1. Visualizations of Cryospheric Data in Virtual Globes at the National Snow and Ice Data Center MODIS Mosaic of Antarctica (MOA) Image Map Kara Gergely, Terry Haran, Brendan Billingsley – National Snow and Ice Data Center Special thanks to Jonathan Bamber, University of Bristol, for ERS 1 & 2 and IceSat DEM Project Summary Virtual globes do a fantastic job of rendering geolocated imagery on a 3D earth. For just the cost of creating compatible imagery, members of the scientific community can develop data visualization capabilities with features such as 3D perspective, zoom, variable transparency, overlays, and time series animation. The National Snow and Ice Data Center (NSIDC) has been creating imagery to visualize our data holdings in virtual globes for several years with considerable success. But since different kinds of data have different visualization needs, we are constantly looking for new ways to use virtual globe technologies to help the earth science community. This presentation highlights virtual globe visualizations of the MODIS of Antarctica ( MOA) image map. Background Staff from the National Snow and Ice Data Center (NSIDC) and the University of New Hampshire have assembled a digital image map and a snow-grain-size image of the Antarctic continent and surrounding islands. The Moderate Resolution Imaging Spectroradiometer (MODIS) Mosaic of Antarctica (MOA) image map is a composite of 260 swaths comprised of both Terra and Aqua MODIS images acquired between 20 November 2003 and 29 February 2004. MOA provides a cloud-free view of the ice sheet, ice shelves, and land surfaces at a grid scale of 125 m and an estimated resolution of 150 m. All land areas south of 60° S that are larger than a few hundred meters are included in the mosaic. Also included are several persistent fast ice areas and grounded icebergs. MOA consists of two MODIS-derived image data sets: a digitally smoothed red-light image, which was compiled using Band 1; and a snow-grain-size image, which was compiled using the normalized difference of calibrated data from Band 1 and Band 2 data. Images were destriped, georeferenced, and resampled using the MS2GT software available at NSIDC. Acquisition times were limited to 0500 - 1300 Universal Time continent-wide to provide a more uniform illumination direction of ice surface morphology across the image seams, while maintaining the ability to capture linear snow features of every orientation. The 125 m grid geolocation is identical to the Radarsat Antarctic Mapping Project Antarctic Mapping Mission 1 (RAMP AMM-1) 125 m mosaic. The MOA image map is complimentary to the radar-image-based RAMP mosaic, highlighting true surface morphology without a subsurface volume backscattering component. This results in a better discrimination between accumulation- and crevasse-related subsurface changes and surface features. The digitally smoothed red-light images are available via FTP at two spatial grid scales: 750 m (112 MB) and 125 m (4 GB), and via a Web-based map server capable of creating manually-selected JPEG images. A variety of pre-processed contrast stretches are available for the JPEG images. The snow-grain-size images are available only at 750 m resolution via FTP. Image data on the FTP site include a 16-bit digitally smoothed red-light image to preserve the radiometric content of the scenes. This image was the input for the creation of virtual globe compatible files. The KML file is available on NSIDC’s Virtual Globes Technical Experiments Web site. -National Snow and Ice Data Center, http://www.nsidc.org/data/moa 18 December 2008 This is an example of a mosaic image at nominal contrast (15000 – 17000) displayed on NSIDC’s MOA map server. An inset of Byrd Glacier at the base of the Transantarctic Mountains and part of the Ross Ice Shelf showcases the detail available in the MOA image map. Byrd Glacier Ross Ice Shelf Transantarctic Mountains Comparisons of MOA Image Map Visualizations MOA image map can be visualized in various ways including NSIDC’s Map Server or virtual globe applications such as Google Earth™ or ArcGlobe. Below is a discussion on the advantages and disadvantages of these applications. MOA Image Map in Map Server The map server Web tool is designed to permit rapid browsing of the MOA data set, and provide readily accessible 2D images for figures, field planning, presentations, etc. This tool was created at the University of New Hampshire by Dr. Mark Fahnestock and Mr. Norman Vine. The MOA Web-based map server contains several pre-stretched versions of the MOA data set, allowing a user to enhance the features of interest to their application of the data. Low-contrast features, such as subtle ice topography on ice shelves, are best revealed by the moa_uhc (ultra-high contrast) stretch (Figure 1). High-contrast features, such as mountains and valleys, are best represented by the very low (moa_vlc) or ultra- low contrast (moa_ulc) stretches of MOA (Figure 2). MOA Image Map in Virtual Globes The creation of MOA data files for use in virtual globes allows for greater exploration of the data set and an increase in potential applications. Data can be rendered in 3D (Figure 2a and 2b), viewed at customized extents and even flown through. NSIDC created a KML file and a GeoTiff file for input into Google Earth™ and ArcGlobe respectively. Google Earth™ is a user friendly virtual globe application available as a free download. The application only displays KML files. KML files require that source data be provided in a lat-long grid and does not allow for user defined topography. Google Earth™ does not correctly display MOA data which was converted from a polar stereographic projection (Figure 1a and 2a). Artifacts and distortions appear in increasing severity approaching the pole. Considerable effort and multiple manipulations were required to convert the input data into a KML file. The 16-bit input image was resampled to an 8-bit image required by KML with a single stretch of 15096-17283. The image was run through Google’s Reginator which took several hours to process and produce a KML. ArcGlobe is another virtual globe application which has a more technical interface and requires a license. It does offer several advantages in both the creation and visualization of virtual globe data. The robust application allows for the input of KML files as well as 16-bit data in GeoTiff format. Users can define projections and topography. A 1 km resolution Digital Elevation Model (DEM) derived from ERS 1 and 2 Radar Altimetry and ICESat Laser Altimetry was loaded into ArcGlobe. This DEM allowed for more accurate 3D rendering than the lower resolution DEM used in Google Earth™. ArcGlobe performs pyramidization of the image and DEM in considerably less time than the Regionator. The image can be custom stretched on the fly by the ArcGlobe user. Figure 2 MOA surface image, ultra-high contrast 15875 – 16125 Figure 1 MOA surface image, ultra-low contrast 1 – 32000 Vertical Exaggeration of MOA Image Map Vertical exaggeration can be used to emphasize subtle changes in a surface which is helpful for viewing the relatively flat Antarctic landscape in 3D. ArcGlobe allows for 10 levels of vertical exaggeration (Figure 2) compared to three levels in Google Earth™. Flying through the image provides another perspective in which to explore the Antarctic map (Figure 3 and 4). Figure 1a MOA Image map displayed in Google Earth™ Figure 1b MOA Image Map displayed in ArcGlobe Figure 2a MOA Image Map displayed in 3D in Google Earth™ Figure 2b MOA Image Map displayed in 3D in ArcGlobe Figure 2 MOA Image Map displayed in ArcGlobe with level 10 vertical exaggeration. Figure 4 MOA Image Map displayed in ArcGlobe with level 10 vertical exaggeration. View from Byrd Glacier. Figure 1 MOA Image Map displayed in ArcGlobe without vertical exaggeration. Figure 3 MOA Image Map displayed in ArcGlobe with level 10 vertical exaggeration. Looking toward the Ross Ice Shelf. http://nsidc.org/moa http://nsidc.org/data/virtual_globes