Search

The evolution of hydrochemical compositions influenced by long-period interactions between groundwater and the geo-environment is a fundamental issue for exploring groundwater quality and vulnerability. This study systematically investigated the hydrochemical processes and anthropogenic interference occurring in the river basin by bivariate plots, Gibbs diagrams, saturation index, and the major ions ratios. Apparent changes in groundwater hydrochemistry have been observed in the study area, illustrating the origins of major ions are affected by various internal and external factors. Results highlighted that TDS varied from freshwater to brackish water, ranging between 187.90 and 2294.81 mg/L. Ca2+ and HCO 3 − are the dominant ions in the studied samples. The results gained by Gibbs diagrams, bivariate plots, saturation index, and the major ions ratios demonstrated that minerals dissolution/precipitation, cation exchange, and human inputs play crucial roles in the unconfined aquifers. Moreover, the overuse of nitrogen fertilizer, livestock manure, and industrial/domestic sewage led to nitrate and nitrite contamination and brought significant challenges to the surrounding hydrogeo-environment. The present study could make an unambiguous identification of natural processes and anthropogenic interventions influencing groundwater hydrochemistry’s long-period evolution and create a preliminary strategy for groundwater resources management.

The quantification of trade-offs between social-economic and environmental effects is of great importance, especially in the semi-arid coastal areas with highly developed agriculture. The study presents an integrated multi-objective simulation-optimization (S-O) framework to evaluate the basin-scale water-environment-agriculture (WEA) nexus. First, the variable-density groundwater model (SEAWAT) is coupled to the reactive transport model (RT3D) for the first attempt to simulate the environmental effects subject to seawater intrusion (SWI) and nitrate pollution (NP). Then, the surrogate assisted multi-objective optimization algorithm is utilized to investigate the trade-offs between the net agricultural benefits and extents of SWI and NP while considering the water supply, food security, and land availability simultaneously. The S-O modeling methodology is applied to the Dagu River Basin (DRB), a typical SWI region with intensive agricultural irrigation in China. It is shown that the three-objective space based on Pareto-optimal front can be achieved by optimizing planting area in the irrigation districts, indicating the optimal evolution of the WEA nexus system. The Pareto-optimal solutions generated by multi-objective S-O model are more realistic and pragmatic, avoiding the decision bias that may often lead to cognitive myopia caused by the low-dimensional objectives. Although the net agricultural benefits in Pareto-optimal solutions are declined to some extent, the environmental objectives (the extents of SWI and NP) are improved compared to those in the pre-optimized scheme. Therefore, the proposed multi-objective S-O framework can be applied to the WEA nexus in the river basin with intensive agriculture development, which is significant to implement the integrated management of water, food, and environment, especially for the semi-arid coastal aquifers.

Deep Neural Network(DNN) techniques have been prevalent in software engineering. They are employed to facilitate various software engineering tasks and embedded into many software applications. However, because DNNs are built upon a rich data-driven programming paradigm that employs plenty of labeled data to train a set of neurons to construct the internal system logic, analyzing and understanding their behaviors becomes a difficult task for software engineers. In this paper, we present an instance-based visualization tool for DNN, namely NeuralVis, to support software engineers in visualizing and interpreting deep learning models. NeuralVis is designed for: 1). visualizing the structure of DNN models, i.e., neurons, layers, as well as connections; 2). visualizing the data transformation process; 3). integrating existing adversarial attack algorithms for test input generation; 4). comparing intermediate layers' outputs of different inputs. To demonstrate the effectiveness of NeuralVis, we design a task-based user study involving ten participants on two classic DNN models, i.e., LeNet and VGG-12. The result shows NeuralVis can assist engineers in identifying critical features that determine the prediction results. Video: https://youtu.be/solkJri4Z44

A clear identification of the hydrochemical processes in groundwater systems is critical for the management and protection of alternative water resources. The Dagu River Basin (DRB) has become one of the most concerning areas due to the typical geographical conditions and intensive human disturbance. In this study, the spatio-temporal variations of groundwater hydrochemical characteristics were discriminated based on the self-organizing maps (SOM) clustering results from 2001 to 2017. Simultaneously, their main governing factors were also identified according to the composite properties of hydrogen and oxygen isotopic evidence, hydrochemical facies evolution diagram (HFE-D), chloro-alkaline indices (CAI) values, and the enrichment of nitrogen. Analysis results indicated that the long-term evolution of hydrochemical signatures basically conformed to the trends of rising in the dry season and falling in the rainy season. Total dissolved solids (TDS) fluctuated from 187.90 mg/L to 2294.81 mg/L, with an average of 881.77 mg/L. Moreover, the nitrate concentrations in the river basin varied significantly from 0.70 mg/L to 605.32 mg/L. The unconfined aquifer beneath DRB was experiencing freshening gradually, while some other samples were still subjected to the interference of environmental pollution. Spatio-temporal variations in the respective clusters demonstrated that natural factors including precipitation infiltration, evaporation, and water–rock interactions played vital roles in the samples of Clusters I-III, V, and VIII, while the samples of Cluster VII were influenced by seawater intrusion; and those of Clusters IV and VI were mainly affected by nitrate and nitrite contamination. This study suggests that the SOM connected with stable isotope analysis, HFE-Diagram, CAI values, and correlation analysis could be successfully used to interpret nonlinear and high-dimensional multivariate systems and provide supplementary information about the dominant mechanisms controlling the regional groundwater evolution.