Visual analytics research at ITCJun 11th, 2010 | By Prof. Dr. Menno-Jan Kraak | Category: Featured Articles, News
The research focusses on methods and techniques to integrate data from different sources and dimensions (2D, 3D, Time). The data are offered via a variety of map representations in multi-functional visual (online) environments that allow exploration and data analysis. To justify the solutions produced, usability research is an integral part of the activities.
The research approach
Geographic data is characterized by a locational, attribute and temporal component. The time component can be a typical indicator of change of all sorts. For instance, movement data holds information about where the movement took place (the path), what was moving (the object), and especially when it was moving (the time frame). Examples are travel time between cities, the daily trips of commuters, or the possible average speed along a path etc. Many of these movements (trajectories) occur through existing transport networks, and methods and techniques are used to aTnalyze and understand movement patterns.
Maps have an important role to play in visualizing these patterns of change. They have the ability to represent the real world and enable synthesis, analysis and exploration. In the past they only functioned as an inventory of human and natural resources. Today maps form an integral part of a rich GIScience environment to inform, help understanding, increase insight, support decision making and stimulate thinking related to global and local issues.
The theme’s main question addressed is related to the kind of visual representations required to best represent change / movement in a particular application or use context. The representations should allow ‘temporal tasks’ to be executed and support reasoning, improve insight, and contribute to problem solving. It becomes increasingly important that these visualizations can deal with high data volumes, and better represent the dynamics of the process behind the data.
What is the most suitable graphic representation?
Cartographers have developed a set of map design guidelines that guarantee comprehensible maps. However, today’s maps depend on geocomputational support. They are increasingly interactive in nature based on algorithms and linked to models representing knowledge of a certain discipline, like time geography’s space-time-cube.
What is the best working environment?
Executing user tasks requires a environment with specific functionality that allows for a visualization strategy. Most environments are web-based and operate in a spatial data infrastructure (SDI). Work on the Dutch national atlas is an example. The atlas embedded in a SDI requires options for multi- resolution data integration, because data sets with similar or at least related content, at both similar and different scale levels will be used. This is based on the philosophy ‘collect once, use many times’.
Does it work?
A solution is of little value if users are unable to work with it properly. Therefore the visualizations created are always judged on three aspects. Their efficiency (task should be executed with reasonable effort), effectiveness (accurate execution of the task in relation to objective), and satisfaction (users should have a positive attitude to the visualization). The research follows the user-centred design approach where every project starts with a requirements analysis and ends with a usability evaluation.
An example: a web-based tool for exploring iceberg movement
A geo-service to study the movement of icebergs over time combines information from different online databases. It has interactive views for both temporal and spatial dimensions. The combined map and timeline can offer an overview, allow zoom and filtering and give details on demand. Queries can be posed via the map, timeline or menu.
Scope and Objective
Geographic information technologies are evolving from stand-alone systems to a distributed model of independent web services. In parallel, voluminous geographic data are being collected with modern data acquisition techniques such as remote sensing and GPS. Example of such data is iceberg trajectories. Icebergs are tracked using remote sensing images (from 1976 to present) and are made freely available on the Internet (click).
A broad research community is interested in the distribution and behaviour of icebergs. The research is stimulated by phenomena like global warming and climate change, environmental problems like changing habitats, and navigation and engineering activities that are endangered by increasing numbers of floating icebergs. Among the characteristics of icebergs that are studied are their spatio-temporal distribution, movement dynamics, and calving events: splitting from ice shelves of from each other.
There is a need to satisfy such a broad community and to provide exploratory tools for this large data that operate in a distributed (web) environment. The importance of geovisual analytics to explore large sets of geospatial data cannot be overemphasised. However, in geovisual analytics web services are currently hardly used. In the following sections we will describe an example of utilising web visualization libraries on a client side for the manipulation and exploration of iceberg trajectory data.
Methods and Results
Iceberg data are placed in Postgres/PostGIS Data Base Management System. To visualize the integrated iceberg data from the DBMS in a client browser, we chose PHP as middle ware to read the data in the Data Base, restored them in KML files, and then send the files to the client side through an Apache server. The files are dynamically generated: i.e.: if changes are made in the Database, new KML files are created and transferred to the client.
The visualisation of the Antarctic iceberg case is focused on the display of important iceberg events, like first appearances, iceberg splitting (or calvings) and last appearance (or disappearances) and on trajectories, in space and time. To provide overview in the spatial view, it was decided to emphasise at low spatial zoom levels. Details (intermediate iceberg positions and portions of the trajectories) are visible from spatial zoom levels of 5 and higher. In the spatial view, trajectories are distinguished by colour: a random selection out of five colours is made, except if a split occurs (e.g. where iceberg A20 splits into A20A and A20B), then that part of the trajectory is represented in red. Similar colours appear for events and trajectories in the Timeline. The sizes of iceberg (classified) are also represented at each of the displayed positions.
- iceberg name (ID)
- iceberg size (small, medium or big)
- lifetime (short, medium or long)
- average speed (slow, normal or fast)
- travel distance (short, medium or long)
- prevalent wind (eight main directions)
The prototype can be explored here. Offering such a visualization client as a Web Service is an advantage for the research community that is interested in iceberg patterns and dynamics.