Home > Naval History Magazine > 1999 > December Issue
Products Site Content

Pearl Harbor—Attack from Below

By Commander John Rodgaard, U.S. Naval Reserve (Retired), Peter K. Hsu, Carroll L. Lucas, and Captain Andrew Biache Jr., U.S. Naval Reserve (Retired)

Naval History, December 1999

After careful analysis of a photograph taken by a Japanese aviator over Pearl Harbor, a team of experts is convinced that the 7 December 1941 attack was more than an air raid.


For years historians have relied on survivor and eyewitness accounts, material testing, and, if available, declassified intelligence reports to help gather evidence to piece together an event. Advances in computer technology now have allowed photo imagery analysts to enhance and sharpen images, changing the way the world is viewed. For example, digital photo imagery analysis techniques used to measure an inward-bent hull plate have re-energized the debate about the reasons for the destruction of the USS Maine (see April 1998 Naval History).

History has recorded that five Imperial Japanese Type- A-class midget submarines, deployed as an advance force to attack the Pacific Fleet in Pearl Harbor, failed to inflict any damage during the 1941 attack. Results from digital imagery, engineering, and forensic analyses of a single dramatic photograph taken that infamous December morning by a Japanese aviator, however, now refute that view.

Japanese Attack Plan

Admiral Isoroku Yamamoto's Pearl Harbor operational strike plan was one of combined arms, involving the use of carrier-based aviation in concert with the fleet, including the five midget-type submarines. The carrier strike was to consist of two waves of torpedo and bomber aircraft, with Zero fighters as escorts. The fleet submarines would engage certain capital ships from Pearl Harbor, and they would interdict the sea-lanes between Hawaii and the U.S. mainland. The five midget submarines would enter the harbor. Figure 1 depicts the approach routes of the aerial strike and the operational areas assigned to the nearly three dozen fleet and midget-type submarines. According to Yamamoto's plan, the five Type-A-class midgets were to launch from their mother submarines in the early hours of 7 December. They were then to proceed from their launch areas to Pearl Harbor before the aerial strike (see Figure 2). Each midget would enter the narrow (400 yards wide) and shallow (50 feet deep in the middle) channel, and pass under the antitorpedo nets that hung down to a depth of 35 feet. Once inside Pearl Harbor, the midgets would wait for the aerial attack to begin; then they would launch their torpedoes against targets of opportunity (preferably battleships) as the attack unfolded.

Captain Hanku Sasaki was in command of the special attack force of the five mother-fleet submarines, each carrying a single, two-man midget sub. The original orders given to the 10 Ninja submariners on 14 November 1941 were specific: Attack the surviving capital ships on the night of 7 December and escape. Yamamoto did not want this portion of the operation to be suicidal. Lieutenant Naoji Iwasa, the leader of the midget submarines, asked permission to launch his attack immediately after the air strike instead of waiting for night. Admiral Yamamoto granted this permission.


Figure 1: Operation Hawaii—the Japanese attack plans for 7 December 941 formulated by Admiral Isoroku Yamamoto (right) and analyzed by Commander Minoru Genda—included nearly three dozen fleet and midget-type submarines. For clarity, the midget submarines are not depicted in this graphic.

Japan's Special Weapon

As designed in 1938, the Type-A midget submarine was a remarkable innovation. She was a two-man sub, consisting of a junior officer who commanded the boat and a petty officer who manipulated the control valves and ballast. Figure 3 shows that the Type-A measured 80 feet in overall length, and her hull measured 6 feet in diameter. She displaced 46 tons when submerged. Two bolted hull joints permitted the submarine to be separated into three sections. She was propelled by a single 600-horsepower electric motor, driving a single-geared shaft with a pair of contra-rotating propellers. Acid-cell batteries supplied power, with recharging accomplished by a mother submarine. Top speed of this class was 23 knots on the surface and 19 knots submerged. At top speed she had an endurance of 55 minutes. At a submerged speed of 4 knots, however, the midget had an effective range of 100 miles or 25 hours. Each carried two 18-inch torpedoes mounted one above the other. The size of each torpedo warhead approximated 1,000 pounds of explosive—more than twice the destructive firepower of those carried by the Japanese attacking aerial force.

Determining the Presence of the Submarine

In 1993, while visiting Hawaii on business, coauthor Rodgaard, a senior image analyst at Autometric, Inc., met Dan Martinez, historian of the National Park Service's USS Arizona (BB-39) Memorial, and Burl Burlingame, a noted Pearl Harbor historian and local journalist. Martinez showed a copy of what many consider to be one of the 20th century's most dramatic photographs. It was taken from a Nakajima B5N2 “Kate” torpedo bomber during the first attack wave. The Japanese government distributed this aerial photograph to show the successful attack to the world. But Martinez saw something intriguing in it and asked Rodgaard if he could determine whether a tiny object in the photograph was one of the Japanese midget submarines (See Figure 4).

Rodgaard subsequently secured an original, higher-quality duplicate image of the photograph from the U.S. Navy Historical Center in Washington, D.C. Then he, together with Carroll Lucas, Dewey Houck, and Tim Hosek of Autometric, Inc., interpreted the image and measured the object to determine whether evidence supported the premise that a midget submarine was present at Pearl.

