The continuous and effective monitoring of the water quality of small rural rivers is crucial for rural sustainable development. In this work, machine learning models were established to predict the water quality of a typical small rural river based on a small quantity of measured water quality data and UAV hyperspectral images. Firstly, the spectral data were preprocessed using fractional order derivation (FOD), standard normal variate (SNV), and normalization (Norm) to enhance the spectral response characteristics of the water quality parameters. Second, a method combining the Pearson’s correlation coefficient and the variance inflation factor (PCC–VIF) was utilized to decrease the dimensionality of features and improve the quality of the input data. Again, based on the screened features, a back-propagation neural network (BPNN) model optimized using a mixture of the genetic algorithm (GA) and the particle swarm optimization (PSO) algorithm was established as a means of estimating water quality parameter concentrations. To intuitively evaluate the performance of the hybrid optimization algorithm, its prediction accuracy is compared with that of conventional machine learning algorithms (Random Forest, CatBoost, XGBoost, BPNN, GA–BPNN and PSO–BPNN). The results show that the GA–PSO–BPNN model for turbidity (TUB), ammonia nitrogen (NH₃-N), total nitrogen (TN), and total phosphorus (TP) prediction exhibited optimal accuracy with coefficients of determination (R²) of 0.770, 0.804, 0.754, and 0.808, respectively. Meanwhile, the model also demonstrated good robustness and generalization ability for data from different periods. In addition, we used this method to visualize the water quality parameters in the study area. This work provides a new approach to the refined monitoring of water quality in small rural rivers.

Polychlorinated diphenyl ethers (PCDEs) are a class of synthetic halogenated aromatic compounds, which have gradually attracted widespread attention due to potential environmental risks to humans and ecosystems. This paper presents a literature review of research on PCDEs using PubMed, Web of Science and Google Scholar as search engines/databases with no constraints on publishing year or number. A total of 98 publications on the sources, environmental levels, environmental behavior and fate, synthesis and analysis and toxicology of PCDEs were retrieved. Existing studies have shown that PCDEs widely exist in the environment with the ability of long-range transport, bioaccumulation and biomagnification, which are almost comparable to polychlorinated biphenyls. They can elicit adverse effects including hepatic oxidative stress, immunosuppression, endocrine disorders, growth retardation, malformations, reduced fertility and increased mortality in organisms, among which some seem to be related to the activation of the aryl hydrocarbon receptor. PCDEs can be metabolized into other organic pollutants, such as hydroxylated and methoxylated PCDEs and even polychlorinated dibenzo-p-dioxins and furans through biotransformation, photolysis and pyrolysis reactions in the environment. Compared with reviews on PCDEs published previously, some new information and findings are summarized in this review, such as new sources, current environmental exposure levels, main metabolism pathways in aquatic organisms, acute toxicity data for more species and relationships between structural parameters and toxicity and bioaccumulation potentials of PCDE congeners. Finally, current research deficiencies and future research perspectives are proposed to facilitate the assessment of health and ecological risks of PCDEs.

Despite growing concerns over antibiotic pollution, studies on the effects of low-level antibiotic exposure on plant growth remain limited. This study investigated the bioaccumulation and adverse effects of three typical antibiotics, enrofloxacin (ENR), sulfamethazine (SM2) and oxytetracycline (OTC), on lettuce (Lactuca sativa L.) at environmentally relevant concentrations (1, 10 and 100 µg/L). The results indicate that antibiotics were minimally accumulated in lettuce tissue (roots and leaves), with their bioconcentration factors (BCFs), root concentration factors (RCFs) and transfer factors (TFs) ranging from 0.05 to 10.35 L/kg, 1.14 to 236 L/kg, and 0.004 to 0.207, respectively. For each antibiotic, the BCF, RCF, and TF values generally decreased with increasing exposure concentrations (except for TF following OTC exposure). All three antibiotic treatments disrupted photosynthesis, reduced photosynthetic rates, triggered oxidative stress, and resulted in growth inhibition. Moreover, soluble protein and sugar contents declined, while antioxidant enzyme activities and lipid peroxidation products increased. In addition, microscopic observations revealed cellular structural damage, and elevated reactive oxygen species indicated intensified stress responses. Furthermore, transcriptomic analysis combined with functional enrichment showed that genes involved in photosynthetic processes and antioxidant defense were significantly downregulated with rising antibiotic concentrations, with KEGG pathway analysis highlighting the inhibition of key pathways related to energy metabolism and stress response.

