Search

Landslides pose a constant threat to the lives and property of mountain people and may also cause geomorphological destruction such as soil and water loss, vegetation destruction, and land cover change. Landslide susceptibility assessment (LSA) is a key component of landslide risk evaluation. There are many related studies, but few analyses and comparisons of models for optimization. This paper aims to introduce the Tree-structured Parzen Estimator (TPE) algorithm for hyperparameter optimization of three typical neural network models for LSA in Shuicheng County, China, as an example, and to compare the differences of predictive ability among the models in order to achieve higher application performance. First, 17 influencing factors of landslide multiple data sources were selected for spatial prediction, hybrid ensemble oversampling and undersampling techniques were used to address the imbalanced sample and small sample size problem, and the samples were randomly divided into a training set and validation set. Second, deep neural network (DNN), recurrent neural network (RNN), and convolutional neural network (CNN) models were adopted to predict the regional landslides susceptibility, and the TPE algorithm was used to optimize the hyperparameters respectively to improve the assessment capacity. Finally, to compare the differences and optimization effects of these models, several objective measures were applied for validation. The results show that the high-susceptibility regions mostly distributed in bands along fault zones, where the lithology is mostly claystone, sandstone, and basalt. The DNN, RNN, and CNN models all perform well in LSA, especially the RNN model. The TPE optimization significantly improves the accuracy of the DNN and CNN (3.92% and 1.52%, respectively), but does not improve the performance of the RNN. In summary, our proposed RNN model and TPE-optimized DNN and CNN model have robust predictive capability for landslide susceptibility in the study area and can also be applied to other areas containing similar geological conditions.

In this study, Yulin city and Yan’an city in northern Shaanxi Province were taken as the study area. Based on the diurnal dry–wet events abrupt alternation index DWAAI, the joint probability distribution of two characteristic variables of “urgency” and “alternation” of dry–wet events abrupt alternation was established by using copula function, and the characteristics of dry–wet events abrupt alternation were analyzed. DWAAI was calculated from daily precipitation data and the applicability of the index was verified. On this basis, the two characteristic variables of “urgency” and “alternation” were separated, and the appropriate marginal distribution function was selected to fit them, and the correlation between the two variables was evaluated. Finally, the appropriate copula function was selected to fit the bivariate of each station, and the joint cumulative probability and recurrence period of the two variables were calculated. The results show that the DWAAI index is suitable for the identification of dry–wet events abrupt alternation in the study area. Light and moderate dry–wet events abrupt alternation occurs more frequently, while severe events rarely occur in the study area. The frequency of severe dry–wet events abrupt alternation in Jingbian station and its northern area is greater than that in the southern area, and the risk of dry–wet events abrupt alternation of disasters in the northern area is higher. The greater the degree of “urgency” and “alternation”, the greater the joint cumulative probability and the greater the return period. The return period of severe dry–wet events abrupt alternation was more than five years, while the return period of light and moderate dry–wet events abrupt alternation was less than five years.

Land use change is an important driving force factor affecting the river water environment and directly affecting water quality. To analyze the impact of land use change on water quality change, this study first analyzed the land use change index of the study area. Then, the study area was divided into three subzones based on surface runoff. The relationship between the characteristics of land use change and the water quality grade was obtained by grey correlation analysis. The results showed that the land use types changed significantly in the study area since 2000, and water body and forest land were the two land types with the most significant changes. The transfer rate is cultivated field &gt, forest land &gt, construction land &gt, grassland &gt, unused land &gt, water body. The entropy value of land use information is represented as Area I &gt, Area III &gt, Area II. The shift range of gravity center is forest land &gt, water body &gt, cultivated field. There is a strong correlation between land use change index and water quality, which can be improved and managed by changing the land use type. It is necessary to establish ecological protection areas or functional areas in Area I, artificial lawns or plantations shall be built in the river around the water body to intercept pollutants from non-point source pollution in Area II, and scientific and rational farming in the lower reaches of rivers can reduce non-point source pollution caused by farming.

Tea trees are the main economic crop in Zhejiang Province. However, spring cold is a frequent occurrence there, causing frost damage to the valuable tea buds. To address this, a regional frost-hazard early-warning system is needed. In this study, frost damage area was estimated based on topography and meteorology, as well as longitude and latitude. Based on support vector machine (SVM) and artificial neural networks (ANNs), a multi-class classification model was proposed to estimate occurrence of regional frost disasters using tea frost cases from 2017. Results of the two models were compared, and optimal parameters were adjusted through multiple iterations. The highest accuracies of the two models were 83.8% and 75%, average accuracies were 79.3% and 71.3%, and Kappa coefficients were 79.1% and 67.37%. The SVM model was selected to establish spatial distribution of spring frost damage to tea trees in Zhejiang Province in 2016. Pearson’s correlation coefficient between prediction results and meteorological yield was 0.79 (p &lt, 0.01), indicating consistency. Finally, the importance of model factors was assessed using sensitivity analysis. Results show that relative humidity and wind speed are key factors influencing accuracy of predictions. This study supports decision-making for hazard prediction and defense for tea trees facing frost.

