ACADEMIA

Katie Yurkewicz, Science Grid This Week -- The Earth is being systematically scanned. From high above, lasers are being used to create a three-dimensional, high-resolution map of every nook and cranny on the Earth's surface. It might sound like a plot from science fiction—a mad scientist scanning the planet for nefarious purposes, perhaps—but these maps are science fact. Created using LiDAR, an acronym for Light Detection and Ranging, the high-resolution maps are being used today by many (perfectly sane) scientists to study everything from flood and earthquake hazards to vegetation density. Transforming vast LiDAR data sets into useful topographic maps requires sifting through billions of three-dimensional data points and running computationally intensive algorithms. Researchers from the Geosciences Network are working to make this process easier for a larger number of scientists using a data processing workflow accessed through a Web portal. LiDAR technology determines distance to an object using laser pulses. To map the surface of the earth, special equipment mounted in an airplane directs a series of laser pulses at the ground. The pulses reflect back to the airplane almost instantaneously, but with slightly different time lapses depending on the height or composition of the features on the surface. Three digital terrain models for a portion of the Northern San Andreas Fault LiDAR dataset near Fort Ross, California. On the left, a hillshade image produced by imposing artificial illumination on the topography. The height of the vegetation canopy is indicated by the shade of green; darker green indicates taller trees. In the center is a bare earth terrain model for the same area. On the right is a bare earth terrain model with the active trace of the San Andreas fault shown in red. Image Courtesy Christopher Crosby "If you know the travel time of the laser pulse and the position of the plane, you can say where the ground is," says Christopher Crosby, a graduate student at Arizona State University. "The plane is flying along at over one hundred miles an hour, blanketing the ground with tens of thousands of pulses per second, and can acquire huge amounts of data in very short periods of time." A 400 km² scan can result in a data set of over one billion points. But the challenge doesn't come from the raw size of the data set, which may only measure a few terabytes. The challenge is providing scientists with interactive access to the LiDAR data, and with control over the algorithm used to turn that data into a map useful for geoscientists—a digital elevation model, or DEM. "Some commercial software packages exist to construct DEMs from significant volumes of LiDAR data, but they tend to be a big investment," says Crosby. "Many geologists don't have either the computing resources or the finances to use them." Over the past year Crosby and colleagues at ASU and the San Diego Supercomputer Center have built a proof-of-concept version of a system to provide scientists with interactive access to four LiDAR data sets. Scientists use a Web portal to choose their data, DEM algorithm and processing parameters, and then the GEON LiDAR Workflow takes over. Each of the billions of LiDAR data points is stored in a database on the DataStar machine at SDSC. The workflow extracts the user's data, bundles it with the processing parameters and sends the request for computation to a cluster at ASU. "The workflow acts as the middleware between the portal and the grid infrastructure coordinating the distributed grid resources—the database, computation resources and visualization services—through the data analysis pipeline," says the SDSC's Efrat Jaeger-Frank. All scientists can now access the four LiDAR data sets through the GEON portal, and over 100 researchers have already signed up. Crosby, a student in geomorphology, has started using the portal to study how erosion and tectonic forces shape the landscape. Work continues on the workflow and portal, to increase its usefulness to researchers and make it easier for others to link their LiDAR data sets to the portal. A monitoring system is currently under development, and researchers hope to parallelize the computation of the DEM algorithms so that clusters at other GEON institutions can also be used. But most importantly for many researchers, developers are hoping to export the LiDAR database technology to other institutions, which can then expose additional data sets through the GEON portal. "The data is the big draw," says Crosby. "The tool set is pretty stable and works reliably, and we hope this will develop into a resource for the whole geoscience LiDAR community." For more information visit the GEON Web site, or read the publication "A Three Tier Architecture to LiDAR Interpolation and Analysis" by Efrat Jaeger-Frank et al, in the ICCS 2006 proceedings.