Airborne: Next-Day Delivery of Airvborne Multispectral Data
By Marcia E. Wise, Elizabeth Valenti and Joan Davis

If a tornado struck your hometown and you needed multispectral imagery of the devastated area immediately to assess the damage and to assist the relief efforts, could you make one phone call and receive the data within 12 hours? If you're a data provider and you received such a call, could you deliver georeferenced, annotated imagery to your customer the next day?

Demonstrating the Possibilities
If you order multispectral data from a major commercial data provider, you might wait three to six weeks for the data to arrive. But many end users can't afford to wait that long for their data and are looking for alternative providers and immediate data delivery. In order to test whether one-day data delivery is a viable goal, a dedicated group of remote sensing specialists working under NASA's Commercial Remote Sensing Program gathered at Stennis International Airport in southern Mississippi on Aug. 2, 1995 to demonstrate and document the possibilities and pitfalls inherent in executing quick-turnaround data acquisition missions. Team members hoped to discover how quickly multispectral data could be collected and processed and to see where gaps in the existing technology would slow down the procedure. The demonstration was planned and executed by employees of NASA and Lockheed Martin at the John C. Stennis Space Center in Mississippi; TerraSystems Inc. of Hawaii; and Seattle Research and Training of Washington.

Findings
The groundwork for this demonstration - assembling the necessary equipment and personnel, calibrating the sensor, making arrangements with an air service to fly the sensor - was laid over the course of several weeks prior to flying the mission. Once the groundwork was in place, the process of plotting flight lines, collecting aerial and ground-truth data, processing the information collected, and delivering hard-copy images to the "customer" was completed by the demonstration team in only 12 hours. Team members agreed that flexibility and communication were two vital components to accomplishing a quick-turnaround data acquisition and processing mission and were required to act on these principles numerous times throughout the day. By actually performing a quick-turnaround data acquisition mission, demonstration team members received a crash course in the hazards encountered by data providers, learning the infinite number of variables that might break down to slow the process and discovering where the available technology didn't perform up to their needs.

Lesson 1. Equipment Will Break Down, So Be Ready to Implement a Backup Strategy
TerraSystems' Digital Multispectral Video (DMSV) sensor system has four cameras that are precisely aligned toward a single location about a kilometer away from the aircraft. According to Dr. Jonathan Gradie of TerraSystems, "When we shipped it from Hawaii, it got knocked out of alignment; it was looking slightly 'cross-eyed.'" Rather than taking the camera head apart and painstakingly aligning the components, Gradie and Dr. Pamela Blake of Seattle Research and Training coregistered the images in the data preprocessing stage, eliminating any potential alignment problems.
      For this demonstration the DMSV was mounted in a Cessna 206 airplane belonging to Aero-Data Corp. of Baton Rouge, La. The airplane's gyro-stabilized camera mount caused Gradie some difficulty. "Aero-Data has a brand-new Zeiss mount that is beyond the state of the art; it worked yesterday and not today...we unfortunately didn't have any way to perform pitch control," said Gradie. "So what we had to do before we took off was to tilt the airplane up to fool the mount into thinking that it was pitched a certain way electronically and then lock it there."

Lesson 2. Compatible Components Sometimes Aren't
The camera mount was only part of Gradie's equipment problems: "We found that GPS systems, although they're commonly used, are not as integrable as we thought." TerraSystems' Dick Lunn and Gradie connected the aircraft's GPS system to the DMSV. Although the systems appeared to be compatible during a test flight, the connection didn't function on the day of the demonstration. To compensate for this malfunction, the team added a preprocessing step to find the center of each frame and to register it to a topographic map. According to Lunn, TerraSystems will probably acquire its own portable GPS system to resolve the compatibility problem.
      The demonstration team also originally planned to telemeter the data collected by the DMSV from the plane to ground personnel, but the telemetry systems available were costly and their installation would have required permanent modifications to the aircraft. Instead the team had to wait until the aircraft landed at Stennis International Airport to remove the DMSV computer and download the collected data. Lack of a telemetry system affected the ground team's data-collection strategy. According to Rodney McKellip of Lockheed Martin, who led the ground team, "What we needed was a telemetry down-link on the plane so as soon as they acquired the imagery, they could shoot it to us and we could take our points based on visual information. Because we didn't have specific targets, it was difficult for us to know whether the ground control points we were collecting in the field were going to be included in the imagery. That did cause some difficulty in the georeferencing because we had only about a 60 percent success rate of our ground control points actually falling into the frames that were selected for processing." In two Chalmette frames, only two of the four GPS points collected actually fell within the frame. Ultimately, the post-processing team used DeLorme software to find the additional points necessary for processing.