Plant phenology depends largely on temperature, but temperature alone cannot explain the Northern Hemisphere shifts in the start of the growing season (SOS). The spatio–temporal distribution of SOS sensitivity to climate variability has also changed in recent years. We applied the partial least squares regression (PLSR) method to construct a standardized SOS sensitivity evaluation index and analyzed the combined effects of air temperature (Tem), water balance (Wbi), radiation (Srad), and previous year’s phenology on SOS. The spatial and temporal distributions of SOS sensitivity to Northern Hemisphere climate change from 1982 to 2014 were analyzed using time windows of 33 and 15 years; the dominant biological and environmental drivers were also assessed. The results showed that the combined sensitivity of SOS to climate change (SCom) is most influenced by preseason temperature sensitivity. However, because of the asymmetric response of SOS to daytime/night temperature (Tmax/Tmin) and non-negligible moderating of Wbi and Srad on SOS, SCom was more effective in expressing the effect of climate change on SOS than any single climatic factor. Vegetation cover (or type) was the dominant factor influencing the spatial pattern of SOS sensitivity, followed by spring temperature (Tmin > Tmax), and the weakest was water balance. Forests had the highest SCom absolute values. A significant decrease in the sensitivity of some vegetation (22.2%) led to a decreasing trend in sensitivity in the Northern Hemisphere. Although temperature remains the main climatic factor driving temporal changes in SCom, the temperature effects were asymmetric between spring and winter (Tems/Temw). More moisture might mitigate the asymmetric response of SCom to spring/winter warming. Vegetation adaptation has a greater influence on the temporal variability of SOS sensitivity relative to each climatic factor (Tems, Temw, Wbi, Srad). More moisture might mitigate the asymmetric response of SCom to spring/winter warming. This study provides a basis for vegetation phenology sensitivity assessment and prediction.

Among the most frequent and dangerous natural hazards, landslides often result in huge casualties and economic losses. Landslide susceptibility mapping (LSM) is an excellent approach for protecting and reducing the risks by landslides. This study aims to explore the performance of Bayesian optimization (BO) in the random forest (RF) and gradient boosting decision tree (GBDT) model for LSM and applied in Shuicheng County, China. Multiple data sources are used to obtain 17 conditioning factors of landslides, Borderline-SMOTE and Randomundersample methods are combined to solve the imbalanced sample problem. RF and GBDT models before and after BO are adopted to calculate the susceptibility value of landslides and produce LSMs and these models were compared and evaluated using multiple validation approach. The results demonstrated that the models we proposed all have high enough model accuracy to be applied to produce LSM, the performance of the RF is better than the GBDT model without BO, while after adopting the Bayesian optimized hyperparameters, the prediction accuracy of the RF and GBDT models is improved by 1% and 7%, respectively and the Bayesian optimized GBDT model is the best for LSM in this four models. In summary, the Bayesian optimized RF and GBDT models, especially the GBDT model we proposed for landslide susceptibility assessment and LSM construction has a very good application performance and development prospects.

Soil ecological risk caused by compound pollutants is a topic that deserves increasing attention, and soil risk early warning is a more in-depth discussion on this topic. In this study, we collected soil samples from Changchun, a typical industrial city, and determined the contents of 13 heavy metals (HMs) (0.00 mg kg−1-6380 mg kg−1), 16 polycyclic aromatic hydrocarbons (PAHs) (0.00 mg kg−1-27.7 mg kg−1), 7 polychlorinated biphenyls (PCBs) (0.30 μg kg−1-168 μg kg−1), and 8 organochlorine pesticides (OCPs) (0.00 mg kg−1-4.52 mg kg−1). The soil ecological risks of compound pollutants were assessed. The results showed that PAHs were the greatest risk pollutants, followed by PCBs and HMs, and OCPs were the smallest risk pollutants. Most of the ecological risks of compound pollutants were classified as “moderate severity” level according to the (contamination severity index) CSI evaluation criteria. With the help of modern industrial economic theory, through the analysis of the annual accumulation of pollutants, it is possible to predict the future pollutant content in Changchun, and the soil risks could be forewarned. The results showed that if active measures were not taken to reduce the accumulation of PAHs in Changchun soil, the CSI-PAHs would be classified as “ultra-high severity” level in 2035.

