Safety‎ > ‎

Electronic Seduction Delays Rescue

posted 12 Mar 2013, 06:07 by Support SarMobile   [ updated 21 Jul 2013, 18:50 ]
Electronic seduction occurs when an electronic signal is generated or modified, either naturally or artificially, such that the signal guides a recipient to a location other than that intended by the recipient. The recipient is seduced away from the intended target to another location. 

Mid October 2000 A Piper PA-31 Navajo Chieftain, C-GIPB, departed Yellowknife, Northwest Territories, on a night charter flight to Fort Liard1. One pilot and five passengers were on board. At Fort Liard they encountered moderate to heavy snow. The pilot attempted a non-precision approach with a circling procedure to land on Runway 02. The aircraft struck a gravel bar on the west shoreline of the Liard River, 1.3 nautical miles short of the runway, and 0.3 nautical mile left of course. The aircraft sustained substantial damage, but there was no fire. Three passengers were fatally injured, the pilot and two passengers were seriously injured. The emergency locator transmitter activated and was received by the search and rescue satellite system. Canadian Forces dispatched two aircraft to conduct a search. The wreckage was located by homing the beacon the following morning. Help arrived at the accident site approximately 10 hours after the crash.

Even though the crash was located using the signal from the emergency locator transmitter (ELT), confusion about what the electronic signal was telling the search crews would delay the arrival of help by at least 45 minutes. The search and rescue satellite system (COSPAS/SARSAT) first identified the crash site as being approximately 40 miles south of Fort Liard. A second satellite pass indicated the emergency locator transmitter position was approximately 19 miles south east of Fort Liard. The Rescue Coordination Centre dispatched a Hercules aircraft from Winnipeg, Manitoba and a Twin Otter from Yellowknife. Both aircraft equipped with electronic direction finding equipment that would allow them to home the emergency signal. 

The Twin Otter arrived on scene first and overflew the COSPAS/SARSAT provided location at 3,400 feet above ground level. Due to weather conditions in the area, the Twin Otter was being flown on instruments, Fuel considerations forced it to continue to Fort Nelson, British Columbia before attempting to home in on the emergency transmitter signal. Initial communication with the Twin Otter indicated the ELT was transmitting within two miles of the SARSAT predicted position, based on aural null indications with the very high frequency (VHF) squelch off.1 Here we will have to engage in some educated speculation as to what aural null indications are. If they did not have fuel to begin homing the signal, a process much more efficient than the aural null procedure, they certainly didn't have fuel to complete a canonical aural null. We suspect that what is meant here is this: As the Twin Otter approached the location provided by COSPAS/SARSAT they tuned a radio to the emergency transmitter frequency, 121.5 MHz, and selected squelch off to increase their chances of picking up the signal. As they neared the COSPAS/SARSAT location they began to hear the emergency signal then, some time and distance later, the signal disappeared. Some geometric analysis of these two locations provided a fix within two miles of the COSPAS/SARSAT location.

The Hercules arrived in the same area, near Lake Bovie, NWT, a short time later but was unable to pick up a signal from the emergency transmitter. The COSPAS/SARSAT system had allocated a high confidence to the predicted position. With this in mind, and maybe the report from the Twin Otter, it was believed that the Hercules was in the right area. Rather than fly a procedure to acquire the ELT signal, the crew contemplated performing a descent to visual conditions. This is a procedure that takes the skill and dedication of a professional flight crew, and not likely one that most civilian organizations would even consider. Aircraft approach to land through clouds and poor visibility all the time, but they have electronic navigation systems to guide them along paths that have been surveyed and are known to be safe. This crew was deciding if conditions were right for them to descend through bad weather, without any of the normal guidance, in the hope that they would find good visibility before they came too close to the ground. We are very lucky to have these people available to come rescue us when we need it. They dropped flares in the area for about 45 minutes to see if the cloud base was high enough for the descent. In the end the descent procedure was not carried out. The Hercules then flew further to the west where a stronger signal from the ELT was encountered. The Hercules was able to home the signal to determine the crash location was approximately 1 mile south of Fort Laird. A local civilian helicopter, was able to fly under the clouds to the crash site and render aid. 

So, what happened here? Why was the Twin Otter able to pick up a signal, but the Hercules was not? Why was the COSPAS/SARSAT calculated location so far from the actual crash site? Why should we care? Let's take the last question first.