Lesson 3. Ideas that Sound Good in Theory Should be Physically Tested
On the ground, the data processing team ran into a minor snag with the preprocessing computer system. Rather than bringing a separate computer system from Hawaii, Gradie had decided to load the preprocessing software onto the DMSV computer, a 486 with 8 megabytes of memory. When they performed a system test in Mississippi, though, they discovered that some very critical components needed 16 megabytes of memory. "We never tested our software on the exact system we're using here, and it turned out to be significantly different," says Blake. "But we've learned our lesson: next time we'll have a system that's at least as powerful as our laboratory system and probably more so."

Lesson 4. Satellite Availability Isn't Guaranteed
The ground-truth team encountered a few difficulties while trying to collect GPS ground control points for the Slidell study site. "We started having trouble with satellite availability, which means we had to sit at each position as much as 20 minutes," said McKellip. "By the time we got to Slidell, fewer satellites were in view and we were in locations that had heavy tree cover and other obstructions, so we didn't really have a good view of the sky. There were times in Slidell that we were able to track only four or five satellites, and for that reason we had to stay at each station longer. Once we lost lock on four satellites and had to start the whole survey over."

Lesson 5. The Weather Doesn't Care About Your Schedule
Weather conditions play a major role in any airborne data collection mission. This demonstration was originally scheduled for Mon., July 31, 1995, but the remnants of tropical storm Dean caused cloudy skies and isolated showers over the Gulf Coast on Monday and Tuesday. Lockheed Martin Meteorologist Stan Woolley tracked the progress of Dean and then Hurricane Erin's emergence into the Gulf of Mexico: "By Tuesday evening, most of the coastal region was cloud free because the western edge of Hurricane Erin, which was approaching the east coast of Florida, caused subsiding air and more stable atmospheric conditions. Wednesday morning proved to be an ideal weather day for flying - clear skies with no ground fog or haze. We hit it just right between the two tropical systems." By Wednesday afternoon, the National Hurricane Center in Miami extended the Erin hurricane warning westward to include New Orleans, causing a bit of anxiety for the demonstration team stationed at Stennis International Airport. Stennis Space Center employees were instructed to secure their work areas and were released for the remainder of the day, canceling the scheduled backup data post-processing.

Conclusions
By flying this demonstration mission and producing georeferenced, annotated imagery in just over 12 hours, team members proved that a next-day delivery time is a realistic goal (weather permitting), both for end users desiring information quickly and for data providers with the necessary groundwork of sensor and computer equipment, personnel, and air service in place to offer quick-turnaround data acquisition and processing. However, the gaps in the existing technology - unavailability of a portable and affordable telemetry system, non-standardized interfaces for mounting equipment, incompatible GPS data formats, and other gaps that likely exist but were not encountered during this demonstration - impede the data provider's efforts and must be addressed before next-day delivery of data products can become the industry standard.

About the Authors:
Marcia Wise is a writer/editor for Lockheed Martin Stennis Operations in support of NASA's Commercial Remote Sensing Program Office at the John C. Stennis Space Center in Mississippi. Elizabeth Valenti is the EOCAP-Technology Technical Lead for Lockheed Martin Stennis Operations in support of NASA's Commercial Remote Sensing Program Office at the John C. Stennis Space Center in Mississippi. Joan Davis is the EOCAP-Technology Program Element Manager for NASA's Commercial Remote Sensing Program Office at the John C. Stennis Space Center in Mississippi.

Acknowledgments
Thank you to Gene Phillips and the staff of Phillips Aviation at Stennis International Airport. Their cooperation and support made this demonstration possible and was greatly appreciated. This work was supported by the NASA Office of Space Access and Technology, Commercial Remote Sensing Program Office, John C. Stennis Space Center, Mississippi.

Back