The gap between the Gravity Recovery and Climate Experiment (GRACE) and GRACE Follow-on mission (GRACE-FO) is a crucial problem, leading to discontinuous global mass change information from time-variable gravity field solutions. We aim to use the improved Multichannel Singular Spectrum Analysis (MSSA) to fill the gap and reconstruct the continual global mass change signals in this study. The one-year gravity field gap is filled with improved MSSA based on the Release 06 (RL06) monthly gravity field models provided by Center for Space Research (CSR) truncated to d/o 60 and further compared with Swarm monthly solutions (d/o 40). The results show that the infilled gravity field models agree with the Swarm solutions, indicating that improved MSSA can reliably bridge the gap between GRACE and GRACE-FO missions. Moreover, relative to the Global Land Data Assimilation System (GLDAS) Noah model, the infilled gravity field models have higher correlation coefficients and smaller root mean squared errors than Swarm solutions for the mass change signals over Congo, Lena and Fraser river basins. According to the simulation results using the same gap as GRACE and GRACE-FO, improved MSSA can fill the gap very efficiently and reliably in 25 major river basins over the World.

Fine particulate matter with aerodynamic diameters <= 2.5 mu m (PM2.5) is a proxy for atmospheric pollution levels and is detrimental to human health. The linear mixed-effects (LME) model has been extensively utilized to map regional and hourly PM2.5 levels in the daytime through geostationary satellite-derived aerosol retrievals. However, further improving the performance of the model is difficult because of its limited attention to the spatial and temporal heterogeneity of different predictors that in turn restricts its application to large-scale regions. Using a space-time LME (STLME) model, this study aims to produce an hourly PM2.5 map for the Beijing-Tianjin-Hebei region on the basis of advanced Himawari-8 image aerosol retrievals. In situ ground PM2.5 observations and meteorological and geographical variables are utilized to obtain hourly estimations of PM2.5 concentrations in 2018. Three 10-fold cross-validation (CV) methods, namely, temporal-, spatial-, and sample-based CV, are employed for validation. The results reflect the advantages of the STLME model over the traditional LME model, including its high determination coefficient of (0.83 versus 0.68), small root-mean-square error of (20.9 mu g/m(3) versus 28.1 mu g/m(3)), and minimal mean absolute error of (13.0 mu g/m(3) versus 18.3 mu g/m(3)). Thus, the STLME model shows great potential for air pollution applications because of its effective mapping of surface PM2.5 levels by coordinating the space-time information of different predictors. (C) 2021 Elsevier Ltd. All rights reserved.

Intensified use of disinfectants to control COVID-19 could unintentionally increase the disinfection byproducts (DBPs) in the environment. In indoor spaces, it is critical to determine the optimal disinfection practice to prevent the spread of the virus while keeping DBPs at relatively low levels in the air. The formation of DBPs exceed 0.1 mu g/mg while hypochlorite dosed at 10 mg/m3. The total DBP concentrations in highly disinfected places (100-200 mg/m3 hypochlorite) were as high as 66.8 mu g/m3, and the Hazard Index (HI) was up to 0.84, and both values were much higher than those in less disinfected places (<10 mg/m3 hypochlorite). Taking into account the HI, formation yields and the origin of the DBPs, we recommended 10 mg/m3 as the suggested hypochlorite dose to minimize DBPs generation during routine disinfection for controlling the coronavirus. DBPs in indoor air could be eliminated by ventilation, reducing the usage of personal care products, and wiping the solid surface with water before or after disinfection. These results highlighted the necessity to control air-borne DBPs and their associated health risks arising from intensified disinfection, and will guide the further development of evidence-based regulation on DBP exposure during disinfection and improve public health protection.

