling from the depth of 0–20 cm were used for model prediction and validation. The ISA model performance was compared with the geographically weighted regression (GWR), regression kriging (RK), boosted regression trees (BRT) considering mean absolute prediction error (MAE), root mean square error (RMSE), coefficient of determination (R2), and maximum relative difference (RD) indices. We found that the ISA method performed best with the highest R2 and lowest MAE, RMSE compared to GWR, RK, and BRT methods. The ISA method could explain 76% and 83% of the total SOC and STN variability, respectively, 12–40% higher than other models in the study area. Elevation had the largest influence on SOC and STN distribution. We conclude that the developed ISA model is robust and effective in mapping SOC and STN, particularly in the areas with complex vegetation-landscape when limited samples are available. The method needs to be tested for other regions in our future research.