    
|
|
What causes the fierce hurricanes of the 20th century and this year's massive El Niño in the Pacific? Are they natural forces beyond our control, or are they intensified by our activities? The importance of the answer extends far beyond scientific interest alone—it will play a crucial role in decisions that affect our nation, our economy and our relations with other countries.To answer such questions about the environment, NASA has embarked on the Earth Science Enterprise, a long-term program to understand the Earth as a coupled system of ocean, atmosphere and land. The Earth Science Enterprise's series of Earth Observing System (EOS) satellites will soon begin returning over a terabyte of data per day-equivalent to the information stored in the Library of Congress-to repositories at several locations around the country. Managing the data deluge To help scientists analyze the massive amounts of information streaming in from the satellites, Kannappan Palaniappan of the University of Missouri at Columbia Computer Science Department and Fritz Hasler of the NASA Goddard Space Flight Center have developed a visualization analysis tool called the Distributed Image SpreadSheet (DISS). DISS creates images from the mountains of data and moves those images over high-performance networks to the scientists' desktops. There they can see, manipulate and analyze the information. The ability to move these data-intensive images to the desktop presents a tremendous challenge to networks such as the NASA Research and Education Network (NREN), NASA's high-performance wide area network. "NREN and the Earth Science Enterprise have been working closely for several years," said Christine Falsetti, NREN Project Manager. "We collaborate on an ongoing basis to look at advanced networking." DISS retrieves a variety of data quickly from multiple, distributed remote centers, then in near-real-time displays the product images for analysis at the scientists' desktops. The network has to ensure extremely low delay times, as well as access to the sites involved. A picture is worth a billion bits "In the 1970s, we could analyze kilobytes of data, which is comparable to a telephone conversation." said Palaniappan, discussing the magnitude of data involved. "In the '80s we could handle megabytes, roughly comparable to the rate at which TV channels broadcast. In the '90s we are receiving gigabytes, and in the 2000s we will be receiving terabytes." DISS is helping scientists make sense of the data deluge. For example, scientists have been able to perform advanced visualizations of very large datasets such as the comparison of almost 12 years of Total Ozone Mapping Spectrometer data. In the future, tools such as DISS will allow scientists to move quickly among images to analyze the large amounts of data produced by next-generation satellite systems. DISS is a prototyping research experiment that was carried out for the first time in June 1997 at the National Center for Atmospheric Research (NCAR) in Boulder, Colo., at the Global Observation Information Network workshop, a U.S.-Japan bilateral initiative for cooperation in Earth observation applications. The user workstation at NCAR accessed data remotely at Goddard Space Flight Center (GSFC), about 1500 miles away, via a high-performance wide area network. The wide area network connectivity included an NREN connection to NCAR by way of the San Diego Supercomputing Center, and fiber-distributed data interfaces for the server and client. "This research experiment shows improvements from stage to stage in the ongoing collaboration between NREN and the Earth Science Enterprise," said Richard desJardins, NREN EOS prototyping network applications manager. "In the first-stage experimental configuration at NCAR, we solved connectivity issues but exposed issues concerning the workstation operating system, specifically the tuning of the network file server. As a result, we gained a clearer understanding of latency (delay) requirements on the network and configuration of the workstation software." The DISS application experiment was carried out again in September 1997 to address the issues raised earlier, in particular the issue of delay. This second-stage experiment was performed at an NREN workshop at NASA Ames Research Center (ARC) in Moffett Field, Calif. using asynchronous transfer mode (ATM) networking end-to-end. The ATM connection between GSFC and the ARC workshop dramatically improved performance of the DISS by a factor of ten over the configuration used at NCAR. Because of advances made in both networking and scientific visualization, decision makers are better able to formulate policies that will have a tremendous impact on our future. The Earth Science Enterprise is part of a Global Change Research program that has many international partners. The Montreal Protocol, for example, established a schedule for phasing out chlorofluorocarbons because of their effect on ozone in the upper atmosphere. There are also proposals to