This will include:

• Developing multi-resolution methods for visualization of previously uncorrelated data sets
• Leveraging modern GPU-based features on local graphics workstations and web-based clients
• Supporting new and existing interaction techniques for the decision making process by GPU-based methods
• Developing a 3D-web-based visualisation architecture for the deployment of dedicated visualisation applications to (expert and lay) web users
• Establishing new or extending existing visualization driven data formats

The technical evaluation of the feasibility of the various envisaged approaches to transfer GPU-based visualization techniques to the geospatial domain is part of the tasks of this WP.

---

Task 5.1 Composite visualization methods (SINTEF)

There are a plethora of techniques for scientific visualization of different types of scientific data. However, many techniques are designed for an outdated hardware paradigm (e.g., fixed function OpenGL), and they are most often tailored for a specific use and data representation. The efficient combination of existing visualization techniques for uncorrelated data-sets in a modern hardware paradigm is therefore an open research question. For example, the rendering technique for the different data sets might be “incompatible”, in the sense that their composition is highly non-trivial. We will develop multi-resolution methods for interactive scientific visualization of previously uncorrelated data sets, such as 

• • Vector based cartographic information (e.g., hydrography, cadastre, roads, utility lines); 
  • Raster based terrain data (e.g., elevation data, land use); 
  • Vector and tensor based simulation data (e.g. , wind, pollution, flooding). 

The combination of such data-sets into a single common view is essential to get the overview required in decision making.

---

Task 5.2: Visualization techniques utilizing newest GPU features (Fraunhofer, SINTEF)

 In this task we extend our existing visualization frameworks and address the following issues:

• • How can modern GPU-techniques visualize geo-information data sets more efficiently?
  • What is the possible impact of these techniques on the quality of geo-information visualization?
  • How can image-space based GPU-techniques (deferred mapping) be combined with user interaction techniques to support intuitive interactive geo-information analysis?
  • Can image-space based GPU-techniques be a basis for new deferred fetching algorithms to fetch secondary data in an optimal way?

---

Task 5.3: GPU-supported interaction for decision making (Fraunhofer, SINTEF)

The challenge here is to support interaction techniques (new and existing ones) by GPU-based methods to generate new levels of interactivity and improved quality for the decision making process. Based on the IQmulus scenarios and user requirements from WP1, in this task we will investigate, design and implement interaction methods to enable the users to:

• • Select data to be visualized;
  • Look for more details;
  • Display changes over time;
  • Compare data sets;
  • Probe for result values;
  • Inspect simulation results.

---

Task 5.4: 3D-Web-based Visualization (Fraunhofer)

In this task the focus is on automatable processes for web application development by designing and implementing a fully automated Web Service Portal, which provides web services that automatically convert data into interactive 3D visualizations for the Web that can be delivered via an hybrid approach that provides both streaming for low-end clients and direct web-based 3D rendering for high-end machines.

Web Service Architecture with Transcoder: The HUB service extracts a template, which specifies the data and application characteristics and then it invokes the appropriate transcoder service for providing the concrete data converting it into a declarative deployment format for final presentation.

Client-side Rendering: The visualization application can be realized via client-side rendering by utilizing X3DOM for real-time rendering. Important research issues here are scalable methods for binary compression and suitable caching strategies.

Server-side Rendering: Alternatively server-side rendering is promoted, where the rendered images are transmitted/streamed to the client. Interactions are handled via WebSockets or XHR. To cope with latency issues, suitable hybrid rendering and interaction methods have to be explored.

---

Task 5.5: Visualization driven data formats (Fraunhofer, SINTEF)

This is mainly a research, analysis and design task within which we will collect the requirements on data formats, derived from our algorithms developed in the tasks 5.1 to 5.4. These requirements will be mapped and compared to existing data formats. The comparison could be used in the future for developing visualization-aware data formats or extending existing ones so that they are optimally tailored to new graphics hardware architectures and GPU-based algorithms. An immediate application of such new or extended data formats in IQmulus is not foreseen, since most layers would be affected by data format changes, requiring additional adoption and development efforts by the affected project partners.