Landscape Classification System Based on RKM Clustering for Soil Survey UAV Images-Case Study of the Small Hilly Areas in Jurong City.

With the advantages of high accuracy, low cost, and flexibility, Unmanned Aerial Vehicle (UAV) images are now widely used in the fields of land survey, crop monitoring, and soil property prediction. Since the distribution of soil and landscape are closely related, this study makes use of the advantages of UAV images to classify the landscape to build a landscape classification system for soil investigation. Firstly, land use, object, and topographic factor were selected as landscape factors based on soil-forming factors. Then, based on multispectral images and Digital Elevation Models (DEM) acquired by UAV, object-oriented classification of different landscape factors was carried out. Additionally, we selected 432 sample data and validation data from the field survey. Finally, the landscape factor classification results were superimposed to obtain the landscape unit applicable to the system classification. The landscape classification system oriented to the soil survey was constructed by clustering 11,897 landscape units through the rough K-mean clustering algorithm. Compared to K-mean clustering, the rough K-mean clustering was better, with a Silhouette Coefficient of 0.26247 significantly higher than that of K-mean clustering. From the classification results, it can be found that the overall classification results are somewhat fragmented, but the landscape boundaries at the small area scale are consistent with the actual situation and the fragmented small spots are less. Comparing the small number of landscape boundaries obtained from the actual survey, we can find that the landscape boundaries in the landscape classification map are generally consistent with the actual landscape boundaries. In addition, through the analysis of two soil profile data within a landscape category, we found that the identified soil type of soil formation conditions and the landscape factor type of the landscape category is approximately the same. Therefore, this landscape classification system can be effectively used for soil surveys, and this landscape classification system is important for soil surveys to carry out the selection of survey routes, the setting of profile points, and the determination of soil boundaries.

Digital Mapping of Soil Organic Carbon Based on Machine Learning and Regression Kriging.

In the last two decades, machine learning (ML) methods have been widely used in digital soil mapping (DSM), but the regression kriging (RK) model which combines the advantages of the ML and kriging methods has rarely been used in DSM. In addition, due to the limitation of a single-model structure, many ML methods have poor prediction accuracy in undulating terrain areas. In this study, we collected the SOC content of 115 soil samples in a hilly farming area with continuous undulating terrain. According to the theory of soil-forming factors in pedogenesis, we selected 10 topographic indices, 7 vegetation indices, and 2 soil indices as environmental covariates, and according to the law of geographical similarity, we used ML and RK methods to mine the relationship between SOC and environmental covariates to predict the SOC content. Four ensemble models-random forest (RF), Cubist, stochastic gradient boosting (SGB), and Bayesian regularized neural networks (BRNNs)-were used to fit the trend of SOC content, and the simple kriging (SK) method was used to interpolate the residuals of the ensemble models, and then the SOC and residual were superimposed to obtain the RK prediction result. Moreover, the 115 samples were divided into calibration and validation sets at a ratio of 80%, and the tenfold cross-validation method was used to fit the optimal parameters of the model. From the results of four ensemble models: RF performed best in the calibration set (R2c = 0.834) but poorly in the validation set (R2v = 0.362); Cubist had good accuracy and stability in both the calibration and validation sets (R2c = 0.693 and R2v = 0.445); SGB performed poorly (R2c = 0.430 and R2v = 0.336); and BRNN had the lowest accuracy (R2c = 0.323 and R2v = 0.282). The results showed that the R2 of the four RK models in the validation set were 0.718, 0.674, 0.724, and 0.625, respectively. Compared with the ensemble models without superimposed residuals, the prediction accuracy was improved by 0.356, 0.229, 0.388, and 0.343, respectively. In conclusion, Cubist has high prediction accuracy and generalization ability in areas with complex topography, and the RK model can make full use of trends and spatial structural factors that are not easy to mine by ML models, which can effectively improve the prediction accuracy. This provides a reference for soil survey and digital mapping in complex terrain areas.

[Pollution Characteristics and Risk Assessment of Polycyclic Aromatic Hydrocarbons in Agricultural Soils of Different Land Use Types in Nanjing Suburbs].

In order to clarify the pollution characteristics of PAHs in suburban agricultural soils, the content of 16 types of PAHs was measured in agricultural soils with different land use types (paddy fields, vegetable fields, and forest land) in the suburbs of Nanjing. The results showed that acenaphthene (Acy) was not detected in any soil samples. The concentration of 15 PAHs in agricultural soil in suburban Nanjing ranged from 24.49 to 925.54 µg·kg-1, with an average concentration of 259.88 µg·kg-1. In different land use types, the order of PAHs concentration in soil from high to low was:forest land>paddy fields>vegetable fields, and the high-ring PAHs content was dominant in general. The effects of different soil physicochemical properties on PAHs showed that there was a certain correlation between soil organic carbon (TOC) and clay (clay) content and PAHs, whereas pH and total nitrogen (TN) had no significant correlation with PAHs. The toxic equivalence method and CSI index method were used for ecological risk assessment, which showed that the ecological risk of PAHs in agricultural soils in suburban Nanjing was relatively small; however, the ecological risk of PAHs in forest land should be given some attention, and supervision should be strengthened. Health risk assessment using incremental lifetime cancer risk (ILCR) showed that the threat to the health of children was slightly greater than that of adults, and the CR of forest land was significantly higher than that of vegetable and paddy fields, though still within an acceptable range. Uncertain health assessments were performed in adults, showing that risk analyses of deterministic health risks underestimated the health risks of PAHs. The results of sensitivity analysis showed that the input parameter that had the greatest impact on the total variance of the total carcinogenic risk CR was the exposure frequency EF (50.7%), followed by the pollutant concentration CS (43.3%).