restrict the amount of sulfur dioxide emissions that companies emit, and studies are being conducted to see whether such a system will work for limiting carbon dioxide emissions. "To make policy decisions based on such complex phenomena," said Palaniappan, "we need to collect and analyze data and develop models. Visualization analysis tools allow these activities. Data collection and storage is an inherently distributed activity. Many countries are involved in data collection and archiving, and the U.S. has many national centers where data are collected and archived. The Earth Observing System alone has eight centers, called the Distributed Active Archive Centers, each with data on a different subject such as the ocean, atmosphere or Earth." At present with improved network performance and new analytical tools, users have access to several data repositories at once over networks. Mining NASA's treasure "These data constitute NASA's treasure-trove, which is now publicly available," said Palaniappan. "The Distributed Image SpreadSheet tool allows visualization, analysis and manipulation of 2-D satellite data, 3-D model data and time-varying datasets in a single environment, not only of real-time data but also of historical archives." Visualization tools combined with high-performance network connectivity already facilitate public and commercial use of meteorological datasets. For example, broadcast weather services such as NBC's WRC-TV in Washington, D.C. use National Oceanic and Atmospheric Administration (NOAA) and satellite imagery processed by NASA on a daily basis. An example of real-time use occurred during Hurricane Luis when data were collected at GSFC, analyzed using the DISS at GSFC and distributed by NASA Select TV to the public. Information was also available on a web server at GSFC for public access, and the visualizations were shown nationally on ABC's Evening News. Visualization tools are also important to other agencies involved in data analysis and modeling for aeronautics, astrophysics, geophysical studies, environmental monitoring and mapping. An interactive future Palaniappan foresees new methods of interacting with computers that will provide a virtual environment to NASA's wealth of knowledge. "These tools will become more collaborative," he said, "so students and scientists will be able to interact more closely at both the university and the kindergarten through twelfth-grade levels." As an example, he cited the GLOBE project that opened up accessibility to some weather-related datasets to nearly 4,000 K12 schools around the world. GLOBE is a joint program among NASA, NOAA, and the National Science Foundation to promote environmental awareness and to introduce a much richer environmental curriculum into the K12 level. DISS is one of the tools being expanded to look at the student GLOBE data now. Discussing possible scenarios on the horizon, Palaniappan said, "There will also be more virtual environments for interacting with computers. You won't even need a keyboard as an input to your computer. You'll be able to point, speak and receive spoken responses. The computer's camera will provide the interface between you and your computer by observing where you are pointing and what you're looking at. These are the new paradigms being explored in the area of virtual environments. There are virtual worlds you can explore online... as long as you come home. "In the future, you will have terabytes of data at your fingertips in your homes. You won't have to know beforehand which data repository has which data. All the wealth of knowledge that humanity has collected will be available to everyone 24 hours a day. Because everything is interrelated, it allows you to think at a system level on a much broader scale. The average user won't have to worry about the complex steps required to design and build the satellite, launch it, build the ground receiving station, and collect, validate and archive the data. "We've been doing these very detailed steps over the last 25 years. It's only now that we have both enough variety and volume of data and the tools to understand complex phenomena. Those are the steps that have gone into collecting this treasure of data. Now you can begin to ask questions about the long-term natural variations in our environment and how they are influenced by human activity."
NREN's challenge is to research and develop ways to handle NASA's exploding requirements that demand increasingly sophisticated and powerful networking capabilities. Scientists now require not only greater bandwidth and speeds, but also guaranteed performance, low delay, reliability, quality of service and security over widely differing platforms and users. The results of NREN's ongoing program will ultimately be applied to NASA's existing networks. This will allow scientists to routinely retrieve and visualize massive datasets, remotely control instruments and rovers, and remotely conduct medical examinations and consultations on Earth and in space.
|
