Techniques Employed in Imaging
Lake Mead’s B-29 Bomber

INTRODUCTION

In July 1948, a B-29A “Superfortress” Bomber (#45-21847) was conducting classified atmospheric research as part of the Upper Atmosphere Research Project #288. On a low altitude pass over Lake Mead in southern Nevada, the pilot misjudged his altitude and the plane struck the surface of the water at nearly 250 miles per hour. While all of the crew survived the ordeal, the aircraft sank in over 250 feet of water. For more than 52 years, the big bomber eluded searchers and remained hidden in Lake Mead’s chaotic terrain.

RESEARCH

Lake Mead is a man-made reservoir created by the impoundment of the Colorado River in Black Canyon along the Nevada / Arizona border, with the construction of Hoover Dam. The lake is divided into three major basins along the previous Colorado River, and the Overton Arm, a section of the lake extending north from the intersection of the Virgin and Colorado Rivers. Initial research into the loss of the aircraft yielded a series of news sources indicating that the B-29 impacted somewhere in the lower portion of this Overton Arm. (See figure 1)

 
*Figure 1: Map of Lake Mead showing general region where B-29 was thought to have crashed. Map derived from NOAA Nautical Chart 18687.

To narrow the search area further, a series of flight path projections were plotted and the crew accounts from the official accident report considered. Detailed observations were made only by the civilian scientist, Mr. John Simeroth. Clues offered in other crew members’ accounts were weighted subjectively and correlated to Simeroth’s observations. Several topographical features imposed restrictive limits on the flight path projections. These included the mountainous terrain comprising the Overton Arm’s western shoreline and the Overton Islands, located to the north of the assumed impact site and spanning the entire width of this portion of the reservoir. Analysis of this information led to the formation of a primary search area two miles long and ¾ mile wide. Depths within the search grid ranged from 110 feet to nearly 400 feet deep. The bottom of Lake Mead in this region consists of particularly rough terrain, characterized by steep cliffs, buttes, deep valleys and crevices. The flatter areas are deeply carved alluvial floodplains. Sediment buildup varies throughout the entire reservoir. Based on the work of Twichell et al. (1999) conducted in the western portion of Lake Mead, post impoundment sediment accumulation in the majority of the search area was anticipated to be on the order of centimeters to perhaps a meter. The exception would have been that part of the search grid which overlapped the Virgin River basin, where accumulations on the order of meters to tens of meters would have been expected. The following cross section shows sediment distribution in an area topographically similar to the search area. (See figure 2) Had the aircraft come to rest in the Virgin River bed region, it likely would have been engulfed completely, negating sidescan as an appropriate search device. As more than 75% of the search grid was either alluvial floodplain or elevated topography, it was assumed that at least some portion of the bomber would be visible to sonar.

**Figure 2 (Twichell et al. 1999) - Post impoundment sediment distribution in a portion of the Boulder Basin of Lake Mead. This area is topographically similar to the B-29 search area in the Overton Arm. Thick sediment cover, up to 15-20m is seen to have accumulated in the old river basin, with the raised terrace areas showing no significant sediment buildup. This same profile was expected along the Virgin River basin area in the Overton Arm, where the B-29 survey would take place.

SURVEY

The search tool employed for this survey was a Marine Sonic Sea Scan PC system utilizing a 300khz 2 channel towfish with 170 meters of available cable. Normally, for deep survey work with this type of lightweight system, weights would be attached to the towfish or a depressor vane would be utilized to aid in compensating for the drag created by the length of cable in the water. However, the terrain in this region presented some interesting challenges which could be addressed by deep-towing the unit without modification. The first survey runs were conducted with a depth sounder only on the identical tracks to later be run using the sidescan towfish. GPS coordinates were taken on all major geological features within the search area where the bottom profile deviated 50 feet or more vertically over a 1000 foot distance. Additionally, depth ranges were noted into the sonar’s navigational plotter for each search track. These steps were taken to provide the greatest safety margin in avoiding towfish impacts. The initial scans were made at a range setting of 200m (656ft) per channel. This setting was anticipated to provide sufficient resolution to detect the aircraft even if it had been broken up or partially buried in the lake bottom. Search tracks were spaced 100m (328ft) apart to provide 50 percent overlap. This was done to account for the lake floor’s rough topography. Even an object as massive as a B-29 Bomber could have been easily concealed from sonar by the unevenness of the terrain. This high degree of overlap allowed relatively complete imaging of the bottom and minimization of blind spots created by geological features. The sonar towfish was “flown” approximately 20m above the bottom, or 10% of the horizontal range setting as suggested by Fish & Carr (1990). As average depths within the search grid approached 300 feet, nearly all of the 170m of cable available was utilized and drag considerations limited survey vessel speed to under 2 knots. The towfish layback was in excess of 200 feet, which had both a positive and negative impact on the survey. Unfortunately, with all of the towfish altitude corrections which needed to be made for terrain, layback became an estimated and somewhat dynamic value. The layback figure is manually input; towfish position and thus target location is extrapolated by the sonar computer, based on the assumption that the towfish is located directly aft of the GPS receiver. Errors in this distance estimate could be compensated for by making reciprocal passes and comparing target locations within the navigational plotter. Once the same target is marked on reciprocal passes, even with a layback error or no layback set, the actual target location will rest along the line connecting the marked points. Horizontal offsets, however, if not entered into the layback calculation precisely, will translate into a non-linear positional error for both the towfish and any detected target. Thus from a precision standpoint, very straight survey lines need to be maintained or the great distance between the GPS antenna and towfish would magnify even the smallest horizontal error. There was a positive side to this method as well, however. By deep-towing a lightweight sidescan towfish, it became possible to attain the proper configuration to accurately image the bottom, while at the same time maintaining adequate control of towfish altitude to adjust for the at times rapidly changing depths. This control was found to be very precise in a calm lake environment. Use of the depth sounder at a 50khz setting caused minimal interference with the sidescan and provided at least 30-45 seconds lead time on depth changes in which to make an altitude adjustment to avoid an impact. The towfish was raised or lowered by making speed adjustments to the survey vessel. While it may have been possible to utilize a forward scanning sonar and thus modify the towfish with a depressor or weights to operate with less layback, forward scanners generally work at 200khz, which would have created a great deal of interference with the sidescan and necessitated use of a band pass filter. Additionally, operating with, for example, a depressor vane to shorten layback in this scenario, could cause unpredictable towfish response to velocity changes.