[Vertical Pollution Characteristics and Source Analysis of Soil PAHs in Different Land Use Types].

In order to study the vertical pollution characteristics of polycyclic aromatic hydrocarbons (PAHs) in soils of different land use types in suburban areas of Nanjing, 15 types of controlled PAHs were studied in each section (0-100 cm) of soils from six different land use types, including a vegetable field, forestland, residential area, urban land, paddy field, and industrial area. The vertical distribution and composition characteristics, influencing factors, and sources of PAHs were analyzed. The results showed that:the total concentrations of Σ15PAHs in the six sampling site profiles were as follows:vegetable field (69.3-299.2 µg·kg-1), forestland (20.8-128.3 µg·kg-1), residential area (30.7-142.1 µg·kg-1), urban land (185.6-1728.7 µg·kg-1), paddy field (208.3-693.0 µg·kg-1), and industrial area (165.6-739.2 µg·kg-1). There was no pollution in the residential area or forestland and a light pollution level in the vegetable field, medium pollution level in the paddy field and industrial area, and more serious pollution in the urban land. Soil PAHs were mainly distributed in the surface or subsurface layer, except in the residential area and urban land; however, they were still detected in the deep layers, and high-molecular-weight PAHs were dominant in most depths and sampling sites. The vertical distribution and migration of PAHs in soils were affected by molecular characteristics and component concentrations of PAHs, soil physical and chemical properties, and land use types. PMF source analysis indicated that coke sources, traffic sources, and coal combustion sources from human activities were the main sources of PAHs in this study region.

[Predicting soil property in hilly regions by using landscape and multiscale micro-landform features]. / 基于景观及微地形特征的丘陵区土壤属性预测.

To assess the high-resolution digital soil mapping method for small watersheds in hilly areas, we explored the role of landscape classification and multiscale micro-landform features in predicting soil pH, soil clay content (SCC), and cation exchange capacity (CEC). Geomorphons (GM) terrain classification method was used to create landform units. The traditional digital elevation model (DEM) derivatives and remote sensing variables were employed for different combinations with landscape and micro-landform classification variables, with further compa-rison and analysis being conducted. In addition, three machine learning techniques, including support vector machine (SVM), partial least squares regression (PLSR), and random forest (RF), were used to build prediction models. The best method was then selected, and then combined with regression kriging by modeling spatial structure of the model residuals. The results showed that the application of landscape and multiscale micro-landform classification variables effectively improved the prediction accuracy of pH, SCC, and CEC by 18.8%, 8.2% and 8.7%, respectively. The map of landscape classification that contained vegetation coverage information had greater model contribution than land use data. The GM classification map with 5 m resolution was more suitable for high-precision DSM than those with lower resolution. The composite model of RF performed the best in predicting SCC, while the pH and CEC were not suitable for adding the residual regression kriging on the basis of RF model. Finally, the combination of landscape and multiscale micro-landform classification variables, DEM derivatives and remote sensing variables had the highest prediction accuracy for all the three soil properties. This result indicated that multivariable contained more effective soil information than single data source for rolling areas. The landscape variables composed of GM and surface classified data explained about 40% of the spatial variation of tested soil attributes in hilly area. Therefore, multi-resolution GM and landscape classified variables could be included into the construction of prediction model in research of soil mapping.

[Pollution Characteristics and Risk Assessment of Polycyclic Aromatic Hydrocarbons in a Suburban Farmland Soil].

In order to assess the pollution of polycyclic aromatic hydrocarbons(PAHs) in a suburban farmland soil, 29 sampling sites were collected around Nanjing according to the grid method, and the contents of 15 different PAHs were determined. Acenaphthene(Ace) was not detected in any of the samples. The total content of PAHs in farmland soil ranged from 24.49 to 750.04 µg·kg-1, with an average of 226.64 µg·kg-1. The spatial distribution of high-ring PAHs, the main PAHs in the farmland soil, was similar to that of total PAHs. There was no significant correlation between PAHs and soil organic matter(SOM), pH, cation exchange capacity(CEC), and total nitrogen(TN), whereas bulk density and low ring PAHs were significantly correlated. The results of source apportionment show that the main source of PAHs in the farmland soil is a mixture of combustion and petroleum. The contamination severity index(CSI) index shows that the PAHs does not pose an ecological risk. The results of the health risk assessment show that there is no potential carcinogenic risk to children or adults, and the main sequence of exposure is skin contact>ingestion>inhalation.