1. Abstract: The recent proliferation of virtual globes seems to have captured the public imagination to a degree seldom seen in the Earth Sciences. Virtual globes, such as Google Earth and World Wind, do such a fantastic job of rendering geolocated imagery on a 3D Earth it would be almost criminal not to take advantage. For just the cost of creating compatible imagery the scientific community can offer data visualization capabilities with features like zoom, variable transparency, overlays, etc... But for the climate change community 3D is not enough. To show change over time one needs 4D capabilities. Using 20-30 year time series data from the National Snow and Ice Data Center this presentation explores methods for incorporating four dimensional data into virtual globes. Introduction: Virtual Globes are clearly a concept whose time has come. With half a dozen already in existence, and several more in development, it’s difficult to keep up with all the activity. New functionality seems to arrive on a weekly basis, and a plethora of new datasets are made compatible everyday. The majority of our work with Virtual Globes has involved Google Earth simply because it has proven the most popular. Initially the intent was to create a different kind of access to our data and popular software enables the widest access. Largely because of the popularity of Google Earth, it looks like KML is becoming the de-facto standard for the creation of compatible datasets, so we may soon be able to use other systems with little or no modifications to our existing files. For more information on NSIDC’s work with virtual globes see: http://nsidc.org/data/virtual_globes Ross Swick (swick@nsidc.org), Vincent J. Troisi, John Maurer, I-Pin WangAGU 2006 #IN43A-0894 National Snow and Ice Data Center Visualization of Four Dimensional Data on Virtual Globes The September Arctic sea ice minimum is shown in this flipbook style animation. Both the data image and the date overlay are changed on the server every three seconds. The KML is set to refresh Google Earth every three seconds. The 27-year time series repeats every 81 seconds. The March Arctic maximum and the full time series (all months) are also implemented as flipbooks. Giving the user control of the flipbook would require some additional work to generate a custom KML file and server side image rotator for each user, and an interface to coordinate the two. The same sea ice time series is shown using the new timeline widget. The elimination of the image rotator on the server side means less maintenance and less chance of something going wrong. The widget gives the user control over the animation. The arrow on the right controls pause/play. The white bar controls the range of data displayed at any one time. And the little circle on the left opens a panel to control the speed and cycling of the animation. It’s still beta so there’s a few kinks to work out yet, but it does greatly simplify the task of animating time series data. NSIDC has been generating browse imagery of the gridded MODIS daily snow product for several years, so we already had the images we needed to load into Google Earth. Writing a perl script to generate monthly KML files for the five years of available browse imagery took about half a day. The advantage for the user is the ability to view the browse images quickly, rather than having to click on each daily image. Moreover, users interested in a specific region can zoom in and inspect the browse for just that region, looking for the conditions they want. NISDC’s GISMO subsetter has been generating subsets of our SSM/I and AVHRR gridded data since the turn of the century, but we have not offered quick-look browse in the past. Some minor additions to the processing system that does the subsetting enabled the creation of Google Earth compatible browse images and a KML file to animate them. As an added bonus it is possible to animate two parameters simultaneously and adjust the transparency of the top layer to see how they correlate or interact. If that proves useful we may investigate creating false color animations in a similar manner. Point data is much easier to animate as one does not have to generate the imagery and both network and memory requirements are minimized. This sequence shows an animation of hourly data gathered from a buoy for almost a year. Again it took about half a day to write a perl script to create the KML from the data and that script can now be reused for any similar data. As an added bonus, the animation highlights faulty locations in the data that might be harder to find through simple inspection of the table. As the last frame illustrates current speed is visualized quite well in the animation Using KML 2.0 the only practical way to animate time series data in a virtual globe is by creating a flipbook using the refresh tag. Frequent refreshes in the KML, coordinated with an image rotator on the server side, can be an effective way to visualize changes over time. The flipbook method does require a fairly good connection to constantly refresh the image, and giving the user control of the animation is somewhat difficult. The forthcoming KML 2.1 includes tags specifically for creating time series. The timestamp and timespan tags are used by Google Earth 4.0.2 beta to order the imagery or placemarks, and the new timeline widget gives the user control over the animation. The entire time series is loaded into memory once, so the network connection isn’t used continuously. 4D Visualization: Future Work: The examples to the right have all been prototyped and tested in the beta version of Google Earth. Currently we’re seeking feedback from the user community to determine what kinds of functionality they like most, and what kinds of data they find most useful, in a virtual globe environment. As a next step we plan to explore linking Virtual Globes to a Web Map Server as a way to create a more dynamic data visualization capability. In general, the idea of creating virtual globe compatible datasets on demand is appealing because the visualization tool already exists. Moreover, because virtual globes are a community resource, they are constantly being added to and improved upon, and our users benefit from those improvements.