Shough Report

1. Abstract

2. General Description of the Radarscope Photographs

3. Frame-by-Frame Description of Radarscope Photographs

4. Radar Specifications and Mode of Operation

5. Reconciling Time and Distance Data

6. Interpreting the Unidentified Echoes

7. Conclusions



Download full
report as PDF

Anomalous Echoes Captured by a B-52 Airborne Radarscope Camera

Martin L. Shough

6. Interpreting the Unidentified Echoes

In Section 5-3. it was found convenient to divide the photo sequence into two phases - Phase A (frames 771-773), the ostensible rapid motion of the blip from off the right nose of the B-52 to a point behind the right wing before crossing to a point off the left wing; Phase B, (776-782), the reappearance of the blip stationed persistently off the left wing. We follow the same convention. The method here is eliminative, an attempt to determine beyond reasonable doubt what the "echoes" are not. Some reflections and conjectures will be offered in Section 7.

6-1. Meteors

Whilst the persistent Phase B echo has no similarity at all to a meteor return, echoes such as those to the right of the aircraft in Phase A could conceivably be due to a meteor or meteors. Meteors generate a high temperature plasma due to ram heating of the air, which can be detected on certain radars. Generally it is the long trail or wake ionisation which is detected, acting as an efficient re-radiator when favourably oriented in relation to the radar. The trail will be scanned as an effective point target on a single sweep because the recombination time of the plasma is very short and the typical flight time is less than the rotation period of most surveillance radars.

In the present case the successive echoes are far apart (~100 degrees of azimuth) and a relatively fast 20 rpm rotation rate means that a single unusually long-lived meteor detected on one scan (frame 771) might still have been within the coverage pattern when the antenna rotated back towards it approximately 3.8 seconds later (frame 772). We can show (see Section 7) that the detail of the echo presentation is not inconsistent with two consecutive echoes from a single fast-moving target passing through the drum, provided that the effective target echoing area for 3cm radar is in the order of several hundred feet long on a major axis alligned with the direction of motion. However unlikely, in principle this could be two returns from the head echo of a very large fireball.

Taken at face value the echo displacement would indicate a maximum speed of 1870 kts (~ 2160 mph), which is between one and two orders of magnitude too low for an ordinary shower meteor and requires a flat trajectory at zero degrees relative elevation. The radar coverage pattern, having a top-edge elevation of only about 8 degrees (a maximum, remember, since the characteristic target for this pattern is a large jet aircraft; see Section 5-2), also implies this: An elevation 8 degrees above flight level requires a detectable1st-trip target at 1.62 miles real range to be at about 11,000 ft or less - i.e., a spectacular slow fireball roughly co-altitudinal with the B-52.

Such a fireball implies an abnormally slow meteoroid that has been further dramatically slowed by tropospheric braking and has a very good chance of surviving to the ground. Could a fire caused by an impacting meteorite, or fragments from an air-detonating meteorite, explain the “landing” and the bright glow observed later at ground level from the B-52? Leaving aside for the moment the fact that this wouldn’t help to explain the immediately consecutive Phase B radar echo, several arguments make this theory very unlikely.

The position of the B-52 when the echo finally disappeared is given in the original reports as 296 degrees radial 14 NM from the Deering TACAN beacon adjacent to Minot AFB runway. The “landing” location to which the B-52 returned and where it overflew the ground light is also given with fair accuracy. Col. Werlich gives this position as “320 radius, 16 NM” from the TACAN beacon, corroborated by ground-visual reports. These locations are some 7 miles apart. The relative position of echo 772 is also known accurately - about 3 miles at 300 degrees from the final Phase B echo position on frame 782 and therefore some 10 miles from the “landing” location. So a meteor on a shallow trajectory (see below) travelling almost 1 mile/second on a heading of about 338 degrees magnetic would have to arrive at an impact point some distance even further away to the NW than this, leading (conservatively) to a minimum distance of maybe 15 miles from the nearest likely impact point to the nearest possible ground-visual position. Given that the B-52 approached to within a mile or two of the location of the ground light at only about 1500 ft this discrepancy seems far too large to reconcile. (If Claude Poher's calculation of the B-52 position at frame 783 is accurate, then the discrepancy between the two locations is even increased by several miles.)

No sign of impact or fire damage was discovered from later helicopter survey of the site, or anything else to explain the structured object seen from the B-52. There is no evidence of reports from farmers or claims of damage, and nothing was recovered. A search of various on-line meteor resources produces no record of a meteorite fall on this date in N. Dakota. Moreover there was no visual report from the B-52 flightdeck of a spectacular fireball streaking past the right wing below the clouds, nor do any of the many ground observers who were watching the skies at that time describe a possible fireball.

So a close-range fireball seems to be ruled out. If we forget the "landing" these problems might be evaded by invoking a reduction of displayed speed due to multiple-trip returns from a remote meteor passing beyond the unambiguous range of the radar. At 2nd-trip distances of ~70 miles slant range the angular rate corresponds to a more reasonable velocity of about 170,000 mph (3rd-trip would double this rate). On the other hand a remote meteor would make a proportionately very poor target (signal attenuation going as the inverse 4th power of the range) becoming problematic in terms of the 3cm wavelength of the ASB-9, which is already two orders of magnitude shorter than optimum for returns from meteor ionisation. Moreover the implied trajectory for first-trip passage through the shallow drum of the radiation pattern is a shallow path only about 20 degrees off a reciprocal heading, therefore a large angle away from the favourable "radiant condition" which occurs with meteor trails orientated normal to the radar line-of-sight.

So once again this implies strong head echoes from a very large fireball-type meteor, rather than the more usual echo from a favourably oriented particle trail, and now the requirement for a large radar cross-section is even more stringent. An angular displacement of 100 degrees between paints implies a meteor detectable over a track length of around 200 miles for several seconds, picked up at 2nd-trip range and on an inefficient wavelength.

Simple geometry shows that a 2nd-trip track detected twice on successive scans at about 70 miles passes within a slant range of about 45 miles from the radar (whilst the antenna is "blind" and rotating through the reciprocal sector). Even though 2nd-trip on radar, this is still a "nearby" visual meteor in the local sky since the vast majority of meteors burn out at altitudes well above 50 miles, and it should have been a prominent visual object low in the SW sky (low elevation angle implied by radiation pattern) streaking westwards at ~ 35 deg/sec for several seconds.

It seems possible that the reported presence of haze (the aircraft was probably flying within or close to haze and/or patchy overcast at the time of the photographs) and a second layer of broken overcast at about 25,000 ft (about 3 miles above the flight level) could have prevented visual observation. However, ground observers apparently were in a position to see Sirius and/or other astronomical bodies in the southern sky according to the Blue Book hypothesis, so the degree of likely obscuration is arguable. Most observers reported seeing some stars. Even with broken cloud cover one might expect that so colossal a bolide would be seen by at least one of the many people watching the southern skies from around Minot, and if not by them then by somebody somewhere in N. Dakota. The incident took place within the time frame of the annual Orionid meteor shower, about 15-25 Oct., and meteor showers are routinely observed by professional and amateur astronomers, but no reports are findable of a remarkable fireball seen during the Orionid shower of 1968.

Taking a different tack, note that we cannot necessarily infer continuity from two or three widely separated paints. It is also possible that two different meteors could be detected on successive scans. From the region of Minot ND the Orionid radiant (RA 92 degs; Dec.15 degs N) culminated at about 50 degrees elevation due south at about 0400 local time on the morning of Oct 24 1968. The typical Orionid rate at maximum is about 20 meteors per hour.

Suppose that successive Orionids pass within only a few miles of the airborne radar and so are detected as first-trip targets. Detection might then occur even at the unfavourable 3cm wavelength of the ASB-9, because although the returned power varies as the cube of the wavelength it varies as the 4th power of the range, and the gain due to very close proximity could outweigh the loss due to short wavelength. Travelling at perhaps 50 miles per second a meteor could pass through a 2-degree radar beam (a few hundred feet wide at the indicated first-trip ranges) in a few milliseconds and a wake echo could be scanned as a short streak at almost any azimuth.

But again, 1st-trip echoes from meteors only a few miles from the radar would still imply large meteors that survived ablation down to below about 12,000 ft. Such meteors would definitely be bright visual fireballs, and now we have two, in startling proximity, instead of one only moderately close by. Second-trip ranges would allow ordinary shower meteors at altitude; but two ordinary Orionids at 2nd-trip ranges are not likely to have been detected on this X-band radar in the first place, so again we are back two fireballs instead of one. This is not an attractive alternative.

In summary the least unlikely meteor scenario to explain echoes 771 and 772 would be second-trip echoes from a single very large fireball, passing within about 45 miles south of the radar on a heading of approximately 290 degrees true.

There is no visual evidence consistent with a fireball, despite large numbers of ground and air observers sensitized to "see UFOs", but this is not conclusive owing to the presence of broken layers of cloud and haze above 10,000 ft. On the other hand, these clouds were apparently not dense enough to prevent observation from the air and/or from the ground of the stars Sirius and/or Vega (according to the Blue Book hypothesis).

