Predicting the passenger flow of metro networks is of great importance for traffic management and public safety. However, such predictions are very challenging, as passenger flow is affected by complex spatial dependencies (nearby and distant) and temporal dependencies (recent and periodic). In this paper, we propose a novel deep-learning-based approach, named STGCNNmetro (spatiotemporal graph convolutional neural networks for metro), to collectively predict two types of passenger flow volumes—inflow and outflow—in each metro station of a city. Specifically, instead of representing metro stations by grids and employing conventional convolutional neural networks (CNNs) to capture spatiotemporal dependencies, STGCNNmetro transforms the city metro network to a graph and makes predictions using graph convolutional neural networks (GCNNs). First, we apply stereogram graph convolution operations to seamlessly capture the irregular spatiotemporal dependencies along the metro network. Second, a deep structure composed of GCNNs is constructed to capture the distant spatiotemporal dependencies at the citywide level. Finally, we integrate three temporal patterns (recent, daily, and weekly) and fuse the spatiotemporal dependencies captured from these patterns to form the final prediction values. The STGCNNmetro model is an end-to-end framework which can accept raw passenger flow-volume data, automatically capture the effective features of the citywide metro network, and output predictions. We test this model by predicting the short-term passenger flow volume in the citywide metro network of Shanghai, China. Experiments show that the STGCNNmetro model outperforms seven well-known baseline models (LSVR, PCA-kNN, NMF-kNN, Bayesian, MLR, M-CNN, and LSTM). We additionally explore the sensitivity of the model to its parameters and discuss the distribution of prediction errors.

The increase of temporal resolution from weekly to daily and spatial resolution from ~1° to ~(1/4)° in merged altimeter sea surface height data along with their over two decades of time series opens an unprecedented opportunity to reveal the “spectrum” of eddy variabilities from days to years in time and from mesoscale to semimesoscale in space. Eddies with a lifetime shorter than 1 month or longer than 1 year are classified here as short lived and long lived, respectively. Contrasting the variabilities of short‐ and long‐lived eddies could be an effective way to explore their geographic origins and dynamic formations. In the time domain, the population fluctuations of short‐ and long‐lived eddies are basically dominated by annual and interannual component, respectively. The magnitude of kinematic and dynamic properties of oceanic eddies is positively correlated with their lifetime in general. Statistically, the properties of short‐lived (long‐lived) eddies experience a symmetric (an asymmetric) growth and decay with a single flat peak during their life cycles. In the space domain, eddy occurrence is observed as a high probability event, which can reach every corner of the ocean. However, homes of short‐ and long‐lived eddies are highly regionalized and are geographically separated. A prominent “young eddy belt” is observed in the tropical oceans for the first time. These findings suggest two fundamental characteristics with regard to short‐ and long‐lived mesoscale eddies: residing in largely separated geographic zones under different mechanisms while following a similar pattern of intrinsic life cycle in terms of property evolution.

Among the various challenges that spaceborne radar observations of the ocean face, the following two issues are probably of a higher priority: inadequate dynamic resolution, and ineffective vertical penetration. It is therefore the vision of the National Laboratory for Marine Science and Technology of China that two highly anticipated breakthroughs in the coming decade are likely to be associated with radar interferometry and ocean lidar (OL) technology, which are expected to make a substantial contribution to a submesoscale-resolving and depth-resolving observation of the ocean. As an expanded follow-up of SWOT and an oceanic counterpart of CALIPSO, the planned “Guanlan” science mission comprises a dual-frequency (Ku and Ka) interferometric altimetry (IA), and a near-nadir pointing OL. Such an unprecedented combination of sensor systems has at least three prominent advantages. (i) The dual-frequency IA ensures a wider swath and a shorter repeat cycle which leads to a significantly improved temporal and spatial resolution up to days and kilometers. (ii) The first spaceborne active OL ensures a deeper penetration depth and an all-time detection which leads to a layered characterization of the optical properties of the subsurface ocean, while also serving as a near-nadir altimeter measuring vertical velocities associated with the divergence, and convergence of geostrophic eddy motions in the mixed layer. (iii) The simultaneous functioning of the IA/OL system allows for an enhanced correction of the contamination effects of the atmosphere and the air-sea interface, which in turn considerably reduces the error budgets of the two sensors. As a result, the integrated IA/OL payload is expected to resolve the ocean variability at submeso and sub-week scales with a centimeter-level accuracy, while also partially revealing marine life systems and ecosystems with a 10-m vertical interval in the euphotic layer, moving a significant step forward toward a “transparent ocean” down to the vicinity of the thermocline, both dynamically and bio-optically.