Debris flow events often pose significant damage and are a threat to infrastructure and even livelihoods. Recent studies have mainly focused on determining the susceptibility of debris flow using deterministic or heuristic/probabilistic models. However, each type of model has its own significant advantages with some irreparable disadvantages. The random forest model, which is sensitive to the region where the terrain conditions are suitable for the occurrence of debris flow, was applied along with the steady-state infinite slope method, which is capable of describing the initiation mechanism of debris flow. In this manner, a random-forest-based steady-state infinite slope method was used to conduct susceptibility assessment of debris-flow at Changbai mountain area. Results showed that the assessment accuracy of the proposed random-forest-based steady-state infinite slope method reached 90.88%, however, the accuracy of just the random forest model or steady-state infinite slope method was only 88.48% or 60.45%, respectively. Compared with the single-model assessment results, the assessment accuracy of the proposed method improved by 2.4% and 30.43%, respectively. Meanwhile, the debris-flow-prone area of the proposed method was reduced. The random-forest-based steady-state infinite slope method inherited the excellent diagnostic performance of the random-forest models in the region where the debris flow disaster already occurred, meanwhile, this method further refined the debris-flow-prone area from the suitable terrain area based on physico-mechanical properties, thus, the performance of this method was better than those of the other two models.

Precipitation is low and drought occurs frequently in Northern Shaanxi. To study the characteristics and occurrence and development of drought events in Northern Shaanxi is beneficial to the prevention and control of drought disasters. Based on the monthly rainfall data of 10 meteorological stations in Northern Shaanxi from 1960 to 2019, the characteristic variables of drought events at each meteorological station in Northern Shaanxi were extracted by using run theory and copula function. The joint probability distribution and recurrence period were obtained by combining the duration and intensity of drought, and the relationship between drought characteristics and crop drought affected area was studied. The results show that (1) from 1960 to 2019, drought events mainly occurred in Northern Shaanxi with long duration and low severity, short duration and high severity, or short duration and low severity, among which the frequency of drought events that occurred in Yuyang and Baota districts was higher. The frequency of light drought and extreme drought was more in the south and less in the north, while the frequency of moderate drought and severe drought was more in the north and less in the south. (2) The optimal edge distribution of drought intensity and drought duration in Northern Shaanxi is generalized Pareto distribution, and the optimal fitting function is Frank copula function. The greater the duration and intensity of drought, the greater the cumulative probability and return period. (3) The actual recurrence interval and the theoretical recurrence interval of drought events in Northern Shaanxi were close, and the error was only 0.1–0.3a. The results of the joint return period can accurately reflect the actual situation, and this study can provide effective guidance for the prevention and management of agricultural dryland in Northern Shaanxi.

Landslides are among the most frequent natural hazards in the world. Rainfall is an important triggering factor for landslides and is responsible for topples, slides, and debris flows&mdash, three of the most important types of landslides. However, several previous relevant research studies covered general landslides and neglected the rainfall&ndash, topples&ndash, slides&ndash, debris flows disaster chain. Since landslide hazard mapping (LHM) is a critical tool for disaster prevention and mitigation, this study aimed to build a GeoDetector and Bayesian network (BN) model framework for LHM in Shuicheng County, China, to address these geohazards. The GeoDetector model will be used to screen factors, eliminate redundant information, and discuss the interaction between elements, while the BN model will be used for constructing a causality disaster chain network to determine the probability and risk level of the three types of landslides. The practicability of the BN model was confirmed by error rate and scoring rules validation. The prediction accuracy results were tested using overall accuracy, Matthews correlation coefficient, relative operating characteristics curve, and seed cell area index. The proposed framework is demonstrated to be sufficiently accurate to construct the complex LHM. In summary, the combination of the GeoDetector and BN model is very promising for spatial prediction of landslides.

Residents in industrial cities may be exposed to potentially toxic elements (PTEs) in soil that increase chronic disease risks. In this study, six types of PTEs (Zn, As, Cr, Ni, Cu, and Pb) in 112 surface soil samples from three land use types&mdash, industrial land, residential land, and farmland&mdash, in Tonghua City, Jilin Province were measured. The geological accumulation index and pollution load index were calculated to assess the pollution level of metal. Meanwhile, the potential ecological risk index, hazard index, and carcinogenic risk were calculated to assess the environmental risks. The spatial distribution map was determined by the ordinary kriging method, and the sources of PTEs were identified by factor analysis and cluster analysis. The average concentrations of Zn, As, Cr, Ni, Cu, and Pb were 266.57, 15.72, 72.41, 15.04, 20.52, and 16.30 mg/kg, respectively. The results of the geological accumulation index demonstrated the following: Zn pollution was present in all three land use types, As pollution in industrial land cannot be neglected, Cr pollution in farmland was higher than that in the other two land use types. The pollution load index decreased in the order of industrial land &gt, farmland &gt, residential land. Multivariate statistical analysis divided the six PTEs into three groups by source: Zn and As both originated from industrial activities, vehicle emissions were the main source of Pb, and Ni and Cu were derived from natural parent materials. Meanwhile, Cr was found to come from a mixture of artificial and natural sources. The soil environment in the study area faced ecological risk from moderate pollution levels mainly contributed by As. PTEs did not pose a non-carcinogenic risk to humans, however, residents of the three land use types all faced estimated carcinogenic risks caused by Cr, and As in industrial land also posed high estimated carcinogenic risk to human health. The conclusion of this article provides corresponding data support to the government's policy formulation of remediating different types of land and preventing exposure and related environmental risks.