We care because by trying to understand what happened in this case, we will understand more about the complex nature of performing electronic searches. With this understanding we may be able to avoid similar problems in the future; this is an important aspect of flight safety. It is all too easy to be a Monday morning quarterback, but we don't fault the Hercules crew for taking the actions they did. We know something of what is generally believed about emergency locator transmitter signals, and while these beliefs are broadly true they also contain some subtle misdirection which came to the fore in this case because of the circumstances. One of these beliefs is that the ELT will be located within the confidence circle of the COSPAS/SARSAT calculated location. It is likely this belief, combined with the twin otter crew's indication that the ELT was within two miles of the COSPAS/SARSAT location, that lead the Hercules crew to remain in that area rather than follow a procedure that would have them cover more ground in an attempt to acquire the ELT signal.

Why was the Twin Otter able to pick up the signal, but the Hercules was not? First let's set the stage. C-GIPB was eventually found 1.3 nautical miles from the threshold of the Fort Liard runway, 0.3 nautical miles left of the approach path to the airport. This put it on a gravel bar near the left bank of the Liard River. Here we have a topographic map2 with the crash site marked with a red dot (as in all images on this page you may click on this map to see a larger version):

C-GIPB Crash Site

The river valley and other terrain features provide a complex environment in which to home a signal. With the two aircraft operating in the area at different times and altitudes, and constrained by normal en route navigation accuracy one could easily receive the ELT signal while the other did not. 
To visualize this we need a larger map. Below we see a second topographic map2 of the search area. Fort Liard and the crash site are located just up from the center, Lake Bovie (where the Hercules concentrated initial efforts) towards the bottom right.

 

You will notice the terrain, indicated by the brown contour lines each representing an elevation change of 20 meters, forms a fairly narrow valley. In the same way that the river may be hidden from view at a distance because it is down in the valley, the crash site and ELT may be 'hidden' from the view of an aircraft receiver. Between Lake Bovie and the crash site the terrain rises to 480m before plunging down to below 200m at the river edge. Even if the Hercules was flying high enough to be well clear of the hills around Lake Bovie, the crash and the ELT could have been, and probably were, obscured by terrain closer to Fort Liard. To the left is a photograph [source: Wikipedia] that shows how the Fort Liard runway looks on final. You can see the gravel bar where C-GIPB came to rest in the middle distance at the right edge of the picture. You can also see the rising terrain to the south (left) of the airport.

So why was the COSPAS/SARSAT location so far away from the actual crash site? The accident investigation report points to magnetic interference and previous anomalies in ELT locations but makes no determination as to the cause. The report goes on to discuss how ELT frequency stability affects the quality of the position generated. This is quite correct, but the stability of the ELT on board C-GIPB must have been quite good to generate a high confidence position. Not as good as a TSO-C126 (also known as 406 MHz) beacon, but good enough to convince the Hercules crew the crash was very close to Lake Bovie. The documentation for COSPAS/SARSAT positions we have does not specify a high confidence value but gives a numeric rating from 1 to 4:

 Confidence   Accuracy 95% Confidence Interval
 1  >50 Miles
 2  20 to 50 Miles
 3  5 to 20 Miles
 4  <5 Miles

We assume that the high confidence level from the accident report corresponds to a confidence level of 4 from the chart above; any other level should not have lead the Hercules crew to believe that the crash site was so close to Lake Bovie and limit the search area as they did. All other levels of confidence would have included the actual crash site in the probable location area. To understand this we need to take a detour through refraction and scintillation.

refraction Gritty Refraction Refraction is the change in direction of a wave due to a change in the propagation medium. It doesn't matter if the wave is in a body of water, an acoustic wave in the air, a light wave or a radio wave. As the wave moves from one medium to another the waves are refracted at the interface between the two media causing the direction of propagation to change. In the photograph4 to the left you can see what happens as light propagating from the diagonal lines, through the glass, into the water and back through the glass to the air and finally to the camera is refracted causing a distortion of the lines. A more familiar effect can be seen in the photograph5 on the right. The pencil appears to be discontinuous because the light waves travelling from the pencil to the camera are refracted when passing from the water to the glass and to the air. In order to determine where the lines, or the pencil actually are located we need to know how much refraction the various media are imposing on the light.