This study is to quantify the effects of mesoscale eddies on air‐sea heat fluxes and related air‐sea variables in the South China Sea. Using satellite observations of sea surface temperature (SST) and sea surface height anomaly and a high‐resolution air‐sea heat flux product for the 16‐year period from 2000 to 2015, we conducted the composite patterns of air‐sea fluxes and variables associated with anticyclonic eddies (AEs) and cyclonic eddies (CEs). It is found that the SST‐sea surface height correlations over eddies are not always positive. Only 56% of AEs are corresponded with positive SST anomalies (SSTA), that is, SST⁺ AEs, and 58% of CEs with negative SSTA, that is, SST⁻ CEs. The percentage of these eddies increases with eddy amplitude and shows slight seasonal variations, higher in winter and lower in summer. Composites of SSTA, air‐sea variables, and fluxes are constructed over all eddies, including both SST⁺ eddies and SST⁻ eddies. All composites show asymmetric patterns, showing that the centers (where the extrema are located) of the fluxes and variables shift westward and poleward (equatorward) relative to the AEs (CEs) cores. Besides, composites of latent heat flux (LHF), sensible heat flux (SHF), and air temperature show monopole patterns, while composites of wind speed and specific humidity show dipole patterns. For SST⁺ AEs, the coupling strength is 39.6 ± 6.5 W/m² (7.2 ± 1.7 W/m²) per degree increase of SSTA for LHF (SHF). For SST⁻ CEs, the coupling strength is 39.0 ± 2.0 W/m² (9.0 ± 0.96 W/m²) per degree decrease of SSTA for LHF (SHF).

Marine debris impacts negatively upon the marine environment and the survival of marine life because they are some difficult-to-degrade substances, and most of them will sink into the deep sea and continue to exist in the ocean. Autonomous underwater vehicles can clean up the deep-sea debris to some extent. However, the efficient detection method plays a critical role in the collection rate. This article establishes an efficient deep-sea debris detection method with high speed using deep learning methods. First, a real deep-sea debris detection dataset (3-D dataset) is established for further research. The dataset contains seven types of debris: cloth, fishing net and rope, glass, metal, natural debris, rubber, and plastic. Second, the one-stage deep-sea debris detection network ResNet50-YOLOV3 is proposed. In addition, eight advanced detection models are also involved in the detection process of deep-sea debris. Finally, the performance of ResNet50-YOLOV3 is verified by experiments. Furthermore, the applicability and effectiveness of ResNet50-YOLOV3 in deep-sea debris detection are proved by the experimental results.

The Surface Water and Ocean Topography (SWOT) mission provides a good opportunity to study fine‐scale processes in the global ocean but whether it can detect balanced submesoscale eddies is uncertain due to the “contamination” by unbalanced inertial gravity waves. Here, based on concurrent observations from SWOT and a mooring array in the northwestern Pacific, we successfully captured two submesoscale cyclonic eddies with negative sea level anomalies (SLAs) in spring 2023. We find that the SLA amplitude and equivalent radius of the first (second) eddy are 2.5 cm and 16.0 km (2.0 cm and 18.8 km), respectively. For both eddies, their vertical scales are around 150 m and their horizontal velocities and Rossby numbers exceed 15.0 cm/s and 0.4, respectively. Further analysis suggests that similar submesoscale eddies can commonly occur in the northwestern Pacific and that SWOT is capable to detect larger submesoscale eddies with scales greater than ∼10 km.