|
meteoblue Dynamic Map Server Mathias D. Müller On today's supercomputers, numerical weather prediction models can be run at very high resolution producing gigabytes of data. A user friendly and powerful visualization system is needed to access this information. Traditionally, a predefined set of maps is generated and made available to the forecaster. Only a few national weather services can afford interactive visualization systems where users themselves can control which maps and diagrams to produce. Mathias Müller has developed an internet based dynamic map server as an alternative to expensive workstation solutions, available for experimental use at www.meteoblue.ch. This "mymeteoblue dynamic map server" (Figure 1) makes it possible for everyone to interactively visualize numerical weather forecasts. All that is needed is a web browser with Java plug in. The system takes advantage of IDL for the generation of maps and graphs. It uses ION together with Java, JavaScript and HTML to create dynamic web pages under user control.
Generating a personal weather map with the dynamic map server is a four step process: First the user selects one of several available forecast models that cover geographical regions of different size. The dynamic map server gives access to the meteoblue high resolution forecasts for Europe. These cover all of Europe with a 6-day forecast at a resolution of 12 km and central Europe with a 3-day forecast at 3 km resolution. In addition, the map server gives access to the lower resolution NOAA/NCEP GFS forecast with global coverage. The user then chooses the desired forecast day and hour to visualize. The third step is selecting the type of visualization. Mymeteoblue can generate maps of up to two selected variables (Figure 2), vertical cross sections (Figure 3), and meteograms showing the temporal evolution of the weather at a given point (Figure 4). Furthermore and often used by pilots, vertical soundings (Figure 5) and forward trajectories can be generated. In the fourth step the user defines a rectangular geographical region of interest by simply drawing it on a map on the screen. The last three steps can be taken in any order.
The core of the system is a high-performance database engine that is proprietary to meteoblue. The engine uses smart compression algorithms that enable the dynamic map server to instantly extract relevant information from Gigabytes of model output. The database itself was developed using IDL. The visualization system relies on IDL object graphics to provide maximal performance in rendering. The heart of the visualization system is a proprietary meteoblue mapping library programmed in IDL object graphics. The user front-end of the web-enabled visualization system uses ION Script in combination with Java and JavaScript. The geographical region selection window is realized as a Java applet. As the many possible settings for each map can take time to choose, the map server comes with a user management system that allows users to securely store their personal map settings on the server. User authentication is done using a mixture of HTTP basic authentication, proprietary IDL-based routines for 128 bit encryption and ION Script. An IDL-based credit system limits the number of visualizations a user can make per day. The meteoblue dynamic map server is fairly complex. Using IDL significantly sped up the development process, helping the system's creator Mathias Müller to develop the present functionality within 14 months. Contact: |
