| At present, the research about tsunami numerical compute is already basically mature.Tsunami numerical model adopts explicit staggered leap frog finite difference schemes to computeover time and space based on nested grid system, then, it can obtain more accurate data.Additional, choosing different control equations in different sea areas is also one of elements thatwill impact the accuracy. Here, tsunami model adopts finite difference methods to solve shallowwater equations to simulate the whole tsunami process which begins to generate, propagate and goup near to land. According to the characteristics of tsunami propagation in different regions, wechoose different meshes with different resolutions and numerical models. When tsunami ispropagating, here systematical depict the calling relation between many grids, and introduce howto move coastline when tsunami waves are near to land. During doing tsunami experiment, usingthe data of fault modes happened about M7.2earthquake at shallow area of Taiwan Strait tosimulate. Meanwhile, the simulation is picking a shallow area near to the monitoring station as anobservation position. After repeatedly research and improvement, we can get more accurate datain order to do visualization later period.With more and more accurate tsunami numerical simulation, three dimensional emulationsabout tsunami scenes also acquired widely research and accepted. It emulates tsunami by vividlyperceptive feeling. However,3D tsunami is also limited since it can mainly show spatialinformation. How to expand more dimensions to show more details about tsunami becomes anemphasis of our research project. The data of the water surface elevation and the volume fluxes inx direction and y direction and so on based on numerical simulation should be visualized as manydimensions as possible. Then, it can immediately get more practically valuable information abouttsunami from users’ vision perception.Scientific computing and visualization are already widely used in many fields and become animportant research orientation in the field of computer graphics. But, the use of texture in the areaof computer graphics concerned with data visualization is not as mature and studied as the use oftexture in realistic image synthesis. Flow visualization is an example of a visualization methodthat uses texture to visualization both direction and strength of a flow field. Nevertheless, flowvisualization is one of few examples of texture use in data visualization. This thesis focuses on thedefinition of a multidimensional texture synthesis model. Then, taking use of perceptive texturemodels visualizes multidimensional data obtained by tsunami numerical simulation. Taking use ofthe characteristics of texture, i.e. orientation, frequency, color and contrast can respectivelyindicate orientation, magnitude of the vector, the variation of water surface elevation and depth.Meanwhile, several controlled texture parameters can be used to control orientation, spatial frequency, contrasting and so on. Finally, it will obtain multi dimensional visualization texture.Additionally, this paper use the conditional entropy as the measurement to select the optimalmodel parameter values, leading to flow visualization results that are consistent with the originalvector field according to human visual perception.During visualization, this paper tells us that using many patches with overall direction canindicate vectors’ orientation. Then, another question needs to be solved that the output texture isdiscovered many seams. The texture patches with different characteristics can sew up to generate abig texture based the strategy of image quilting algorithm. Since each splicing texture blocks arechosen by the location of vector data and can’t be rotated, it can lead to that there is bigcharacteristic difference between the output texture and the chosen texture. In spite of theimprovement, it still produces rhombus seams. After many experiment, filter can make thevisualization better and satisfied human visual perception. |
PDF Full Text Request