A large fireball is a priori an improbable event, and a complicated relationship between speed, mass and the altitude of ablation means that fireballs in the N hemisphere have a maximum frequency in Spring and in the evening. In North Dakota, an early morning hour, in the Autumn, is exactly the least likely time to observe a fireball. There is no connection to the culminating Orionid radiant, which is at this time 30 degrees of azimuth south of the southernmost possible origin of the implied radar track, so we have a pure coincidence with the Orionid shower.

Fireballs are observable over a wide area. I can find no other reports of possible fireballs for the date and area, either in UFO report lists or in meteor observation records. Admittedly the local time would not conduce to large numbers of potential observers; nevertheless the absence of other reports adds to the cumulative improbability of a fireball explanation already coincident with unrelated local visuals and a Minot AFB weather-radar target.

Finally, given the coincidence of an immediately consecutive persistent ASB-9 radar echo (Phase B) which can have nothing to do with meteors, the meteor hypothesis for 771 & 772 should be rejected as highly unlikely.

6-2. Aircraft & Missiles

In this case the explanation is conceivable (in principle) for Phase B but is rather more difficult to apply to Phase A. The implied speed of about 2000 mph between frames 771 & 772 appears to rule out successive paints from one conventional aircraft. There were a few aircraft flying in late 1968 capable of Mach 3 (e.g. the SR-71 or the new Soviet Mig-25), but only at high altitudes many times the radar-implied altitude of under about 10,000 ft. (Invoking multiple-trip echoes is no help in this case as displayed rates will always be slower than true rates.)

This leaves the possibility of different aircraft passing sequentially through the radar cover, each being painted for only one scan. The shortest distance through the complete cover at the displayed ranges would be steeply up or down, normal to the boresight angle.

If we say that the vertical cover is nominally 60 degrees (the actual profile is of course a complicated function of range and elevation defineable only in terms of a probability of detection for a given radar cross-section) and the renewal rate is nominally 3 seconds, then we have the very approximate limit values shown in Table 4 below.

graphic from shough analysis

Table 4. These speeds are certainly lower than the ~2000 mph rate we are trying to explain away, but this result is not very helpful inasmuch as no aircraft could possibly exhibit such rates of near-vertical ascent, or descent, at heights under 10,000 ft - certainly not survivably (Note 9)


Actually because of the near-saturated ground echo filling the scope beyond about 2.1 NM it would not be necessary for an aircraft to pass in and out of the entire radar cover (slant range 5 NM) during one scan. The shortest path in and out of the ground echo would in each case be the chord passing through the echo position at right angles to the radius. For 771 this path is 2.45 NM long, and for 772 it is 3.66 NM long, allowing aircraft to be painted only once whilst passing through in either direction, re-entering ground echo after total sweep rotation angles of 395 degrees and 420 degrees respectively, at about 1633 mph (1420 knots) and about 2128 mph (1850 knots) respectively. But this is a highly artificial hypothesis since it requires each aircraft to be painted at the middle of its track and each track to be at right angles to the line of sight, so these rates are improbable minima, and even so they are still excessive - target 782, particularly, is now even more of a problem than it was before.

On the other hand if the tracks are radial then an aircraft could travel directly outwards from echo position #771 into the surrounding ground echo on the shortest path of about 0.35 NM in 3 secs at a speed of only about 480 mph (450 knots), and another aircraft could travel 1.05 NM outward from #772 into the ground echo in 3 secs at about 1450 mph (1260 knots). These figures look better, but they are not really. An aircraft presumably has crossed the scope diametrically to reach the start points of these radial tracks in the first place, and therefore should have appeared in the opposite sector of the same scan 1.5 seconds before. This is not an issue for 771 inasmuch as the photo exposure does not record this scope sector; but once again for 772 this only exacerbates our problem, leading to a minimum average speed (assuming level flight) of about 2900 mph (2520 knots).

We can suppose any arbitrary kinds of circuitous climbs and dives to try and evade these issues, but the result becomes more contrived and improbable. In general, two aircraft must have come from and gone to somewhere, and this activity was taking place close to the terminal manoeuvring area of a SAC air base with a B-52 positioning itself for final approach, limiting the plausibility of the idea that aircraft might have performing manoeuvers at high speed in the vicinity. According to the AFR 80-17 report telex to Blue Book:

j. Location, approximate altitude, and general direction of flight of any air traffic or balloon releases in the area that might account for the sighting.
Basic Reporting Data (TELEX 0008-4,5)

All of the ASB-9 echo positions indicate altitudes no more than a few thousand feet above or below the B-52 which, even if marginal for the ASB-9 coverage, should have been well inside the coverage of any airfield surveillance radar. An unidentified aircraft flying below the B-52 could be unresolvable on the PPI of a ground radar from the echo of the B-52 itself, but the aircraft has n the RAPCON transcript) of a contact with one unknown target near the B-52 on the Minot weather radar for an unknown duration prior to the airborne radar contact, there is no evidence in the file indicating the presence of any aircraft that Col. Werlich, SAC or Blue Book are inclined to acknowledge.

Notwithstanding that there was "no other [known] air traffic", the lighted object seen either close to or on the ground from the flight deck suggests that the object could have descended below any radar cover, and even landed, consistent with what the official report describes as "a simultaneous ground sighting [of an apparent landing] in approximately the same location." But this could obviously not have been a fixed-wing aircraft unless it crashed, and according to Col. Werlich daylight reconnaissance from the air revealed nothing in the area. Covert recovery seems highly implausible. So this leaves the remote possibility of an incursion by an unidentified helicopter.

According to ground observer SSgt. James Bond at N-1 "the object acted like a helicopter". But during the air radar episode the minimum altitude of a helicopter in the radar cover would have been (slant range about 5000 ft and assuming a maximum depression angle of -52 degrees) about 4000 ft below the B-52, or a similar altitude above the ground. For long periods during arrival, pacing and descent a helicopter should have been inside the ground surveillance cover, and helicopters with their bulky geometry and the large swept-area of lift and tail rotors tend to be very prominent targets. Also, 250 mph is extremely fast, and although several ground observers clearly heard jet noise from the B-52 at some thousands of feet, there were no reports of rotor noise from a low-level helicopter. In short, known fixed wing or rotor aircraft seem very implausible.

There remain the remote possibilities of a) one or more experimental stealth vehicles with radar aspect ratios designed to minimise ventral ground radar cross-section but still detectable side-on or in dorsal plan from the relative altitude of the B-52, or b) deliberate spoofing using small unmanned jet drones with vanishing radar cross-sections, augmented by onboard active jamming against the B-52 radar to explain why the ASB-9 Phase B echo was said to be "larger than a KC-135 tanker" or comparable to another B-52. It might be consistent with this that unusual responses were also claimed to have been detected on ECM gear in the plane (although this is a second-hand report uncorroborated by the plane's EW officer) at the same time as its two UHF transmitters were blocked (see also Section 6-6 below).

The state of the art in secret experimental stealth techniques in 1968 is not known to this author. Presumably an early full-scale concept demonstrator of a stealth design is a possibility. Remote regions of N. Dakota were apparently used for test flying and special tactical training, and what is known as an "oil burner" run for high-speed low-level flights was reportedly maintained in the Montana border area west of Alexander, where SR-71 trials were conducted. This run is over 100 miles W of the sighting location however.

Ground observers near Minot generally reported bright lights, or a "wiener-shaped" object; but in one case an observer looking directly overhead described an object looking "similar in outline to a stingray fish" accompanied by jet sounds steadier and lower pitched than a normal engine. This is intriguing; but cruising "real slow when overhead" at low altitude and generally behaving "like a helicopter" does not suggest any known fixed-wing jet. Would any experimental stealth vehicle be flying around over an ICBM missile farm, brightly lit, in full view of many potential observers? No protoype VTOL version of the stealth fighter is known to have been developed, and presumably the crash of such a vehicle would spark a major incident.

Perhaps a small RPV is more likely. Many and varied military RPV programmes did exist in the US during the late 60s, Army, Navy, Air Force and private venture, and some of them have exotic-looking shapes. Teledyne Ryan, for example, had been building a prolific series of pilotless jet targets and ELINT platforms since 1951, and some carried multiple jammers as well as other ECM devices. The B-52 itself was designed to carry several small pilotless ECM decoys. From about 1960 until the late '70s the decoy used was the ADM-20 Quail, which could fly for 30 minutes at up to 500 mph whilst using onboard ECM equipment to simulate the radar signature of a B-52.

If one or more experimental RPVs had been deployed during a covert test of the B-52‘s EW systems they might have used directional active jamming tailored to the ASB-9, and possibly basic stealth techniques such as non-metallic construction and radar-absorbent paints to aid in suppressing their already-small ground radar skin-paints. (This brings to mind the apparently mobile small "negative echoes" in the ground return behind the B-52, discussed as features #2 and #3 in Section 5-1 above. Stealthed targets might show up as holes in the ground return. But there is evidence that #3, at least, must be a system artifact.) Such a small and/or stealthy object could be a marginal target for ground radars whilst generating a very large false ‘echo’ on the airborne radar. If an experimental RPV had failed and crashed near Minot it could account for the bright orange glow on the ground seen later from the flight deck.

