Search

[Display omitted] • A machine learning method assisted by auto hyperparameter optimization was developed. • TPE-XGBoost model has the best performance for spatial prediction of soil heavy metals. • Air quality and hyperspectral variables have the high contributions for prediction accuracy. • There is a source-receptor coupling path for the accumulation of soil heavy metals. • Soil heavy metals with high concentrations are concentrated around rivers. Machine learning (ML) models have been widely used for predicting spatial variability of soil heavy metals. However, it is impossible to explore the entire hyperparameter space of ML models by artificially trial-and-error experimentation. Here, an auto hyperparameter optimization-based machine learning (HPO-ML) method with three search algorithms and random forest (RF) and extreme gradient boosting (XGBoost) models was developed to predict the heavy metal content in soil with multiple environmental variables. The tree-structured Parzen estimator (TPE) algorithm outperformed other search algorithms in identifying the optimal hyperparameters of RF and XGBoost models. The model prediction results showed that the TPE-XGBoost had the highest accuracy for predicting the As (RMSE = 3.06 mg kg−1 and R2 = 70.35%), Cd (RMSE = 0.10 mg kg−1 and R2 = 75.43%), Cr (RMSE = 13.86 mg kg−1 and R2 = 82.11%), Ni (RMSE = 3.19 mg kg−1 and R2 = 75.20%), Pb (RMSE = 3.75 mg kg−1 and R2 = 74.79%), and Zn (RMSE = 6.83 mg kg−1 and R2 = 70.05%) contents. The TPE-XGBoost mapping result showed that areas with high concentrations of soil heavy metals were concentrated in the central and eastern areas (As), the mainstream of the Yellow River (Cd), the northeast area (Cr), the ancient watercourse of the Yellow River (Ni and Pb), and the central and northeastern areas (Zn). The SHapley additive explanation (SHAP) and structural equation model (SEM) were used to interpret the drivers of environmental variables. It is found that the variables with the highest contributions were CO, PM 2.5 , O 3 , PC3, PC1, and PC4 for predicting the As, Cd, Cr, Ni, Pb, and Zn contents, respectively, and there was a significant source-receptor coupling path. The results demonstrate the feasibility of using the HPO-ML approach in hyperparameter-limited conditions, which providing data-driven pathways and options to support the high-quality development of agriculture and the protection of farmland soil ecosystem. [ABSTRACT FROM AUTHOR]

Lanzhou New District is the first and largest national-level new district in the Loess Plateau region of China. Large-scale land creation and rapid utilization of the land surface for construction has induced various magnitudes of land subsidence in the region, which is posing an increasing threat to the built environment and quality of life. In this study, the spatial and temporal evolution of surface subsidence in Lanzhou New District was assessed using Persistent Scatterer Interferometric Synthetic Aperture radar (PSInSAR) to process the ENVISAT SAR images from 2003-2010, and the Small Baseline Subset (SBAS) InSAR to process the Sentinel-1A images from 2015-2016. We found that the land subsidence exhibits distinct spatiotemporal patterns in the study region. The spatial pattern of land subsidence has evidently extended from the major urban zone to the land creation region. Significant subsidence of 0-55 mm/year was detected between 2015 and 2016 in the land creation and urbanization area where either zero or minor subsidence of 0-17.2 mm/year was recorded between 2003 and 2010. The change in the spatiotemporal pattern appears to be dominated mainly by the spatial heterogeneity of land creation and urban expansion. The spatial associations of subsidence suggest a clear geological control, in terms of the presence of compressible sedimentary deposits; however, subsidence and groundwater fluctuations are weakly correlated. We infer that the processes of land creation and rapid urban construction are responsible for determining subsidence over the region, and the local geological conditions, including lithology and the thickness of the compressible layer, control the magnitude of the subsidence process. However, anthropogenic activities, especially related to land creation, have more significant impacts on the detected subsidence than other factors. In addition, the higher collapsibility and compressibility of the loess deposits in the land creation region may be the underlying mechanism of macro-subsidence in Lanzhou New District. Our results provide a useful reference for land creation, urban planning and subsidence mitigation in the Loess Plateau region, where the large-scale process of bulldozing mountains and valley infilling to create level areas for city construction is either underway or forthcoming. [ABSTRACT FROM AUTHOR]

Air quality is a vital environment variable which determines spatial accumulation of soil heavy metals. It is very important to estimate the contribution of air quality for soil heavy metals in oil mining area. For the end, we collected 116 samples from surface soil of oil mining in the Yellow River Delta (YRD) of China, and analyzed the content of As, Cr, Ni, Pb, and Zn. Furthermore, 40 monitoring stations data of air quality were collected in study area, including CO, NO 2 , SO 2 , O 3 , PM 2.5 , and PM 10. Spatial estimation and mapping of heavy metals in soil were carried out by hybrid geostatistical models, including multiple linear regression-ordinary kriging (MLROK), support vector machine-ordinary kriging (SVMOK) and random forest-ordinary kriging (RFOK). RFOK exhibited the highest estimation accuracy (R2) for As (65.76%), Cr (77.85%), Ni (61.47%), Pb (74.64%), and Zn (71.35%) in comparison with other models. And relative R2 of RFOK improved 30%, while MLROK and SVMOK increased over 100% for Zn (RI o = 121.90% and RI o = 121.64%) based on their original R2 of machine learning models. In addition, mapping results by RFOK showed the high concentrations of heavy metals were focused in the central and northeastern (As), northern (Cr), northeastern and northwestern (Ni), central and eastern (Pb), and northern (Zn). Especially, compared with vegetation index and topographic factors, PM 2.5 is the highest driving variable for As (18.34%) and Zn (12.91%), and CO is the most important variable for Cr (18.22%), Ni (14.28%). The above results indicated that there is a mechanism of sources-receptor relationship between air quality and soil heavy metals, that is, oil well and factory in study area discharge heavy metal particles into the atmosphere, and then enter the soil through atmospheric deposition and precipitation. Enlightened by this study, variable selection should be focused on important sources for the accumulation of heavy metals in study area, who must take decisions to prevent and to early warn heavy metals pollution in mine soil. [Display omitted] • Air quality data are used to predict the content of heavy metals in oil mining soils. • Random forest-ordinary kriging has the best predicted performance for soil heavy metals. • The high concentrations of soil heavy metals are the northern of study area. • Air quality variables have effective feasibility for predicting soil heavy metals. • PM (for As and Zn) and CO (for Cr and Ni) is the highest driving variable. [ABSTRACT FROM AUTHOR]