In this paper, we propose a prototype-based classification model for evolving data streams, called SyncStream, which allows dynamically modeling time-changing concepts, making predictions in a local fashion. Instead of learning a single model on a fixed or adaptive sliding window of historical data or ensemble learning a set of weighted base classifiers, SyncStream captures evolving concepts by dynamically maintaining a set of prototypes in a proposed P-Tree, which are obtained based on the error-driven representativeness learning and synchronization-inspired constrained clustering. To identify abrupt concept drifts in data streams, PCA and statistical analysis based heuristic approaches have been introduced. To further learn the associations among distributed data streams, the extended P-Tree structure and KNN-style strategy are introduced. We demonstrate that our new data stream classification approach has several attractive benefits: (a) SyncStream is capable of dynamically modeling the evolving concepts from even a small set of prototypes. (b) Owing to synchronization-based constrained clustering and P-Tree, SyncStream supports efficient and effective data representation and maintenance. (c) SyncStream is also tolerant of inappropriate or noisy examples via error-driven representativeness learning. (d) SyncStream allows learning relationship among distributed data streams at the instance level. The experimental results indicate its efficiency and effectiveness.

In this study, 6 widely used precipitation products APHRODITE, CPC_UNI_PRCP, CN05.1, PERSIANN-CDR, Princeton Global Forcing (PGF), and TRMM 3B42 V7 (TMPA), were evaluated against gauge observations (CMA data) from 1998 to 2014, and applied to streamflow simulation over the Upper Yellow River basin (UYRB), using 4 hydrological models (DWBM, RCCC-WBM, GR4J, and VIC). The relative membership degree (u), as the comprehensive evaluation index in the hydrological evaluation, was calculated by the optimum fuzzy model. The results showed that the spatial pattern of precipitation from the CMA dataset and the other 6 precipitation products were very consistent with each other. The satellite-derived rainfall products (SDFE), like PSERSIANN-CDR and TMPA, depicted considerably finer and more detailed spatial heterogeneity. The SDFE and reanalysis (RA) products could estimate the monthly precipitation very well at both gauge and basin-average scales. The runoff simulation results indicated that the APHRODITE and TMPA were superior to the other 4 precipitation datasets, obtaining much higher scores, with average u values of 0.88 and 0.77. The precipitation estimation products tended to show better performance in streamflow simulation at the downstream hydrometric stations. In terms of performance of hydrological models, the RCCC–WBM model showed the best potential for monthly streamflow simulation, followed by the DWBM. It indicated that the monthly models were more flexible than daily conceptual or distributed models in hydrological evaluation of SDFE or RA products, and that the difference in precipitation estimates from various precipitation datasets were more influential in the GR4J and VIC models.

Traditional semi-supervised learning tasks assume that both labeled and unlabeled data follow the same class distribution, but the realistic open-world scenarios are of more complexity with unknown novel classes mixed in the unlabeled set. Therefore, it is of great challenge to not only recognize samples from known classes but also discover the unknown number of novel classes within the unlabeled data. In this paper, we introduce a new open-world semi-supervised novel class discovery approach named OpenNCD, a progressive bi-level contrastive learning method over multiple prototypes. The proposed method is composed of two reciprocally enhanced parts. First, a bi-level contrastive learning method is introduced, which maintains the pair-wise similarity of the prototypes and the prototype group levels for better representation learning. Then, a reliable prototype similarity metric is proposed based on the common representing instances. Prototypes with high similarities will be grouped progressively for known class recognition and novel class discovery. Extensive experiments on three image datasets are conducted and the results show the effectiveness of the proposed method in open-world scenarios, especially with scarce known classes and labels.

Stream flow plays a crucial role in the environment, society, and the economy, and identifying the causes of changes in runoff is important to understanding the impact of climate change and human activity. This study examines the variation trends in recorded runoff for the Xinshui River, a tributary of the Yellow River on the Loess Plateau, and uses hydrological simulations to investigate how climate change and human activity have contributed to those trends. Results show that the recorded runoff at the Daning station on the Xinshui River declined significantly from 1955–2008 with an abrupt change occurring in 1973. The Simplified Water Balance Model (SWBM) simulates monthly discharge well with a Nash–Sutcliffe efficiency (NSE) coefficient of 78% and a relative error of volumetric fit (RE) of 0.32%. Runoff depth over the catchment in 1973–2008 fell by 25.5 mm relative to the previous period, with human activity and climate change contributing 60.6% and 39.4% of the total runoff reduction, respectively. However, the impacts induced by climate change and human activities are both tending to increase. Therefore, efforts to improve the ecology of the Loess Plateau should give sufficient attention to the impacts of climate change and human activity.