Groundwater Contamination Source Identification (GCSI) is a crucial prerequisite for conducting comprehensive pollution risk assessments, formulating effective groundwater contamination control strategies, and devising remediation plans. In previous GCSI studies, various boundary conditions were typically assumed to be known variables. However, in many practical scenarios, these boundary conditions are exceedingly complex and difficult to accurately pre-determine. This practice of presuming boundary conditions as known may significantly deviate from reality, leading to errors in identification results. Moreover, the outcomes of GCSI may be influenced by multiple factors or conditions, including the fundamental information about the contamination source boundary conditions of the polluted area. This study primarily focuses on contamination source information and unknown boundary conditions. Innovatively, three deep learning surrogate models, the Deep Belief Neural Network (DBNN), Bidirectional Long Short-Term Memory Networks (BiLSTM), and Deep Residual Neural Network (DRNN), are employed for identification and validation and to simulate the highly no-linear simulation model and directly establish a mapping relationship between the outputs and inputs of the simulation model. This approach enables the direct acquisition of the inverse identification results of the variables based on actual monitoring data, thereby facilitating rapid inverse identification. Furthermore, to account for the uncertainty of noise in monitoring data, the inversion accuracy of these three deep learning methods is compared, and the method with higher accuracy is selected for uncertainty analysis. Multiple experiments were conducted, such as accuracy identification tests, robustness tests, and cross-comparative ablation studies. The results demonstrate that all three deep learning models effectively complete the research tasks, with DBNN showing the most exceptional performance in the experiments. DBNN achieved an R² value of 0.982, an RMSE of 3.77, and an MAE of 7.56%. Subsequent uncertainty analysis, model robustness, and ablation study further affirm DBNN adaptability to GCSI research tasks.

Direct sewage discharge can cause severe damage to the water environment of the river. However, the impacts of dissolved organic matter (DOM) in the discharge on the original pattern of DOM and the distribution of heavy metals (HMs) in the river are little known. How to monitor such areas in a long-term and systematic manner also needs to be urgently addressed. In this paper, we characterized the DOM of the sediments in the WWTPs (wastewater treatment plants)-river integrated zone by ultraviolet-visible absorption spectroscopy (UV-vis), three-dimensional excitation-emission matrix (3D-EEM) combined with parallel factor (PARAFAC) method. The effects of WWTP on receiving waters were investigated, and the potential link between DOM and HM pollution was explored. Hg (I_geo: 3.94 ± 0.65; EF: 44.83 ± 31.11), Cd (I_geo: 1.81 ± 0.69; EF: 8.02 ± 2.97), Cu (I_geo: 1.61 ± 0.83; EF: 6.85 ± 2.37), Zn (I_geo: 1.55 ± 0.54; EF: 7.24 ± 3.58), and Ni (I_geo: 1.46 ± 0.56; EF: 6.12 ± 1.99) in rivers were the primary risk sources of HM. The combined pollution risk indicates that the WWTPs-river integrated area is in a high pollution risk state. Moreover, a(254) has a significant correlation with pollution indicators and can be used as a proxy indicator. These results help to understand better the impact of WWTPs on receiving water bodies and the potential connection between DOM and HM pollution and provide new ideas for monitoring the water environment in highly polluted areas.

The Songliao River Basin (SLRB) is a key agricultural region in China, and understanding precipitation variations can provide crucial support for water resource management and sustainable development. This study used CN05.1 observational data and the Coupled Model Intercomparison Project Phase 6 (CMIP6) data to simulate and evaluate the precipitation characteristics within the SLRB. The optimal model ensemble was selected for future precipitation predictions. We analyzed the historical precipitation characteristics within the SLRB and projected future precipitation variations under SSP126, SSP245, and SSP585, while exploring the driving factors influencing precipitation. The results indicated that EC-Earth3-Veg (0.507) and BCC-CSM2-MR (0.493) from MME2 effectively capture precipitation variations, with MME2 corrected data more closely matching actual precipitation characteristics. From 1971 to 2014, precipitation showed an insignificant increasing trend, with most precipitation concentrated between May and September. Precipitation in the basin decreased from southeast to northwest. From 2026 to 2100, the increasing trend in precipitation became significant. The trend of precipitation growth over time was as follows: SSP126 < SSP245 < SSP585. Future precipitation distribution resembled the historical period, but the area of semiarid regions gradually decreased while the area of humid regions gradually increased, particularly under SSP585. The long-term increase in precipitation will become more pronounced, with a significant expansion of high-precipitation areas. In low-latitude, high-longitude areas, more precipitation events were expected to occur, while the impact of altitude was relatively weaker. From SSP126 to SSP585, the response of precipitation changes to temperature changes within the SLRB shifts from negative to positive. Under SSP585, this response becomes more pronounced, with average precipitation increasing by 4.87% for every 1 °C rise in temperature.