Three-Dimension Visualization Of Weather Data

Visualization in Scientific Computing is a hot spot of Computer Graphics. It has greatly accelerated the procedure of data processing and enhanced the quality of the results. Weather Forecast plays an important role in the lives of hundreds of millions of people and the development of the national economy. Accurate Weather Forecast will largely reduce people's lives and property losses. The accuracy of Weather Forecast relies on the large-scale weather data acquisition and analysis. Weather Forecast requires quick response and atmospheric research center must analysis weather data in time. The application of visualization techniques in the area of Weather Forecast turns the indigestible large-scale data into 2D and 3D images, which significantly improves the understanding of the data sets and helps people to make accurate judgments quickly. Therefore, it is worthwhile to investigate three-dimension weather data visualization.Classification is one of the two key steps of visualization techniques. Only through good classification, users can quickly and accurately find the implied information of data sets and make reasonable judgments. In the classification step, the task is to define a mapping between data values and the corresponding colors and opacities. After classification, the regions of interest and volume of interest are more evident and the unnecessary information is hided. There are some special properties of weather data, such as time-varying and multi-valued, so it is difficult to get a good classification result. The main part of this paper will discuss how to achieve a good classification result.We employed two approaches to the volume classification problem. One method is based on 2D scatter plot. Any two fields of the weather data can be axes of the 2D scatter plot. 2D scatter plot is used as interaction interface. The other method is based on machine learning. This method is firstly used in medical visualization area and we extended it into weather data visualization application. There are some differences between medical data and weather data. We will discuss these problems in this paper. We also implemented a flexible and scalable classification system. It is based on the volume rendering system of our laboratory. The visualization results demonstrate that our proposed classification methods are very effective.