An altered total seminal plasma glycosylation has been associated with male infertility, and the highly abundant seminal plasma glycoprotein prostate-specific antigen (PSA) plays an important role in fertilization. However, the exact role of PSA glycosylation in male fertility is not clear. To understand the involvement of PSA glycosylation in the fertilization process, analytical methods are required to study the glycosylation of PSA from seminal plasma with a high glycoform resolution and in a protein-specific manner. In this study, we developed a novel, high-throughput PSA glycopeptide workflow, based on matrix-assisted laser desorption/ionization-mass spectrometry, allowing the discrimination of sialic acid linkage isomers via the derivatization of glycopeptides. The method was successfully applied on a cohort consisting of seminal plasma from infertile and fertile men (N = 102). Forty-four glycopeptides were quantified in all samples, showing mainly complex-type glycans with high levels of fucosylation and sialylation. In addition, N,N-diacetyllactosamine (LacdiNAc) motives were found as well as hybrid-type and high mannose-type structures. Our method showed a high intra- and interday repeatability and revealed no difference in PSA glycosylation between fertile and infertile men. Next to seminal plasma, the method is also expected to be of use for studying PSA glycopeptides derived from other biofluids and/or in other disease contexts.

Domain Adaptation (DA) aims to generalize the classifier learned from a well-labeled source domain to an unlabeled target domain. Existing DA methods usually assume that rich labels could be available in the source domain. However, we usually confront the source domain with a large number of unlabeled data but only a few labeled data, and thus, how to transfer knowledge from this sparsely-labeled source domain to the target domain is still a challenge, which greatly limits its application in the wild. This paper proposes a novel Sparsely-Labeled Source Assisted Domain Adaptation (SLSA-DA) algorithm to address the challenge with limited labeled source domain samples. Specifically, due to the label scarcity problem, the projected clustering is first conducted on both the source and target domains, so that the discriminative structures of data could be exploited elegantly. Then label propagation is adopted to propagate the labels from those limited labeled source samples to the whole unlabeled data progressively, so that the cluster labels are revealed correctly. Finally, we jointly align the marginal and conditional distributions to mitigate the cross-domain mismatching problem, and optimize those three procedures iteratively. However, it is nontrivial to incorporate the above three procedures into a unified optimization framework seamlessly since some variables to be optimized are implicitly involved in their formulas, thus they could not benefit to each other. Remarkably, we prove that the projected clustering and conditional distribution alignment could be reformulated into other formulations, thus the implicit variables are embedded in different optimization steps. As such, the variables related to those three quantities could be optimized in a unified optimization framework and benefit to each other, and improve the recognition performance obviously. Extensive experiments have verified that our approach could deal with the challenge in the SLSA-DA setting, and achieve the best performances across different real-world cross-domain visual recognition tasks. Our preliminary Matlab code is available at https://github.com/WWLoveTransfer/SLSA-DA/ . (c) 2021 Elsevier Ltd. All rights reserved.

Cross-modal retrieval aims at retrieving relevant points across different modalities, such as retrieving images via texts. One key challenge of cross-modal retrieval is narrowing the heterogeneous gap across diverse modalities. To overcome this challenge, we propose a novel method termed as Cross-modal discriminant Adversarial Network (CAN). Taking bi-modal data as a showcase, CAN consists of two parallel modality-specific generators, two modality-specific discriminators, and a Cross-modal Discriminant Mechanism (CDM). To be specific, the generators project diverse modalities into a latent cross-modal discriminant space. Meanwhile, the discriminators compete against the generators to alleviate the heterogeneous discrepancy in this space, i.e., the generators try to generate unified features to confuse the discriminators, and the discriminators aim to classify the generated results. To further remove the redundancy and preserve the discrimination, we propose CDM to project the generated results into a single common space, accompanying with a novel eigenvalue-based loss. Thanks to the eigenvalue-based loss, CDM could push as much discriminative power as possible into all latent directions. To demonstrate the effectiveness of our CAN, comprehensive experiments are conducted on four multimedia datasets comparing with 15 state-of-the-art approaches. (C) 2020 Elsevier Ltd. All rights reserved.Y