But this is a rather desperate speculation. After a 10-hour flight the B-52 crew were preparing to make a final approach for landing, systems winding down, ECM gear not operational (according to both the contemporaneous Air Force report and the EW Officer, who remarked that he was probably taking a routine nap at this stage of the flight!) and the pilot evaluation flight virtually over. This is an odd moment to choose to begin such a potentially risky deception, and an odd location, too, in the midst of the Minuteman missile field. And as mentioned, Col. Werlich searched the reported landing area from a helicopter finding only empty farmland, "nothing there that would produce this type of light". How, when, why, and by whom, would an RPV have been recovered in secrecy from the area under the noses of numerous Minuteman security and maintenance teams and SAC investigators?

6-3. Precipitation

The short 3cm wavelength of the ASB-9 radar makes it more likely than typical S-band surveillance radars to detect a sufficient density of small precipitation particles. Thick haze and broken cloud is reported above about 10,000 ft and there are indications of increasing noise speckling on the PPI which could be caused by weather. But such weather cannot explain extremely anisotropic and compact echoes of the strength observed.

It is true that hail showers especially have been observed on radar to form in quite discrete short-lived cells of an order of size not much above the likely first-trip resolution cell in this instance (the resolution cell for a 1.6-degree wide 0.25 microsecond pulse at a range of 2 miles would be about 300 feet in azimuth by 123 feet in range). However a hail cell would not normally form in such extreme isolation, and as mentioned the USAF weather report states that there was no precipitation in the area, let alone the thunderstorm updrafts with which hail showers are generally associated.

Multiple-trip returns from an intense storm beyond the 67.5-mile first-trip unambiguous range of the radar, in the vicinity of the Turtle Mountains massif, could conceivably explain the persistent Phase B target off the left wing, since the angular displacement of a point 70 miles away due to the ~2.5-mile travel of the B-52 during the photo sequence is very small (see also Section 6-11 below). The vertical recirculation of hail cells lofts the particles to altitudes of many thousands of feet (up to 60,000 ft in some cases) and the large vertical extent of echo is characteristic of precipitation. A large storm with hail might conceivably produce a broad echo (the angular width of the #773 echo corresponds to a breadth approaching about 8 miles at the second-trip range) with also a noticeable extent on the range axis due to the vertical height of the storm, which could be as great as 10 miles. The range differential between the top and the base of such a storm from the B-52 altitude of about 1.5 miles would be in the region of 500 feet, which is several times the range resolution of a 0.25 microsecond pulse (123 ft) and might be detectable in principle, but only barely in practice, corresponding to less than 1% of the scope radius (only about 3 mm on the scale of the scope images measured in Section 5-1) when typically about 200 spot diameters might be resolvable along the PPI radius. The radial extent of the #773 echo is approximately 4 or 5 times as large, so much too great to be accounted for by the vertical development of any possible terrestrial storm at 2nd-trip range or greater.

Smearing of echoes on the range axis by ghosting, caused by radiation returned to the antenna by two ray paths of different lengths in (hypothetical) unusual propagation conditions, could account for this degree of radial ellipticity and/or apparent doubling of the persistent Phase B echo. But it seems likely that echoes received in this way, from multiple-trip distances and also via lossy scattering pathways, would require a rather efficient reflector. Therefore types of targets other than weather (for which there is no evidence) might be better candidates for ghost reflections (see Section 6-10 below).

6-4. Echoes from the moon

At first sight this might seem an extraordinary notion for a 250 kW peak power airborne radar, delivering a mean power of a little over 100W in Station Keep mode (0.25 microsec. pulse, 1617 roject Diana) at Ft. Monmouth in 1946 with a continuous wave signal of 3 kW, but they were often observed on early surveillance pulse radars with peak powers of tens/hundreds of kW so it should be considered.

The 0.5 degree diameter of the moon will behave like a point target in azimuth. Because it is smaller than the azimuth resolution of the radar it will present an echo with an angular width comparable to the width of the main beam, appearing at the azimuth of the moon. But because the radar integrates echoes from the entire lunar hemisphere it is not a point target on the range axis. It is also obviously a multiple-trip target (~3500 times the unambiguous range of a 1617 pps repetition frequency). On the scope, echo may appear at any range which is equal to the residual between the true range to a point on the lunar hemisphere and some exact multiple of the unambiguous range of the set, in addition to which the orbital motion of the moon means that a given point on the lunar surface changes range rapidly. When at low elevation near the horizon the range-rate of the moon will be in the order of 1000 mph. In short the echo can have an arbitrary extension on the range axis, might abruptly change displayed range, but will maintain the same bearing from an aircraft in straight flight.

The Phase B echo does appear at essentially the same bearing throughout, and has intermittently a curious elongation or ghosting on the range axis. The Blue Book account describes this echo first appearing off the left wing and closing range abruptly from 3 miles to 1 mile (at 3000 mph according to Col. Werlich); although during the photo sequence the echo tends to approach slowly from 1.05 NM to 0.87 NM. Some qualified similarity to a moon echo can be argued, then, and a bearing of 9 o'clock from the aircraft would be ~30 degrees true, which is within 10 degrees or so of the 19-degree true azimuth of moon at 0400 local time on the 24 Oct 1968.

The long axis of the echo(es) is not in all cases exactly radial, deviating up to about 10 degrees in a clockwise direction. The degree of radial compactness and range consistency of the echo is probably also greater than one might expect. But these are minor issues compared to the fact that running an astronomy PC application for the latitude and longitude of Minot at 0400 CDT on 24 Oct 1968 places the moon at a negative elevation of -67 degrees.

This is measured from the surface of the earth, not from altitude, and moreover the 4/3 earth-radius radar horizon is always about 15% further away than the optical horizon, so we need to correct this figure. But even from 20,000 ft the radar horizon is only 200 miles away, and the relative elevation of the moon doesn't change much in 200 miles. This is equivalent to the change in celestial position due to less than 12 minutes siderial revolution of the earth, or less than 3 degrees. Even allowing for the possibility (unsupported by available radiosonde data) that the horizon might be extended by a strongly superrefractive elevated duct we could still be confident that the moon must have been some tens of degrees below the radar horizon of the ASB-9.

6-5. Echoes from Lightning Channels, Lightning Sferics & Ball Lightning

Echoes from lightning channels can be detected as discrete targets, or sferics due to RF radiation emitted by rapidly accelerated electrons in lightning channels can generate more widespread display products. The phenomenology throughout is completely inappropriate for sferics in this case. Successive lightning channels (duration about 0.5 sec) might show up as stochastic point echoes on successive scans around the scope as in Phase A if the radar is located in the middle of a storm. But there was no local thunderstorm activity.

The possibility exists of remote lightning strokes being detected by multiple-trip echoes, as plasma column cross-sections are typically in the region of 60m2 at centimetre wavelengths. But any echoes will be much weaker at X-band than at longer S-band or Lband wavelengths (generally a factor 10 increase in wavelength allows a 100-fold decrease in the necessary electron density for critical reflection) so the short 3 cm wavelength in this case is not very favourable; moreover the shorter the wavelength the more likely it is that lightning channels will be masked by echoes from the widespread precipitation that always accompanies thunderstorms. The Phase A photos show nothing resembling this sort of precipitation echo. So these factors combine to make echoes from remote lightning channels at multiple-trip ranges rather unlikely. (Needless to say the duration of the persistent Phase B target rules out lightning channel echoes.)

Blue Book concluded that the most likely explanation for the radar echoes was "a plasma of the ball lightning class". Quintanilla states: "Plasmas can effect electrical equipment and can also be painted on radar" and "Plasmas, such as ball lightning, can occur in clear weather as well as stormy weather." This rather misrepresents the fact that, statistically, ball lightning is very strongly correlated with electrical storms, even though there are some reports in the literature of phenomena nominally classified as 'ball lightning' occurring in clear weather. If such phenomena are physically identical to ball lightning or not is a moot question, inasmuch as explaining the sustained energy density of lightning balls has proved difficult enough for theoreticians even with the power of an active thunderstorm to draw on. In the present case it would be fair to say that the general condition of the weather is not remotely suggestive of ball lightning.

Ignoring the supposed hypersonic approach and transit of the B-52, the behaviour of the Phase B echo might redefine our understanding of ball lightning. Ball lightning duration is typically only a few seconds. A target with a radar cross-section comparable to a large jet (~ 10-1002m) pacing the aircraft at ~250 knots for at least half a minute and probably closer to 6 minutes (contemporaneous witness reports) is unintelligible as ball lightning. Speed, duration and cross-section are all at least one or two orders of magnitude greater than the median reported or inferrable values for ball lightning.

Blue Book also suggested a "possible plasma" as an explanation of the luminous object seen visually from the B-52 on or near the ground some minutes later. Multiple lightning balls are almost never reported. The probability of so rare a phenomenon occurring twice in the same area, in the absence of any sign of atmospheric electrical acitivity, is vanishingly small, and if the suggestion is that the same plasma was responsible for both radar and visual observations then this remarkable plasma is a "UFO" in all but name.