Faced with the intertwined challenges of climate change, pollution dispersion, ecosystem degradation, and resource constraints, the global environment exhibits high nonlinearity, multiscale coupling, and regional variability (Zhu et, al., 2023).Traditional experimental methods and statistical tools are often constrained by high costs, reliance on complex instrumentation, and an inability to capture intrinsic dynamics or provide interpretable decision support across large spatiotemporal scales.In recent years, computational modeling integrated with modern techniques, such as artificial intelligence, machine learning, molecular simulation, and computational fluid dynamics, has offered new ways to address environmental complexities (Jeong and Choi, 2022;Liu et, al., 2022). These tools help assess toxicity and risks, identify pollution sources, and elucidate pollutant behavior across media. Despite significant progress, most current applications rely on off-the-shelf algorithms, while reusable models designed specifically for environmental problems remain scarce (Alotaibi and Nassif, 2024). This limits the discovery of underlying mechanisms and hinders the practical application of models in environmental management and pollution control. Moreover, the "black-box" nature of environmental input-output relationships makes model interpretation difficult (Zhu et, al., 2023;Xu et, al., 2024). Hence, this special issue focuses on three core questions: how to improve the cost-effectiveness and practical applicability of environmental solutions via computational modeling; how to integrate data-driven approaches with domain knowledge for mechanistic interpretation of environmental processes; and how to develop practical frameworks that handle missing data and spatial heterogeneity.This Research Topic features ten articles, including both original research and reviews, on computational modeling applied to pollutant source apportionment, environmental behavior, toxicity, risk, and removal. All of them address diverse environmental problems with a problem-driven modeling approach and provide practical technical solutions. Third, traditional source apportionment methods have long been limited to qualitative or semi-quantitative estimates of source contributions, making it difficult to directly link them to ecological or health risks. Two studies improved the methodology at regional and site scales. Zhai conducted a meta-analysis of volatile organic compounds (VOC) emissions and ozone formation potential (OFP) across 14 cities in the Pearl River Delta, Yangtze River Delta, and Bohai Rim. The study revealed that oxygenated VOCs dominate emissions (44.6%-81.1%), with isopropanol and ethyl acetate as main compounds; OFP ranged from 78.5 to 644 mg/m 3 , highest in Changsha. About 68% of enterprises use inefficient technologies (e.g., granular activated carbon), while efficient regenerative thermal oxidizers are not widely adopted due to cost and policy constraints. This work provides a scientific basis for region-specific VOC reduction and ozone control. Liang et al. focused on a decommissioned coking site, building a quantitative source-risk framework with GIS, positive matrix factorization (PMF), and Monte Carlo simulation. Results showed that ecological risks were mainly from coking industry Hg emissions, while health risks were driven by Cd and As from agricultural activities. The framework demonstrated that "high mass contributions do not equal high risks", offering guidance for environmental management and safe urban renewal.Finally, environmental monitoring and forecasting rely on data completeness, regional representativeness, and the ability to capture complex dynamics. Four studies addressed missing data interpolation, forecasting under regional heterogeneity, and In summary, the articles collected under this research topic highlight and elucidate the role of computational modeling in developing low-cost detection methods, energy-saving remediation strategies, quantitative source-risk frameworks, and reliable forecasting tools for environmental monitoring. This research topic has positive implications for deepening the understanding of complex environmental processes and for developing practical, mechanism-driven modeling solutions that support environmental management and pollution control.