This is similar in principle to a COSPAS/SARSAT satellite locating an ELT. As the satellite passes within range of the ELT it receives the radio wave carrying the distress signal. If the radio wave is subjected to refraction, the ELT will 'appear' to be in a location some distance from where it really is. The atmosphere and local magnetic interference can contribute to refraction of radio waves; but the most significant factor is the ionosphere. The ionosphere is a layer of the upper atmosphere that is ionized by solar radiation. The level of ionization varies by location, time of day and solar activity which causes the solar wind. In the arctic and antarctic the level of ionization is further complicated by the same process that gives us the aurora. The earth's magnetic field deflects particles of the solar wind toward the poles where they interact with the molecules of the atmosphere to create the northern and southern lights, and contribute to greater ionospheric refraction and scintillation. With only one satellite measuring the ELT signal from an unknown location, there is no way to find out how much the radio wave is refracted.

For a more detailed look at refraction you can check out this web site.

In these two previous photographic examples, and our discussion of ionospheric refraction we have only considered stable propagation media. The water has been still. Of course the ionosphere is rarely still. Just like the rest of the atmosphere it can be chaotic, vary in density and composition. If you have watched the aurora you know that they are very fluid and constantly changing shape and color. Atmospheric turbulence causes the stars to twinkle when we look at them at night. The phenomenon we call twinkling arises when variations in the atmosphere cause the star to shift apparent location and brightness due to refraction and diffraction. Astronomers call this scintillation. Ionospheric scintillation also affects radio waves transmitted from the earth into space. This causes the apparent location and strength of the ELT signal to vary as received by the COSPAS/SARSAT satellite. Just like a twinkling star, or coins in the bottom of a water fountain (photograph6 to the left) when the surface of the water is disturbed by spray, the apparent position will vary randomly around a central position. This random variation over time contributes to the COSPAS/SARSAT computation of the confidence level. Other things also contribute to this random variation. One important contributor is the frequency stability of the radio signal. That is why a 406 MHz beacons, which have better frequency control, can usually provide better location accuracy than a 121.5 MHz beacon. But, again with only one satellite making the measurements, even a 406 MHz beacon does not give us the ability to eliminate all refraction and scintillation.

To compute an ELT location the COSPAS/SARSAT system makes several position measurements of the signal over time during a pass. Each position will have error introduced by refraction, scintillation, frequency instability, etc. The position measurements are used to compute an 'average' position. Then the distance from each of the measurements to the average position is used to compute confidence value of the location. The formula used to compute the confidence value is designed so that the actual ELT location should be within the confidence distance of the computed location 90-95% of the time. Any error that is random and varies above and below zero will average out. Some errors that are not random can be accounted for in the computation and removed. If there are errors that are not random and can not be removed by the computation they will introduce an offset. Depending on the source of the error this offset can be quite large, in this case 19 nautical miles. The accident report refers to other ELT anomalies in the area between the Yukon border and Great Slave Lake. This area is within what is known as the auroral oval. This is a good indication that the source of this error is probably related to the effect of space weather on the ionosphere.

So after all this what have we learned? This is a case where, with the best of intentions at heart, a team working to resolve a missing aircraft assumed that a COSPAS/SARSAT location accurately locates an ELT. Even when they could not hear the ELT signal, they assumed the COSPAS/SARSAT location and confidence area were valid. In retrospect, these assumptions are not supported by what is known about radio propagation in the arctic. There are procedures to follow when a search resource arrives at a presumed ELT location and can not hear the ELT. The procedures were not followed in this instance because of the earlier assumptions. This is a case where naivete and assumptions trumped procedure. Mistakes will occasionally be made and we assume the RCAF has learned from this and promulgated those lessons. Never the less this incident stands as an example of why electronic search procedures must have a grounding in a good and thorough understanding of the science and technology. We don't expect search and rescue crews to master this knowledge. Indeed we know their time is better spent mastering the specific aspects of their craft, like the descent to visual conditions. Those who would design search techniques and procedures do require a mastery of the science and technology. Unfortunately this is not always the case. In our next article we will look at what happens when naivete and incorrect assumptions are used to create a procedure; and oversight is unable to detect and correct the problems that arise.

 

[1] Transportation Safety Board of Canada, Aviation Report A01W0261
[2] Natural Resources Canada, Atlas of Canada (atlas.nrcan.gc.ca)
[4] Refraction pattern (some rights reserved www.flickr.com/photos/mrmoaks/7788291338/)
[5] Pencil in glass refraction (some rights reserved www.flickr.com/photos/mohtj/436541295/)
[6] Fountain coins stock 4 by caliconcept-stock (caliconcept-stock.deviantart.com)
Comments