The malfunction of the two UHF radio transmitters coincident with the proximity of the radar target is not really addressed at all by Blue Book, although some vague suggestion of possible ball lightning-related electromagnetic effects is offered. One (not wholly successful) theory of ball lightning formation proposes that the plasma is sustained by ducted high-intensity radio-frequency fields (never observed), presumably associated with the large-scale vertical charge separation occurring in electrical storms. Some analogous mechanism might conceivably cause UHF interference. But there is no apparent likelihood of atmospheric-electrical RF emissions in this case. Moreover interference is one thing; complete transmission failure whilst preserving reception on the same wavelength is quite another.

It is true that a plasma will scatter radio waves. The UHF radio wavelengths concerned are around one metre (~300 MHz). Any plasma with an electron density high enough to efficiently scatter X-band radar (ex hypothesi) will be much more effective at scattering UHF radio. However the only radio waves scattered will be those that are actually radiated in the direction of the plasma. There is no obvious physical reason for a lightning ball that gives a discrete radar echo at 9 o'clock from the aircraft to affect radio waves transmitted forward to a receiver situated at about 12 o'clock ahead of the aircraft.

One can imagine an associated region of sparse ionisation, with a recombination rate not frequent enough to be detectable by visible light emission, which could, if spread over a large enough volume, still have significant opacity at radio wavelengths. If the B-52 were flying within or above such a region its UHF transmissions could be attenuated by absorption. But it is very doubtful that sudden and complete blocking of the transmission "in the middle of a word" could be caused in this way, and again the preservation of UHF reception on the same wavelength is completely unexplained. Moreover there is no evident natural mechanism for sustaining even a rather weakly ionised large volume of air in the absence of electrical storm activity (see Note 7).

6-6. Radio Frequency Interference, Internal Noise & EW Spoofing

Radio frequency interference was apparently not seriously considered by Blue Book. However a summary in the file of a phone conversation between Col. Werlich and an FTD officer contains the remark that according to Werlich "the blip changed shape, round, rectangular, etc". The conversation is dismissed as having contained "nothing of value", which may mean that oddly-mishapen blips are being considered diagnostic of false targets and that this information therefore adds "nothing of value" to what Blue Book had already concluded. A "rectangular" blip might reasonably have been thought to indicate some electronic artefact or interference, but there is nothing in the file that could be construed as an analysis of this possibility.

The Phase B echo recorded on the scope photos does vary in presentation, in a manner that can be construed as elongating or doubling in a radial direction and reverting on 782 to a more compact blip before disappearing. The meaning of "rectangular" is not very clear and may refer to a verbal description offered by the operator in respect of some earlier phase of the incident, or it may refer to the sort of appearance visible in frame # 776. Here the doubled or ghosted echo appears flattened off at the far end where it merges into the range ring, and at a glance it perhaps does resemble a somewhat rectangular "bar". But although these aspects of the presentation are certainly unusual, they are not necessarily diagnostic of an electronic phantom. There are mechanisms that could cause such effects to occur when the radar is detecting an ordinary reflective object (see Sections 6-11 & 6-12 below). In fact this sort of fairly discrete blip is far from the most likely appearance of an RFI display product. Spiral or spoke-like patterns all over the scope are typical.

When powerful radar pulses with foreign characteristics, or powerful continuous wave emissions that are not pulse modulated like radar beams, are picked up throughout the receiving antenna's rotation via sidelobe and spillover gain (or sometimes when a noise source washes directly through poorly shielded receiver circuitry) the display products are usually scrambled because the receiver input bears no relation to the specific modulation for which the display timebase is designed. But it is possible in a special case for interference to emulate a target arc if signals closely comparable to a radar's normal output can be picked up from a similar remote radar only via the antenna main gain - i.e., when the radars are "looking" at each other.

The conditions are: a) for the two radar wavelengths to be closely matched; b) for both scan rates to be closely matched; c) for both p.r.f 's to be very closely matched; ; and d ) for a short pulse train to be rather discretely sampled, which probably requires e) that the two antenna rotations are synchronised 180 degrees out of phase, so that they "look at each other" once per scan whilst sweeping in opposite directions and the simulated "dwell time" is short, and/or f) that there is a highly spatially anisotropic radio duct in the atmosphere that helps by sampling only the strong pulses at the peak of the gain. (There are also anti-ECM sidelobe suppression techniques commonly used in airborne radars that might enhance this selectivity; the ACR version of this radar did have monopulse sidelobe reduction or MSR, but the Tech Order suggests that it was only usable by selecting a distinct anti-jamming mode of the ACR, so we assume it did not affect the situation in Station Keep.)

If all of the conditions are satisfied the display product might resemble the discrete arc of pulses returned from a point target. If we consider two fixed ground radars, then if the two scan rates are perfectly synchronous the "echo" would appear in the same place on each scan, a stationary target. If the scan rates are very slightly asynchronous by an amount shorter than the trace time (which is the light-travel time for the maximum range on the display, about 0.5 millisecond for a 100 mile range) then the echo can progress radially in or out, along approximately the same set of trace radii, varying in intensity and presentation as the two antennas drift towards or away from perfect boresight alignment.

But if the asynchrony per rotation is larger than the trace time then the blip won't progress smoothly but will skip around, first on the range axis and then in azimuth as well. Obviously problems of interference that can be anticipated are normally designed out. Transmitters are tunable and identical sets are not normally sited in the radar line of sight from one another. Nevertheless RFI does occur, and it could happen that two very distant similar radars are able to "see" each other on rare occasions due to anomalous propagation conditions. There are large numbers of off-the-shelf civil and military ATC, airfield surveillance or GCA radar sets in use at airfields large and small across the US, as well as air navigation and weather radars. It is easy to imagine that a false near-stationary target could sometimes be caused by unusual mutual interference between two closely matched radars.

On the other hand this X-band airborne bombing-navigation radar differs from common L-band or S-band surveillance sets, and in the present case we know that at least one of the two radars hypothetically involved is an airborne bombing radar travelling at ~250 knots almost tangentially to an hypothetical line of sight whose bearing from the 1st radar definitely does not change at all within the limits of measurement for 24 seconds (scope photos 773-781), and probably does not change very substantially for around 5 minutes (contemporary witness reports). This could mean either a) that the source radar is so remote that the angular displacement is negligible even at the ground speed of the B-52, or b) that the source radar is also mobile.

The first option might just be supportable for the duration of the extant photos. For example: assume the true bearing is known within error bars of +/-1.0 degree, compounded of an uncertainty of +/-0.5 degree in the PPI bearing indication and a similar uncertainty in the aircraft heading/yaw indication (this may be optimistic). Then 24 secs. flight at 250 mph gives a travel of about 1.66 miles, which would subtend an angle narrower than 1.0 degree from a remote radar at any range greater than ~100 miles.

But a remote fixed ground radar would probably be in conflict with witness testimony, which indicates a duration of target-stationing off the left wing approaching 6 minutes. Distance to the radar would need to be more than about 1400 miles to keep the angular displacement within 1.0 degree in this case, which is around 5 times the maximum unambiguous range of the ASB-9 and in the order of 10 times the radar horizon distance from the B-52 at the time of the photography, implying a relatively powerful radar and trapping or ducting conditions for which there is no evidence (see Section 6-11).

It is true that we lack photo evidence that the echo bearing remained constant to the same accuracy during the entire 6 minutes. If it moved 10 degrees in this time then the emitting radar could have been as close as the second-trip distance of 140 miles. And many types of radars, such as marine radars, some weather radars, fire-control radars or army mobile tactical radars, share the X-band frequency range of the ASB-9. But arguably by far the most likely candidate for an emitter that meets all the conditions of precisely similar frequency, pulse repetition frequency and scan rate, and which also enables the echo to remain at a constant bearing from a moving receiver over an arbitrary period, is another airborne ASB-9 bombing-navigation radar, presumably in another B-52 flying a parallel course many tens of miles away to the NE.

Conceivably, a high level radio duct above the levels sampled by radiosonde could cause some energy to arrive via slightly longer refracted ray paths as a fractionally delayed ghost of the main signal, recieved by standard 4/3-earth radius ray paths, and it is possible that this could explain the elongation of the blip along the range axis of the PPI, with a fainter secondary blip appearing intermittently.

There is also a strong direct correlation between the rate of closure of the Phase B blip on the display and the rate of descent of the B-52 (see Section 7). Now, if there is relative movement between the transmitting and receiving radars, and/or a fluctuating ray path due to anomalous refraction, then the length of the ray path may change and the receiver will "see" this as a changing echo delay. Conceivably, therefore, a change in displayed range could be systematically related to the aircraft altitude.

But ex hypothesi the interference display product on the PPI has an extremely sensitive dependency on the degree of asynchrony in the two radars' intrinsic electromechanical periodicities. The asynchrony has to be tiny in order to produce a display product resembling a discrete echo in the first place; but in order for the change in its displayed range to be overwhelmingly dominated by a changing length of ray path systematically related to the altitude, any underlying blip displacement due to drift in the two antenna rotation rates (in particular) would need to be vanishingly small, approaching microsecond synchrony.

