Aviation turbulence is an essential factor threatening flight safety. However, due to its complex mechanism, which study has been one of the key issues faced by the aviation industry. In recent years, with the development of large eddy numerical simulation (LES), which has become an important method to solve aviation turbulence problems. This study reviews the research progress of LES technology in the past few decades, focusing on the impact of aircraft trailing vortex wakes and low-level turbulence during the takeoff and landing phase, as well as convective induced turbulence, mountain wave turbulence, and clear air turbulence during the cruise phase on aircraft turbulence. It also summarizes and prospects the urgent problems to be solved in the application of LES technology and future key research directions. Overall, LES simulation research on aviation turbulence has got much achievement, which can clarify the source and lifecycle of aviation turbulence more clearly, significantly improving the mechanism cognition, quantitative diagnosis, prediction and warning capabilities. However, in terms of mechanism, the interaction mechanism of various complex turbulent processes is still unclear; in terms of numerical model techniques, the predictive skill of LES of aviation turbulence is still limited by errors in initial conditions, boundary conditions, and the models themselves (e.g., parameterizations, dynamical methods). In the future, the development of nesting and dynamic grid technology between LES and mesoscale regional models, high-resolution ensemble prediction methods and probability prediction approaches, as well as the combination with deep learning methods will further improve the computational efficiency and prediction ability of LES on aviation turbulence simulation and forecasting.

Massive ocean data acquired by various observing platforms and sensors poses new challenges to data management and utilization. Typically, it is difficult to find the desired data from the large amount of datasets efficiently and effectively. Most of existing methods for data discovery are based on the keyword retrieval or direct semantic reasoning, and they are either limited in data access rate or do not take the time cost into account. In this paper, we creatively design and implement a novel system to alleviate the problem by introducing semantics with ontologies, which is referred to as Data Ontology and List-Based Publishing (DOLP). Specifically, we mainly improve the ocean data services in the following three aspects. First, we propose a unified semantic model called OEDO (Ocean Environmental Data Ontology) to represent heterogeneous ocean data by metadata and to be published as data services. Second, we propose an optimized quick service query list (QSQL) data structure for storing the pre-inferred semantically related services, and reducing the service querying time. Third, we propose two algorithms for optimizing QSQL hierarchically and horizontally, respectively, which aim to extend the semantics relationships of the data service and improve the data access rate. Experimental results prove that DOLP outperforms the benchmark methods. First, our QSQL-based data discovery methods obtain a higher recall rate than the keyword-based method, and are faster than the traditional semantic method based on direct reasoning. Second, DOLP can handle more complex semantic relationships than the existing methods.