Vehicle emission is a major source of atmospheric secondary organic aerosols (SOA). Driving condition is a critical influencing factor for vehicular SOA production, but few studies have revealed the dependence on rapid-changing real-world driving conditions. Here, a fast-response oxidation flow reactor system is developed and deployed to quantify the SOA formation potential under transient driving conditions. Results show that the SOA production factor varies by orders of magnitude, e.g., 20–1500 mg kg-fuel−1 and 12–155 mg kg-fuel−1 for China V and China VI vehicles, respectively. High speed, acceleration, and deceleration are found to considerably promote SOA production due to higher organic gaseous emissions caused by unburned fuel emission or incomplete combustion. In addition, China VI vehicles significantly reduce SOA formation potential, yield, and acceleration and deceleration peaks. Our study provides experimental insight and parameterization into vehicular SOA formation under transient driving conditions, which would benefit high time-resolved SOA simulations in the urban atmosphere.

No abstract

There are many potential hazard sources along high-speed railways that threaten the safety of railway operation. Traditional ground search methods are failing to meet the needs of safe and efficient investigation. In order to accurately and efficiently locate hazard sources along the high-speed railway, this paper proposes a texture-enhanced ResUNet (TE-ResUNet) model for railway hazard sources extraction from high-resolution remote sensing images. According to the characteristics of hazard sources in remote sensing images, TE-ResUNet adopts texture enhancement modules to enhance the texture details of low-level features, and thus improve the extraction accuracy of boundaries and small targets. In addition, a multi-scale Lovász loss function is proposed to deal with the class imbalance problem and force the texture enhancement modules to learn better parameters. The proposed method is compared with the existing methods, namely, FCN8s, PSPNet, DeepLabv3, and AEUNet. The experimental results on the GF-2 railway hazard source dataset show that the TE-ResUNet is superior in terms of overall accuracy, F1-score, and recall. This indicates that the proposed TE-ResUNet can achieve accurate and effective hazard sources extraction, while ensuring high recall for small-area targets.

Based on the records of people’s political inquiries, comments from public sources on the Internet and the data of the relevant departments’ responses to some people’s messages, this study uses text analysis, text feature extraction, model building, text mining, and other evaluation methods to study and evaluate the three aspects of government services: analysis of public comments, mining of hot issues and evaluation of replies, which aims to prompt the government to understand the needs of the people quickly and solve the relevant problems in a timely and effective manner. The results show that the final classification accuracy using BERT is 3.4% and 1.8% higher than that using TF-IDF and Word2vec, respectively. Multi-classification of message data was realized by BERT combined with the LinearSVC algorithm, and the crowd message was accurately divided into seven types of problems, with an accuracy of 96.7%. It is intended to be transferred to relevant departments for processing. For problems related to people’s livelihood, law, economy, and other aspects, different departments should take countermeasures to solve them and achieve systematic, departmental, and regional coordination. This will enhance the ability of government platforms to deal with problems. Through the definition of hot indexes, hot issues mining can timely find the outstanding problems reflected by the masses. At the same time, the feedback evaluation system can comprehensively evaluate the work of relevant departments from the perspectives of relevance, completeness, and interpretability. Big data analysis technology based on text mining is a feasible way to solve the difficulties of text data analysis. The analysis model constructed in this study is suitable for mining and analyzing unstructured data with short text features, and the results can provide guidance for government decision-making.

No abstract