This seems highly unlikely, even assuming identical radar installations. Indeed, especially assuming this: The ASB-9 antenna is driven by a hydraulic motor whose nominal 20 RPM rate is specified in the Tech Order as 17.5 - 22.5 RPM, margins of +/- 12.5%, leading to a possible 25% (5RPM) disrepancy in the rotation rates of two otherwise identical ASB-9 radars, or as much as 0.75 sec per 3-second scan period. In fact the radarscope clock in the photos gives evidence that this antenna rotation rate is slightly adrift from its nominal 20 RPM setting and has an error of possibly about 2% (see Section 2 and Note 3) which, if correct, would limit the total possible discrepancy to somewhere in the range 10.5% - 14.5% depending on whether the two drifts add or subtract, or about 0.3 - 0.4 sec per rotation. Since even a discrepancy ten times as small as this would still be three orders of magnitude too large, we should probably conclude that a systematic reference to the local flight level is more naturally explainable if the displayed range represents a genuine echo delay from some nearby reflector having an approximately constant relation to the ground level (such a model is developed in Section 7).

But several facts can be brought forward to suggest that there was something unusual about the electromagnetic environment. The B-52 navigator, Pat McCaslin, recalls that the plane's Electronic Warfare officer received unusual responses on his equipment during the time that the unidentified echo was being detected (although this does not appear in the Blue Book documents and is not recalled by Goduto). UHF radio transmission from the aircraft is also known to have been affected during the same period. The possibility arises that these electromagnetic anomalies are symptoms of a deliberate ECM jamming exercise carried out against the B-52 by other elements of the USAF. But if the intention is to simulate a convincing aircraft target then this deception jamming (from an unidentified source; see also Section 6-2) was not a very effective spoof. Also, jamming does not simply silence radio transmissions, as was described in this case, but fills the frequency band with noise; and how would it block UHF transmissions selectively but not block UHF reception on the same (multiple) radio sets?

Finally there is the possibility of internal radar system noise due to component degradation or something similar. There is no evidence in the file that any internal radar fault was discovered, or even suspected, either by the operators or by investigators in the ensuing days and weeks. There is no specific record of an electronics check (part of the reason for this may have been the undismissable ground- and air-visual reports as well as the report of a target on the electronically independent Minot weather radar) but presumably neither routine operations nor maintenance uncovered any persistent fault.

The Blue Book investigation does nevertheless seem to have considered the idea that an electronic fault implicated in the UHF failure might somehow have caused the radar blip (no mention of the EW equipment responses occurs in the file), but this was discarded quite quickly. Col. Werlich satisified himself that the UHF hiatus was not due to an equipment fault - the transmission was suddenly interrupted "in the middle of a word", affecting both independent UHF sets, and both sets worked perfectly afterwards. If the radar blip was an internal noise track, or remote interference, then in either case there is a pure coincidence with the UHF interruption. Blue Book's reasoning seems to have been that it was better to seek a common external cause of the radar blip and the UHF interruption, hence the reliance on "ball lightning". The logic of this position is probably sound even if the explanation isn't.

In summary, interference from another B-52 bombing-navigation radar may be physically possible, but the operators had seen nothing like this before, neither had the radarscope photoanalysts, so presumably it must be very rare. Hypothetical propagation conditions might help to account for this, but still the combination of special circumstances required is undeniably very fortuitous. Simultaneous UHF failure and air/ground visual observations add further levels of coincidence. The scenario is at least very improbable.

6-7. Balloon

The Blue Book file mentions that Lt. Marano raised the possibility of hot-air balloons as an explanation of the sightings. There is also some ambiguous reference to "trouble we have had with hot air balloons" although the context of this remark is very unclear. This notion was dismissed by Col.Werlich on the grounds of local geography and the fact that there were only handful of remote farmhouses in hundreds of square miles. Whether Marano was offering this idea to explain the radar echoes as well as the air and ground-visual sightings is not clear, but it should be considered. The copilot's description of the grounded object as an orange-glowing oval, with a "molten", "translucent" look to it and a greenish appendage on one end, could (with some effort, it has to be said) be squared with a very large hot-air balloon.

Obvious objections are the implied radar cross-section of the "balloon" and its velocity. In the case of a hot air balloon it is interesting to speculate that the flame and/or associated turbulent hot air column might themselves contribute to the radar signature, and one might imagine the intermittent "doubling" of the radar target as indicating a constant echo from the bulk of a balloon somewhat below the B-52, supplemented with an occasional secondary echo at slightly greater slant range when the flame generator below it is switched on. But the height of the rig implied by the displayed range differential between "balloon" and "gondola" (order of 1000 ft) would be unrealistic. And in any case neoprene or polyethylene balloon envelopes are generally not radar reflective, implying that the constant echo would have to come from conductive components in the payload, with the intermittent echo at greater displayed range being caused by the flame and hot air column rising into the envelope above it. But this would have to mean that the rig was considerably above the B-52's probable altitude of about 9000 ft and implies that it disappeared in frame 782 by climbing out of the radiation pattern, not by descending to the ground, which is contrary to the reported "landing" scenario that presumably gave rise to Lt. Marano's hot-air balloon speculation in the first place. And a radar cross-section comparable to a large jet is wholly inconsistent with any believable balloon payload.

As for relative velocity: The range rate and the azimuth rate of the Phase B echo relative to the B-52 are very small. Winds from 320 degrees would be directly behind the B-52, but obviously the vector sum of the highest likely rate of balloon ascent (say about 15 mph) and a 50-knot wind (the strongest winds at any altitude of the aircraft during the incident, @ 20,000 ft) cannot remotely match the likely aircraft ground speed. If a near co-altitudinal balloon falls behind the B-52 at a plausible rate of 140 knots then during the 24 second Phase B photograph sequence the bearing to the balloon should drop back by some 30 degrees. The bearing of the radar echo changes only 1 degree between frames 773 and 782. The relative angular rate alone seems sufficient to rule out a balloon as a cause of the radar episode.

6-8. Auroral ionisation

Blue Book makes only passing mention of auroral phenomena. Discussing lightning plasmas that might cause electrical effects and be detected on radar, Quintanilla adds the remark that "Aurora Borealis is quite often seen from Minot AFB at this time of the year and is an electrical atmospheric phenomenon", apparently implying in a vague way that auroral phenomena might be stirred into the explanatory mix.

Auroral ionisation can reflect radio waves and generally does so in an echo pattern that correlates quite closely with the visual pattern of the auroral glow, i.e. in broad swathes and streaks spanning many degrees of arc. It might in some cases cause discrete small echoes on some radar scopes but this seems most unlikely in the present case. Neither of the echoes on frames 771 and 772 is likely to be due to aurora since a) neither echo is in the auroral quadrant and b) detectable auroral echoes at 3cm are very unlikely anyway because the frequency dependency of auroral echoes is similar to that of other ionisation phenomena such as meteor trails and lightning channels.

The true ranges to aurorae will be comparable to the ranges of most meteors, in the order of hundreds of miles at low elevations (therefore multiple trip echoes), and the electron densities will be much lower than in meteor trails. Power reflectivity from ionisation falls off rapidly through L-band and S-band, and an X-band radar such as ASB-9 has little chance. Metric wavelengths in the order of 100 times the 3cm length of the ASB-9 are favoured.

The condition for detection is further crucially restricted by the need for the radar line of sight to be near perpendicular to the magnetic field lines, which generally limits echoes to a well-defined region in the N scope quadrant, regardless of where in the sky visual auroras may be observable. The radar's angle of elevation in this case is not very high, being under the nose of the B-52, and from a simple geometric point of view the antenna would be extremely poorly placed for detecting zenithal auroral streamers even if the streamer orientation and radar wavelength were favourable.

In short, the Phase A radar echoes in the southerly scope sectors (771 & 772) are almost certainly not auroral echoes on the grounds of magnetic field geometry alone, and the Phase B echo is very probably not an auroral echo on the grounds of its discreteness, strength, stability in the same approximate location, and persistence at a very unfavourable radar wavelength.

6-9. Birds, Insects

The persistence of the Phase B echo at a 40 degree bearing from a B-52 flying on a straight heading at ~250 knots seems sufficient to rule out birds or insects. Could birds account for the Phase A echoes to the right of the plane on frames 771 and 772? Obviously a single bird is ruled out. It is also difficult to conceive of two different birds each rapidly flying in and out of the radar cover for a single scan, especially given evidence in each echo of internal structure indicating either a very high target rate during the brief dwell time of the beam or an elongated target echoing area with a major axis in the order of hundreds of feet (see Section 7).

It might be worth pointing out that larger birds at ranges of just a couple of miles could be bright targets. In fact the inverse 4th power attenuation of echo intensity means that on a normal PPI showing airborne targets out to the limit of the display a nearby bird can be a stronger target than a distant aircraft on the same scan, possibly deceiving an inexperienced operator. But this is not a normal surveillance PPI used (generally) to search for targets out to long ranges. The operator is only looking at airborne targets inside a small altitude "hole" whose maximum radius is never more than about 30,000 ft, and in terms of range he is broadly speaking always comparing like with like.

