Research On Uncertainty Visualization Method Of Transport Variance In Ensemble Vector Field

With the rapid development of high performance computing,large-scale scientific data are produced,and new challenges are put forward to how to analyze large-scale data effectively.Visualization is an important way to analyze and explore large-scale data.It aims to display data in a more intuitive way by mapping data to various graphic elements,and helps users analyze valuable information that is not easy to be found.Vector field visualization is one of the most challenging subdomains in visualization and it plays an important role in the analysis of vector field data.Through displaying vector field data in motion graphics,color and texture,It can help researchers to explore the complex physical phenomena that are difficult to be understood in the vector field.In the process of data acquisition and use,uncertainty will be introduced inevitably,and uncertainty has become an important part of the data.In the process of analyzing the data,uncertainty visualization can help researchers understand the data more comprehensive and make reasonable decisions.At present,uncertainty visualization has become a hot research topic in visualization.Vector field uncertainty visualization has attracted more and more researchers' attention.By generating the set of simulated vector field data and visualizing the uncertainty,it can show the overall situation of the vector field data and the differences among members to help the domain experts to analyze all kinds of scientific phenomena more scientifically.In view of this,in this dissertation a new method of measuring the transport difference among the ensemble members is proposed and the uncertainty of the ensemble vector field is visualized.The research contents are as follows: First,in order to measure the difference between trajectories of different lengths and avoid the effect of noise,a distance measurement method based on EDR(edit distance on real sequence)is proposed.Comparing with the traditional distance metric methods,the accuracy of this method is higher.Secondly,based on the spatial neighborhood structure correlation,a comprehensive uncertainty measurement is proposed.The moving variance between grid points and their neighborhood in different ensemble members is comprehensively considered,and two differences are combined effectively.In order to reveal the uncertainty of grid points,we improved the Canopy-Kmeans clustering algorithm for ensemble pathline cluster.The results of cluster can help domain experts identify and analyze different transport patterns clearly.Furthermore,in order to explore the divergence degree of different clusters,a glyph named shuttlecock is designed to visualize the details in different transport patterns,and to compare the transport differences of different grid points.The Double-Gyre synthetic dataset,ECMWF dataset and Hurricane Isabel dataset are performed in our experiments and the results demonstrate that our method is more effective and accurate than other methods for measuring the uncertainty of the ensemble vector field.