Image 1:  150m single channel image shows the B-29 resting on uneven terrain in the upper right of the image.

Image 2: 75m image showing 21’ tall vertical stabilizer, engine #1 propeller and orientation of wreck on lake floor.

The first hit was detected on the first leg of the survey’s third day and had the appearance of a small cross resting on uneven terrain. A repeated pass was made on a reciprocal heading to confirm the presence of the anomaly and estimate a true location. The cross-like structure of the anomaly was noted again on this repeated pass and even scrolling in real-time the object resembled an aircraft. A third pass was now made at a higher resolution of 150m (492ft) on a single channel. This scan, shown in Image 1, confirmed the profile of an aircraft. In fact, the scan revealed an aircraft with the nearly unmistakable outline of a B-29 Bomber. Image measurements showed the wingspan to be the proper width of about 140 feet, but the fuselage length roughly 20% shorter than a B-29’s 99 feet. Damage to the aft section of the fuselage and empennage, coupled with the orientation of the wreck on a downward slope, was later determined to be the cause of this discrepancy. To obtain higher resolution imagery down to 50m (164ft) per channel, the towfish had to be flown near the wreck within 5-6m (15-20ft) of the rapidly changing bottom. While risk of impact with the bottom was present, the high resolution scans revealed remarkable detail and proved worthwhile. Visible features now included the attached remaining engine and propeller, damage to the nose and empennage and the excellent state of preservation of the wings and majority of the fuselage. It became possible to determine that the tail section was elevated on a slope, the wings had settled over an old flood channel but were still attached and relatively undamaged and the forward section was elevated off the lake bottom. The 21 foot high vertical stabilizer also appeared to be intact. (See Image 2)

Documentation

Visual documentation of the wreckage now became the challenge. Declining water levels over the 2000-2001 seasons made technical diving to the wreckage more practical. Still, the B-29 was located more than 200 feet beneath the surface. Ambient light penetration at these depths is minimal and the dive team quickly discovered that even the powerful HID lights on the camera systems proved a lack of sufficient illumination. Supplemental systems were constructed of sealed beam halogen flood lights, PVC, ABS piping and silicone. The systems were powered by sealed gel batteries contained within ABS pods. The units were designed to be nearly neutrally buoyant and provided an additional 1.2 kilowatts of halogen light, enough to adequately illuminate selected portions of the bomber for filming. (See Images 3-6)

Image 3: Diver illuminating tail number on vertical stabilizer.

Image 4: The nose, damaged upon impact with the lake floor.

Image 5: Pilot’s escape hatch and crumpled nose, illuminated by supplemental light systems.

Image 6:  Flight engineer’s console, with many instruments still readable even after 54 years underwater.

REFERENCES

Fish, J.P. and H.A. Carr “Sound Underwater Images: A Guide to the Generation and Interpretation of Side Scan Sonar Data”, Lower Cape Publishing, Orleans, MA; 1990

Marine Sonic Technology, LTD., “Sea Scan® PC Operator’s Manual Version 1.5”, White Marsh, VA; 1998

Twichell, D.C., Cross, V.A., Rudin, M.J., Parolski, K.F. “Surficial Geology and Distribution of Post-Impoundment Sediment of the Western Part of Lake Mead Based on a Sidescan Sonar and High-Resolution Seismic-Reflection Survey”, U.S. Geological Survey, Open-File Report 99-581, Woods Hole, MA; 1999

* Figure 1 Map derived from NOAA Nautical Chart 18687 12th Ed., US Department of Commerce, Washington DC; 1999

**Figure 2 Courtesy Twichell, David et al. U.S.G.S. Open File Report 99-581