For example, the ratio of returned power between identical targets at 4 miles and 1 mile range inside the 5-mile Station Keep PPI display is only 1 : 256, which is very tiny compared to the ratio of signal levels handled by a typical-based surveillance radar and comparable to the variation in return from a single aircraft due to changing aspect. By contrast, two identical targets at 50 miles and 1 mile range inside a 60-mile airfield surveillance PPI would have an enormous signal ratio of 1 : 6.25 x 106. In the present case an experienced radar-navigator accustomed to the use of the ASB-9 radar for close range Station Keeping on refuelling tankers offered the view that the Phase B echoes indicated a target cross section larger than a KC-135 at comparable range. This is confirmed by the opinion of the Bomb Wing intelligence analyst based on the radarscope photos.

It is possible for a cloud of insects to produce a significant target, indeed the more so at X-band than at more typical surveillance wavelengths. But the issues raised with regard to bird echoes become even more grossly problematic for insects. Therefore neither of these common sources of radar "angels" seems to be relevant on the grounds of implied airspeed and echo intensity and presentation.

6-10. Satellites

First-trip echoes from satellites at high elevations within the unambiguous range of the radar (67.5 miles in Station Keep mode) are almost certainly ruled out by a top edge main beam coverage of only +8 degrees. In any case the speeds and trajectories would be very inappropriate. However the radar could in principle pick up multiple-trip echoes from a distant satellite at low angles of elevation. In this case displayed tangential speeds would be slowed and the track of a satellite in a polar orbit travelling N-S roughly perpendicular to the radar line of sight could be distorted into a curve or a "V" approaching and receding from the scope centre. But there seems to be no sensible application of the theory in this case.

Large satellites at this date could have cross-sections of hundreds of square meters, as large as or larger than a well-aspected big jet. But the inverse 4th-power attenuation of returned energy makes the effect of distance dramatically nonlinear, and it seems inconceivable that any satellite at likely third-trip ranges could present as an echo which was characterised by experienced operators and photoanalysts as stronger than that from a nearby B-52 or "several times the size of KC-135 tanker".

Also, although the apparent groundspeed indicated on the PPI would be several times slower than the typical ~18,000 mph orbital speed the average angular rate would be preserved. This rate will typically be in the order of 100 degs/minute. The angular rate of the persistent Phase B echo is near-zero for far too long. The recorded angular rate over almost half a minute is less than about 2 degs/minute. This is already inconsistent with a satellite echo and testimony indicates that a comparably low angular rate was maintained for several minutes before the camera was switched on.

Finally the displayed range rate of a multiple-trip satellite echo would still be in the order of 1000's of mph, but photo and witness evidence both indicate a negligible range rate maintained for at least tens of seconds and probably for several minutes. In short, satellite ech

oes appear to be ruled out.

6-11. Anomalous propagation

Here 'anomalous propagation' or AP is broadly defined to include various atmospheric refraction and reflection effects that could give rise to phantom targets. Several passages in the Blue Book file materials refer to a "moderate temperature inversion" or a "pretty good inversion". The AFR 80-17 telex report has the entry:

Basic Reporting Data (TELEX 0008-4)

Blue Book typically appealed to any evidence of temperature inversion to write off cases as probable AP, often without much regard to either phenomenology or quantities. And early in the investigation Col. Quintanilla remarked, covering a couple of bases at the same time:

I'm (Col Quintanilla) pretty sure it was either caused by an internal radar malfunction . . . or because of the inversion he might have also picked up an anomalous blip.
Memo for the Record (00010-1)

But unusually, in this case the official evaluation did not in the end place great emphasis on radar AP or electronic phantoms and Blue Book came down in favour of a ball lightning-type plasma.

Nevertheless the conditions need to be investigated, and it should be said first of all that the above reliance on temperature lapse rates alone is not at all meaningful since humidity is a much more important contributor to radar refractivity.

Secondly the upper-air weather data (though not the surface data) given in the Blue Book file were "obtained from Glasgow", an airfield in Montana some 250 miles west of the location of the incident, and so although possibly indicative are not guaranteed to be relevant.

Thirdly the time of Glasgow balloon release is not stated.

Fourthly the original Glasgow rawinsonde data are not presented; instead those given in the file occur only in a Memo for the Record, a typed record (with not-wholly-unambiguous handwritten addenda) of a telephone conversation in which Col. Werlich passed on to Lt. Marano, FTD, information "obtained" from Glasgow in an unspecified manner by Sgt. Dickson of the Minot AFB weather office.

And fifthly, Quintanilla gives no thought to the physics or the ray geometry of these "anomalous blips" that might occur due to inversion conditions when the radar is flying some thousands of feet above the supposed inversion. Commonly energy from a radar on the ground which normally only "looks" at the sky may be bent or scattered back down to detect echoes from reflectors elsewhere on the ground. In this case we have a radar in the sky deliberately inverted to radiate most of its energy earthward in such a way as to be full of ground return in normal operation, so the usual assumptions may be misleading.

6-11-1 Radar Refractivity Profiles

To begin with, correct radar refractive index values were calculated from the Glasgow AFB temperature and dewpoint data in the file and the resulting N-gradient is graphed in Fig. 13 below.

graphic from shough analysis

Fig. 13. N-profile for Glasgow AFB, Montana, Oct 24 1968. Constructed from Blue Book temperature and dewpoint data for five levels (time unknown), altitudes converted to equivalent pressures (36 mbar/kft) and N-values determined by nomogram. The dotted part of the profile connects the Minot AFB surface data for 0855 GMT. The limiting slopes of "normal" refractivity (mean -12N/kft) are indicated at left.

The radiosonde sampling levels are too sparse to give a very meaningful picture, but the main features of the diagram are:

  • the average gradient for the first 2000 ft is just marginally superrefractive, but not significantly at -27.5 N-units per kft (the range 0 N/kft to -24 N/kft is considered the extent of "standard" refractivity);
  • the conditions for trapping (about -48 N-units per thousand feet or greater) are nowhere indicated;
  • above 2000 ft (height above terrain, so about 3680 ft MSL) the refractivity gradient remains quite close to the mean for a standard atmosphere.

[Note that the equivalent pressures calculated here assume a "standard atmosphere" of 36 mbar/kft (Wylie, 1952) chosen to be close to the empirical lapse rate found for the Bismark ND soundings (see Tables 5/6 below).]

graphic from shough analysis

Table 4. Radar refractive index data for Glasgow/Minot. Calculated from Air Force temperature and dewpoint readings for Glasgow AFB, Montana (aloft, time unknown) and Minot AFB, ND (surface, 0855 GMT). Pressures are estimated and indicative only.

So there is no sign in the Glasgow radiosonde data of the elevated anomalous propagation conditions inferred by Blue Book from the temperature figures. Very marginal superrefactivity is indicated through the first 2000 ft above the surface in Fig. 13, but this depends on the validity of importing Glasgow balloon data and Minot surface data into the same diagram. Stratification of stable night-time air can extend over very large horizontal distances, but this assumption is obviously doubtful. Given the limitations of the data and the relative remoteness of Glasgow from Minot some coherent data from a nearer weather station were considered desirable.

Enquiries to the US National Climate Data Centre, Asheville, NC., established that the nearest extant balloon release data for Oct 24 1968 were from Bismark, ND., approximately 120 miles SSE of Minot AFB. Copies of the Bismark data for 0000 hrs and 1200 hrs on the 24th were obtained and used to populate Table 5 and Table 6 below (the complete NCDC dataset is reproduced in Note 10).

graphic from shough analysis

Table 5. Radar refractive index data, Bismark ND, 0000 Oct 24 1968. Temperature, pressure, RH and heights from DS-6201, US Rawinsonde Observations, courtesy of National Climatic Data Centre, Asheville, NC.

The Bismark datasets have temperatures recorded up to heights of 11 and 50 mbar, but no relative humidities are shown above 350 and 308 mbar so these upper levels are unfortunately of no use. In any case the N-profiles graphed in Fig.14 below are terminated at 500 mbar as this was the limit of the refractive index nomogram used. (Informally, it is fair to say that the few levels not graphed indicate continuing trends, with one small inversion - only a fraction of a degree, just off the top of the 0000 hrs diagram at about 450 mbar - and no noteable discontinuities in the dew point.)

graphic from shough analysis

Table 6. Radar refractive index data, Bismark ND, 1200 Oct 24 1968. Temperature, pressure, RH and heights from DS-6201, US Rawinsonde Observations, courtesy of National Climatic Data Centre, Asheville, NC.

The results of these Bismark observations bracket the sighting period of interest. Overall, of 32 pairs of layers compared, only in four cases are N-gradients outside the range for a "standard" atmosphere indicated. These occur in the first few hundred feet above the surface in both diagrams, and at about 4,000 and 10,000 ft in the second.

The earlier 0000 GMT diagram shows an interesting narrow subrefactive surface layer; whilst the most relevant 1200 GMT diagram (~2 hrs 54 mins after the time photographed on the radarscope clock) shows a very marginally superrefractive surface layer of similar depth, in fact not very dissimilar to the surface value shown in the compound Minot/Glasgow profile in Table 4, though over a narrower sample range in this case.

graphic from shough analysis

Fig. 14. N-profiles for Bismark, ND, Oct 24 1968. Heights converted from geopotential meters.

