Plastic-related industrial discharge is suspected as a significant source of microplastics (MPs) in the influent of wastewater treatment plants (WWTPs). However, little is known about the characteristics of MPs in industrial wastewater. Taking the Haicang WWTP in Xiamen, China, as an example, this study compared MPs in industrial wastewater with the domestic one in terms of abundance, particle size, polymer type, shape and color. Wentworth modulus, grain size parameters and principal component analysis (PCA) were performed to describe the MP difference between those two. It was found that the abundance of MPs in industrial wastewater was more than twice that in domestic wastewater, and the flux of MPs discharged into the aquatic environment through industrial wastewater was about 3.2 times that of domestic wastewater. The main shapes of MPs in industrial wastewater and domestic wastewater were fiber and granule, respectively. The proportion of polyester (PES) and polyethylene terephthalate (PET) in industrial wastewater was higher than that in domestic wastewater, related to the type of factories served by the WWTP. Compared with domestic wastewater, the rough surface of MPs in industrial wastewater was more complex and diverse, which might have a high capability of adsorbing other pollutants, thereby causing more significant harm to the environment. Our results supported that industrial sources of MPs are the priority areas in environmental management, and immediate action is taken to prevent industrial-sources MPs from entering the environment.

Semi-supervised learning is a learning pattern that can utilize labeled data and unlabeled data to train deep neural networks. In semi-supervised learning methods, self-training-based methods do not depend on a data augmentation strategy and have better generalization ability. However, their performance is limited by the accuracy of predicted pseudo-labels. In this paper, we propose to reduce the noise in the pseudo-labels from two aspects: the accuracy of predictions and the confidence of the predictions. For the first aspect, we propose a similarity graph structure learning (SGSL) model that considers the correlation between unlabeled and labeled samples, which facilitates the learning of more discriminative features and, thus, obtains more accurate predictions. For the second aspect, we propose an uncertainty-based graph convolutional network (UGCN), which can aggregate similar features based on the learned graph structure in the training phase, making the features more discriminative. It can also output the uncertainty of predictions in the pseudo-label generation phase, generating pseudo-labels only for unlabeled samples with low uncertainty; thus, reducing the noise in the pseudo-labels. Further, a positive and negative self-training framework is proposed, which combines the proposed SGSL model and UGCN into the self-training framework for end-to-end training. In addition, in order to introduce more supervised signals in the self-training process, negative pseudo-labels are generated for unlabeled samples with low prediction confidence, and then the positive and negative pseudo-labeled samples are trained together with a small number of labeled samples to improve the performance of semi-supervised learning. The code is available upon request.

Rapid and accurate dynamic monitoring and quantitative analysis of red tide disasters are of significant practical importance to national economic development. Remote sensing technology is an effective means for monitoring red tides. This paper utilizes GOCI satellite data and employs a quasi-analytical algorithm (QAA) to retrieve the spectral curves of phytoplankton absorption coefficients. On the basis of a detailed analysis of the differences in the spectral curves of the phytoplankton absorption coefficients between red tide and non-red tide waters, we establish a red tide identification algorithm for the East China Sea on the basis of phytoplankton absorption coefficients. The algorithm is applied to multiple red tide events in the East China Sea. The results indicate that this algorithm can effectively determine the occurrence locations of red tides and extract relevant information about them.

Through AMS-14C and optically stimulated luminescence (OSL) dating, and analysis of microfossils, grain size, and element geochemistry from borehole samples in the Shenhu Bay on western coast of the Taiwan Strait, combined with global paleo-climate and sea level change records, this study reveals two marine transgression events in the Shenhu Bay which happened in Marine Isotope Stage 3 (MIS 3) and the Holocene. The MIS 3 transgression, corresponding to the previously named "Fuzhou Transgression" is characterized by extensive oyster reefs development. And before the Holocene transgression, ancient forests grew in present beach area of the Shenhu Bay, and subsequent transgression deposits preserved ancient trees. The research suggests pattern of "restricted channel incursion" in uplift region, which is that the MIS 3 transgression involved seawater intrusion through ancient incised valleys or low-lying channels formed in past glacial low sea levels period, distinct from conventional understanding of transgression for coastline parallelly advancing toward coast when sea level rises.

Five layers of detrital amphiboles in the CJ08-008 sediment core from the northwest Indian Ocean have been found. To analyze their genetic types and provenance, an electron probe microanalysis of 300 amphibole grains from the core was conducted to calculate the numerical and characteristic values of cations in the crystal structure. The results showed that amphiboles with high Si, Ca, and Mg contents and low Na and K contents exhibit a low degree of weathering and that amphiboles mainly comprise tschermakite (46.43~70.69%), followed by magnesiohornblende, in the calcic amphibole subgroup. The types of sources for these amphiboles are mainly different types of metamorphic and magmatic rock. A large proportion of the detrital amphiboles (>60%) are derived from metamorphic rocks, followed by intermediate acid-intrusive rocks. The genetic analysis of amphiboles showed that most of the medium acid-intrusive amphiboles belong to the crust–mantle type, followed by the mantle type. Most of the amphiboles of metamorphic origin are of the low-pressure type. The amphiboles in the CJ08-008 sediment core exhibit characteristics different from those brought by monsoons from surrounding land masses. The variations in the amphiboles indicate different sources, which may have different origins; these origins could include the Carlsberg Ridge, the Owen Fault Zone, or older submarine sediment sequences eroded by turbidity currents.