DGPS outages

The Coast Guard’s differential GPS (DGPS) enhancement is the most accurate electronic aid to navigation available to general users. However, DGPS has significant reliability problems. The Coast Guard’s differential GPS system has reached its “initial operating capability.” Yet after more than a year at IOC, the DGPS system continues to provide poor availability. To put this in context, we must note that the Coast Guard system is not at “full operational capability,” and that the Coast Guard has issued several warnings cautioning mariners about that fact.

The Coast Guard DGPS system uses high-quality GPS receivers with antennas at known locations, along with computers to determine the minute-by-minute error from each satellite. They broadcast the corrections over modified radiobeacons; navigators having suitable GPS receivers and special radiobeacon receivers can apply the corrections and get excellent accuracy.

Depending on several factors, including receiver quality, DGPS fixes usually fall within a two- to 10-meter radius given a stationary receiver. That is very high accuracy. Only private, survey-grade systems have met or exceeded this accuracy. DGPS is two or three times as accurate as military GPS. But accuracy is only part of the picture; the DGPS system has yet to come up to its availability standard. The Coast Guard has a goal of 99.7% usable time for most of the system; in critical harbor areas, the goal is 99.9%.

Analyzing the Coast Guard Navigation Center data for November and the first 17 days of December 1996, we find that the average usable time is about 96.3%, ignoring the experimental stations at Alexandria and Wildwood, and two stations listed as “off air”: St. Louis and Youngstown.

Two more stations have continuing problems: Aransas Pass, Texas, was at reduced power for about a month, and Upolu Point, Hawaii, was off air all of December. If we drop these two stations, the remainder have operated about 98.2% of the time. Even by this overly lenient method of excluding badly performing stations, the unavailable time is still about six times as much as planned.

A few days after our analysis, the 120-foot DGPS/radiobeacon tower at Cold Bay, Alaska, failed. It usually takes a month or two to replace such a tower. Analysis of a similar period this spring showed DGPS signals available about 97.9% of the time.

The first reaction to roughly 98% operation might be, “that’s pretty good.” Compared with other aids to navigation, however, it’s lousy. A loran transmitter typically operates correctly 99.9% of the time, which corresponds to a coverage value of 99.7% for a three-station chain. A GPS satellite typically operates about 98% to 99% of the time, but other satellites in view fill in to give coverage of about 99.7%. Even buoys and lighthouses usually exceed 98% availability. As noted, DGPS performance this past winter had, at best, about six times the planned outage time. There is a complicating factor: the usual DGPS outage lasts for many hours, or even days. In many areas there is no “overlap” for DGPS coverage, so losing one station blanks out differential GPS coverage over a wide area.

Capt. James T. Doherty, commanding officer of the USCG Navigation Center in Alexandria, Va., made the point that DGPS is not fully operational, adding that many of the problems arise from using commercial power for the DGPS stations without immediately available backup. Other problems are due to old radiobeacon transmitters, remote locations, collateral damage from power failures, antenna icing and insulator salting. He said, “We are working very aggressively to improve the system.”

The design of the system, as mentioned above, does not include consistent overlapping coverage. In the northeast part of the country, there are many overlaps. If one DGPS station fails, a navigator can switch to another one. But in many areas, for example most of the West Coast, there are few overlaps. If one station goes down, it leaves a large area without differential corrections.

In addition, receivers have had some problems. There are reports that different brands of receivers react differently to corrections larger than normal. Some apply them, as supplied by the differential system, while others drop all differential corrections and revert to GPS. This is a matter of receiver design, not the signals.

When a satellite’s signal gets so far out of tolerance that it is unusable, the differential system sends a specific and very large “correction.” That is to announce that the particular satellite is unusable. Thus, the DGPS system dramatically improves GPS “integrity,” or freedom from incorrect information. The receivers that the Coast Guard is using, however, failed to recognize the “do not use” significance of this message, and plotted positions that were wrong by a matter of miles. It takes a software revision to correct this receiver problem.

When DGPS is operating correctly, however, it gives a navigator unmatched precision. Just locating an underwater wreck can take some time with GPS or loran, since each of these aids to navigation shows positions that vary significantly with time. It is easy with DGPS. Differential GPS signals at a stationary location vary about one-tenth as much as GPS signals, a tremendous improvement. It is quite practical to put one DGPS waypoint at the bow of a shipwreck, and another at the stern. It’s that accurate; accurate enough to distinguish between waypoints a hundred feet or so apart.

DGPS also gives speed over the ground with great accuracy: about plus or minus 0.1 to 0.2 knots. Again, that’s better than any previous public system. DGPS speed is so accurate that it can even be misleading!

Remember, a boat seldom follows an exact course. Nor does the speed stay precisely even; a boat goes uphill and down, slams into waves, surfs down them, and is carried back and forth by the water motion of the wave. DGPS is measuring the antenna’s motion rather than that of the whole boat. It’s necessary to average the speed and course, not due to deficiencies of DGPS but to the boat’s motion in a seaway.

The other problem relates not so much to DGPS as to charts. Chart surveys are oldolder than most navigators. The coast survey found positions with a variety of methods that were quite good over relatively small areas. Yet these local areas are often slightly out of position with respect to the modern coordinate systems that GPS uses. NOAA has estimated that objects on approach charts may disagree with GPS positions by as much as 65 meters, for example. That presumes that the GPS receiver is set to match the chart’s coordinate system. Overseas, particularly at isolated islands, the error may be several miles. People unaware of them have found out the hard wayby going aground.

There is good news ahead. The FAA is designing a Wide Area Augmentation System (WAAS) to broadcast differential corrections from geostationary satellites. These broadcasts will be on the GPS frequency that civil receivers use, so there will be no need for an additional beacon receiver to get the WAAS corrections. Most GPS receivers will have to be modified or replaced, however.

Recent experiments have shown that the current loran network could actually carry DGPS corrections over long distances with high accuracy and high availability.

Contributing editor Bill Brogdon is a retired Coast Guard captain and former head of the Coast Guard’s office of navigation.

By Ocean Navigator