The elevated 1200 GMT discontinuities are both subrefractive (i.e., tending to bend raypaths upwards away from the surface of the earth) and neither is more than a few Nunits outside the "normal" range. None of these features appears likely to cause an increased likelihood of anomalous ground echoes on an airborne radar (in general, rather the opposite if anything), and there is no evidence of any RI discontinuities severe enough to even be detectable by direct backscatter, let alone as a very strong discrete echo.

Of course it is impossible to rule out the presence of undetected sharp layers of extreme N-gradient falling between the sample points. Such extreme layers, occurring below the flight level, could conceivably produce very unusual echoes by direct backscatter near normal incidence. Gradients in the order of tens of N-units/cm have been hypothesised in extreme conditions (for perspective, in terms of equivalent temperature - ~1 deg C per N unit - this would be 100,000 times as steep as the steepest gradients responsible for normal optical mirage). Admittedly it strains credibility to suppose that such a backscatter echo, at an angle very far from the peak gain of the antenna (in the order of several 10's of dB below the gain anywhere in the main beam), could yield a strong blip on the PPI of an airborne surveillance radar of modest power (see Note 6). But it is also true that the extreme limits of the power reflectivity coefficients of such layers in nature may be unknown, so the hypothesis should be investigated.

As will now be shown, even granting extraordinary power reflection efficiency it is very hard to see such a mechanism as a primary cause of a strong discrete echo of the kind seen. The main reasons for this conclusion are connected with the strength and discreteness of the Phase B echo (requiring a hot-spot of direct backscatter to the antenna at normal incidence) combined inconsistently with persistence of the echo off-centre at a constant bearing (indicating forward scatter away from the antenna at an off-normal incidence).

6-11-2. Backscatter from a Hypothetical Elevated Layer of Extreme N-Gradient

As mentioned, the detection of local ground echoes by AP in the usual way is not an issue in this case. If the short-range unidentified targets within the altitude hole are local first-trip echoes (they do not have to be, but we come back to that presently), and as long as their displayed slant ranges are less than the aircraft altitude, then they must be echoes from airborne reflectors. For the B-52 altitude values found in the present analysis the minimum geometrically possible altitude of a such an airborne reflector at frame #773 is approximately 10,000 because they assume a reflector to be vertically below the aircraft, whereas the radar is emitting very little radiation in this direction because the antenna boresight is elevated in Station Keep mode.

The technical literature (Section 4) and the evidence of the photographs both suggest a rather sharp cut-off in antenna gain at a depression angle of about 45-50 degrees, consistent with the B-52 flight data (Section 4). If the target is within this main beam pattern then it must be significantly higher than ~3500 ft above the terrain. But of course gain will never be quite zero at any angle for any radiator, and we should perhaps consider that this apparent cut-off is not sharp at all ranges.

Because of practical limits of antenna design and mounting, the curve of gain might have minor lobes at undesirable angles far from the boresight that are insignificant in normal use but might pick up an echo from an unusual reflector at close range almost directly below the aircraft. If so the echo will paint on the PPI at the azimuth of the main beam at the time. Nevertheless even in a significant minor lobe the gain will typically be several orders of magnitude weaker than the main beam, and so a strong echo presentation - described by expert witnesses as comparable to or stronger than the echo from a very large jet in the main beam - implies that any reflector near -90 degrees elevation would have to be super-efficient in comparison to a large jet by at least the same several orders of magnitude.

What could this local reflector be, if not a large airborne object? In general one would expect the most efficient direct backscattensity that was (hypothetically) constant per unit solid angle the curve of reflected intensity would peak at the nadir vertically beneath the aircraft and diminish towards grazing incidence at longer ground ranges (Fig.15). If the plane of rotation of the antenna is horizontal above a plane reflecting surface, then in this ideal case the echo would have a "hot spot" perfectly centred on the PPI and diminishing in intensity with radial distance.

In the real case the antenna gain falls rapidly towards the nadir (as far as ground echo is concerned, approaching zero for practical purposes at around -50 degrees) and we know that the reflector must be >3500 ft above the terrain during the photo sequence. The approximate constancy of this altitude figure has been alluded to, and could be taken to suggest that the closure of slant range on the PPI is largely or even entirely due to the rate of descent of the B-52 in relation to a reflector which is roughly stationary in altitude at about 3500 ft or above. Is it possible that the reducing slant range to the UFO echo may be tracking the reducing vertical distance to a sharp layer of extreme N-gradient? There is no radiosonde evidence of such narrow layers, but they could fall between the sparse data points and such a layer could conceivably constitute a radio "mirror" causing direct backscatter to the radar receiver in the form of a hot spot of efficient reflectivity.

However another feature of the real case is that the successive echoes are not coincident with the scope centre. They are not even scattered isotropically about the scope centre as might be expected in the case of some random wander about the mean. In fact they are all tightly confined to a narrow azimuth, which is difficult to explain.

graphic from shough analysis

Fig. 15. Geometry of hypothetical direct backscatter echo due to gain near normal incidence. Although gain near the nadir may be many orders of magnitude smaller than at the peak of the main lobe, direct backscatter efficiency increases rapidly towards normal incidence. In level flight with all other factors equal, the possible echo region would tend to be annular and concentric with the PPI centre.

The problem is trying to understand how to combine an extreme efficiency, which is more difficult to support the further the reflection geometry moves away from normal incidence, and which therefore strongly predicts echoes positioned isotropically in relation to the PPI centre, with the very pronounced anisotropy that we actually see. A concentric hot-spot is possible in principle; but the last thing one ought to expect is a compact echo, eccentric and restricted to one narrow azimuth on scan after scan within a margin of a degree or so, over a minimum of about half a minute and probably for as long as 6 minutes. Ex hypothesi this is hard to explain other than as a systemmatic deviation of the reflection geometry away from normal incidence. How can this occur?

One way in which this might happen is if the aircraft is flying with a slight angle of roll which could favour a consistently anisotropic backscatter, as shown exaggerated for clarity in Fig.16. But this appears to be ruled out because a) the aircraft is proven to be on a straight heading and the wings will be level, and more importantly because b) the antenna tilt is automatically servo-stabilised by pitch and roll signals from the navigation computer and aircraft attitude (to +/- 15 degrees) is irrelevant.

Could the computer compensation have been in error, sending inaccurate signals to the antenna tilt servos and causing an off-kilter rotation which favoured normal-incidence low gain echo returns from beneath the aircraft only when the boresight azimuth was on the left of the aircraft? Almost certainly not, itude hole proportional to the cosine of this angle. For example, given a representative flight altitude of 10,000 ft, a tilt of only 5 degrees would cause the altitude hole radius to expand approximately 770 ft on one side of the PPI and contract by the same amount on the opposite side. This corresponds to fully 25% of the range ring interval, so even a small fraction of this 5 degree tilt should be measurable. There is no detectable asymmetry in the altitude hole - relative to the small overall eccentricity of the display caused by off-axis photography. (See also Section 6-12 for a related issue.)

graphic from shough analysis

Fig. 16. Eccentric display of direct backscatter echo due to failure of antenna servo-stabilisation. (Hypothetical pattern and exaggerated roll angle for illustration only).

So could there be an inhomogeneity in the layer, a domain or bubble of exceptionally high power reflectivity causing a persistent echo? The answer is again "no", because such a nearby feature of this hypothetical layer could not possibly maintain constant bearing from a B-52 travelling at 250 mph for at least 24 seconds (photo) and probably about 6 minutes.

The only options that appear to be left are to assume either a) a linear RI discontinuity parallelling the flight track of the B-52 at a slant range of about a mile, or b) an inclined layer tipped up to the ENE and down to the WSW, i.e. rotated around a horizontal axis parallel with the B-52 flight track (Fig.17), which might behave as a canted plane reflector so that a "hotspot" of extremely efficient reflectivity occurring at normal incidence could appear constantly offset to a bearing of 9 o'clock within a margin of about 1 degree over a recorded distance of about 2.5 miles and a well-reported distance some ten times as long.

Option a) is meteorologically bizarre; and as for option b), given that a layer with the necessary extreme efficiency of backscatter in a hypothetical minor radar lobe is already a reach of speculation, the added coincidence of a systematic reference of the layer inclination to the B-52 flight track, plus the unlikely compactness of the echo presentation, render the theory hardly credible in this writer's opinion.

graphic from shough analysis

Fig. 17. Offset backscatter hot-spot due to a canted layer. (Purely illustrative and not intended to represent real radar coverage or realistic angles)

6-11-3. Multiple-Trip Echoes from a Remote Reflector

If the echo is not a first-trip echo from a target inside the altitude hole (and assuming it is not a phantom due to malfunction or RFI) then it would have to be a target returning echoes from beyond the unambiguous range of the set, which would be 67.5 miles for the pulse repetition frequency of 1617 pps used in Station Keep mode. A reflector at range r would appear on the scope as a second-trip echo at a displayed range of r - 67.5 miles (i.e if r = 70 miles then displayed range = 2.5 miles) and could possibly account for the target falling back from about 40 to 39 degrees (+/- 1.0 degree) on the scope photos. Flight at ~250 mph for 24 secs. gives a travel of about 1.66 miles, which would subtend an angle of ~1.4 degree from a remote reflector at ~70 miles. Given margins of error this is not too bad and might explain the apparent stationing of the echo.

One other factor that might be suggestive of a remote fixed reflector is the behaviour of the echoes when first detected. Very roughly speaking, an echo was detected on the right of the aircraft on the o aircraft. Whilst this behaviour has quite a complex relation to the changing heading of the aircraft if considered as a moving object in the local sky, it has a natural relation to it considered as a multiple-trip echo of a static remote reflector. Basically (in the absence of detailed track data here) we could describe the relative motion by saying that the echo stayed in the NE.

The radar cross-section of the distant target implied by the multiple-trip theory is considerable. If the displayed echo was comparable in width of presentation to a large jet at 1.5 miles (the overall echo was described as larger than a KC-135 or a B-52) then the remote reflector responsible should be treated effectively as a point target and the returned power varies as the inverse fourth power, leading to a truly enormous ratio of efficiencies in the order 10^6. This implies an equivalent echoing area possibly as large as 10^9 square metres. One possibility might be echoes received from an isolated patch of high terrain.

Turtle Mountain, an upland area rising to 767 meters (2515 ft), on the Manitoba border, is at about the correct ~70 mile range, well inside the radar horizon (from an altitude of 20,000' MSL a radar beam intersects zero feet MSL at about 200 miles distance even in normal refractivity), and is near enough the azimuth indicated to be worth looking at. An echo around 7 degrees wide (frame 773) would correspond to a 2nd-trip echo from an area almost 9 miles across at the range of Turtle Mountain, and could conceivably be a reflection from the escarpment on the SW facing edge of the massif, which is the highest side. (Note that this would be especially possible if a Short Time Constant anti-clutter circuit is available to the operator (seeNote 5). However although STC was fitted to some bomb-nav radars of this vintage, there is no direct evidence that an STC switch was fitted in this case. It is also true that the rather large unbroken areas of generally featureless ground echo on all the photos do not suggest the use of such a filter.)

On this theory the doubling and/or elongating of the radar echo might be explained if a portion of the radar energy could be deflected by an elevated layer above the flight level, possibly a tropopausal layer above the usable radiosonde readings, returning a delayed echo from the mountain to the receiver via this slightly longer ray path. The ghost would appear on the same scope azimuth at a slightly greater range and would probably be intermittent as the efficiency of the raypath fluctuated. See Fig.18 below.

graphic from shough analysis

Fig. 18. Schematic diagram of possible dual ray paths, of unequal length, giving rise to primary echo and simultaneous ghost echo of a distant mountain. Model a) is unlikely in view of the radar refractivity profile, but model b) is not ruled out.

It remains unproven that radiosonde readings taken 250 miles W and 120 miles SSE of the sighting location are relevant to conditions tens of miles northeast of the sighting location, of course, and there is no direct evidence at all of the required AP conditions. However with some reservations one could say that some variant of this theory is qualitatively consistent with the photo evidence.

On the other hand, over a period of some 6 minutes, as reported in the RAPCON transcript and elsewhere, this echo would have moved over an azimuth of 20 degrees or more. Whether this could still be consistent with an echo that kept station "off the left wing" as described is debateable. Also the nearer edge of the primary blip is seen to approach the radar over the photo sequence. The expected change in displayed range and the duration of the echo would depend sensitively on the exact relative azimuth, and on possible fluctuations in the radar path length(s) due to changing d to increase slightly from frame 773, not to decrease as shown, and certainly not at a rate equal to the B-52's descent rate over the local terrain.

It is also difficult to make this theory work in the face of evidence that the B-52 was still NW of Minot AFB prior to executing the planned low approach when the photos were taken, because only from positions SE of Minot AFB would Turtle Mountain begin to approach the displayed 40 degree azimuth. See Fig.19 below. If the photos were taken at the very end of the radar event at the position indicated in the official file the discrepancy is about 30 degrees; at any earlier point on the flight track the match gets progressively worse.

graphic from shough analysis

Fig. 19. Relative bearings of Phase B echo and Turtle Mountains upland region. (illustration only approximately to scale)

Could the B-52 possibly have been SE of Minot AFB at the time? It is true that there is an unexplained apparent discrepancy between the times recorded in the RAPCON transcript and those photographed from the radarscope clock (see Note 3) and it would be possible to appeal to this timing ambiguity in order to place the B-52 SE of Minot at 0406. But to support the theory that the aircraft was already climbing out from its low approach over the Minot runway at this time we have to explain not only witness evidence (supported by the contemporanous RAPCON tape traircraft was still on approach some miles to the NW of the unway, and internal photo evidence which indicates an aircraft altitude consistent with Col.Werlich's 1968 reconstruction placing the aircraft at least 16 miles NW of the runway at this time (or even more in Claude Poher's reconstruction which ties the ground feature in 783 to the shore of Lake Darling), but also the rather conclusive evidence (see Section 5-2) that the B-52 is descending towards the runway during the photo sequence.

6-12. Ghost Echoes

Strong ghost echoes produced by multiple reflections ordinarily require first-trip returns from a primary reflector and a secondary reflector quite near the radar, such as another aircraft and an efficient corner-reflector on the ground. The displayed range to a ghost echo on the PPI will be half the total additive out-and-back path length of the signal via all reflectors. The ghost cannot possibly appear closer than the slant range to the secondary reflector, and generally the reflection geometry means that it will be much greater. Such ghosts typically do not last long as the critical geometry is unlikely to be sustainable due to relative motions of the radar and reflectors (Blackmer et al., 1969).

In this case the only evidence of an accompanying "aircraft" is the evidence for a UFO that we are trying to explain away, and the range from the B-52 to the ground at all relevant stages of the flight is far too great for echoes at ranges of a mile or so to be caused by secondary ground reflectors.

An exotic kind of ghost reflection geometry might conceivably arise if we can hypothesise an extremely sharp elevated scattering layer below the flight level, with an extraordinary power reflection coefficient near normal incidence, as we tried in Section 6-11. In this case a part of the B-52's own airframe might act as primary reflector, and the layer as secondary, with a ghost being displayed at essentially the same range as the path length to the layer. The ghost range could therefore be as small as the ranges photographed, and it is also possible that in this way we could explain a very discrete and anisotropic echo which we found impossible to do by invoking a scattering layer alone, since the ghost will appear at the bearing of the primary reflector - in this case a part of the B-52, say a section of the wing or an engine pod.

graphic from shough analysis

Fig. 20. Roll of the aircraft in a turn, bringing wing down into higher-gain region of servo-stabilised antenna pattern.

Fig. 20 suggests how this might occur due to the fact that the plane of rotation of the antenna is servo-stabilised by pitch and roll signals from the flight computer. In other words during manoeuvres the radar stays still, like the eye of a hawk, whilst the rest of the plane oscillates around it (within tilt limits of +/- 15 degrees).

Interestingly if we look at the positions of first detection of the radar UFO shown in Fig. 11 in Section 5-3 we see that it appeared off the right wing about the time when the B-52 would have been beginning to bank into a right hand turn (right wing dropping), and remained on the inside of the turn until about the point where the B-52 would have been banking into a left turn to compensate its overshoot and come back onto the approach heading over the WT beacon fix. At this point, with the left wing dropping, the echo reappeared on the left of the aircraft. As one wing drops it moves into a higher-gain region of the radiation pattern, possibly scattering increased energy down near the nadir, simultaneously as the opposite wing is rising out of the radiation pattern.

This is an intriguing hypothesis but, even given the possible existence of a layer with such extraordinary backscatter efficiency, it fails in several ways.

First, witness reports and contemporary documents describe a rapid closure of the echo off the left wing near the WT point, a speed in the order of at least hundreds of mph (Werlich's map overlay) or thousands of mph (written statements) which can't be explained by any change in the reflection geometry between aircraft and layer in a matter of seconds.

Second, the persistent anisotropy of the echo geometry over the rest of the approach path is unexplained by an elevated layer, as already explained in Section 6-11. (In fact there we saw that this could only be explained - if at all - by a failure of the antenna servostabilisation leading to a canted plane of rotation; yet there is clear photogrammetric evidence that the plane of antenna rotation was horizontal during the radar film sequence, as it should be if functioning correctly.) The plane is during this time flying straight and level with no cross wind and thus zero or negligible roll.

Third, the geometry of a ghost reflection due to a layer below a descending aircraft does not allow displayed range to stay constant when the aircraft is flying at over 18,000 ft above the terrain and when it is less than half this height near the end of its approach.

Finally, and conclusively, the smallest bearing angle from the radar in the B-52 nose to any part of the airframe in the radar pattern (an engine pod) is fully 40 degrees aft (see Fig. 21). Since a ghost echo is always displayed at the azimith of the primary reflector it is not possible for the UFO echo, always many degrees ahead of this angle, to be a ghost produced by reflection from the B-52 airframe.

graphic from shough analysis

Fig. 21. Plan of B-52 showing smallest bearing angle to any part of the airframe from the nose-mounted radar.

7. Conclusions ››