Figure 2: In an artist's conception by coauthor Peter K. Hsu, Ensign Masaharu Yokoyama and Petty Officer Second Class Tei Uyeda launch in a Type-A midget from their mother sub in the early morning of December 1941. Upon reaching Pearl Harbor, they were to wait for the air attack to commence.



Figure 3: Prior to launch, all but two of five tie-downs were removed, and the midget was given positive buoyancy. The mother took on water, achieving negative buoyancy, and maintained 1 to 2 knots' forward speed. Using a telephone connection, the midget skipper signaled the mother, which then released the ties, sending the midget in the opposite direction.


The Autometric team presented its findings to the USS Arizona Memorial on 7 December 1994. In 1998, Peter Hsu conducted a forensic study based on the 1994 findings. This study looked at underwater explosion phenomena and shock-wave propagation, elaborating the 1994 findings. Hsu's results strengthened evidence that a midget submarine fired two torpedoes at the battleships USS West Virginia (BB-48) and USS Oklahoma (BB-37) during the attack.

The 1994 photographic imagery analysis of the Japanese aerial attack photograph consisted of five phases: The first phase involved photoscientists who reproduced the image and enhanced it digitally by applying various computer algorithms and techniques. The second was to establish the sequence of events within the scene. The third was to search the photograph for observables that lent themselves to the presence of a midget submarine. The fourth involved precise photographic measurements of the object that Martinez and Burlingame had thought was the submarine. And the fifth was to compare the size (length and height) of the object with the specifications of the Japanese Type-A-class midget submarine. In 1999, Hsu built on the 1994 findings and his 1998 analyses; his underwater shock analysis provided a sixth phase.

Phase I—Image Enhancement

The image was digitized into a software image file. Next, an edge-sharpening algorithm was applied to the overall image scene. The subject area, where the object was said to exist, was magnified digitally, until the image was about to lose its integrity. At this point, measurements of the object were taken.

Phase II—Chronology of the Image and Sequence of Activity

In the background of the attack photograph, smoke obscuring most of Hickam Field Army Air Base indicates that it already had been struck by the first wave of Japanese bombers (see Figure 4). This indicates that the photograph was taken after the initial attack run, which occurred at 0755. Smoke also can be seen rising from the USS Helena (CL-50) and the USS Oglala (CM-4), indicating that at least one, if not both, ships had been hit. The log of the Helena is specific about the time (0758) she was struck by a torpedo. In addition, history records that the first wave of torpedo bombers attacked Battleship Row within minutes after the attack on Hickam Field. Photo evidence shows that the first torpedo strike against the battleships USS California (BB-44), Oklahoma, and West Virginia had occurred. The California was struck by one torpedo, and she was leaking oil. The Oklahoma had been torpedoed and was listing to port. The West Virginia had been hit at least once, as indicated by several wakes.

The first Battleship Row attack occurred at 0758. The photo shows torpedo entry points and wakes originating from the Southeast Loch and located south-southeast of their targets. Concussion waves on the surface indicated that the West Virginia and the Oklahoma had been struck before the photograph was taken (See Figure 5). Water spouts associated with the torpedo detonation against the side of the ship already had dissipated. Considering this evidence, including calculations from shadow analysis, the best estimate for the time was between 0801 and 0803.

Phase III—Determining the Presence of the Submarine

The image shows a rectangular, black object present in the suspect area. This smaller object appears to be sitting atop a dark linear object. The composite shape is surrounded by the white-water area. Numerous concussion waves also are visible in the image. The concussion waves appear to have affected the object.

Three distinct rooster-tail sprays appear behind the composite shape. Each tail seems to coincide with the passing of a concussion wave. The height characteristics of the tail sprays grow toward the composite shape with the smallest tail, or the oldest, having occurred first. The sprays have a distinct characteristic shape different from the splashes one would see from torpedoes entering the water from the air, or from exploding ordnance.

Six torpedo wakes can be seen in the photograph. Four appear erased partially by concussion waves. The broadest wakes represent greater degradation. No torpedo wakes are visible beyond the easternmost splash points. The two narrowest (newest) wakes (numbers five and six) converge at a point behind the rooster tails. The evidence concluded that a line of convergence can thus be traced through the dark rectangular object, the rooster tails, and the intersection of the most narrow torpedo wakes.


Figure 4: According to photographic imagery and marine forensic and photogrammetric analysis this combat photo—taken during the first wave of the attack from a carrier-based Nakajima B5N2 "Kate" torpedo bomber—verifies the presence of midget submarines in Pearl Harbor. Splash points and torpedo tracks have been highlighted and numbered.

Figure 5: An enlargement of the area in question indicates the presence of black, rectangular object sitting atop a dark linear structure, followed by "rooster tail" plumes. The analysis have determined these images to be a midget submarine, its conning tower and contra-rotating propellers exposed by the concussion of a torpedo hit on the U.S. battleship West Virginia.

Phase IV—Photograph Measurements

Two objectives had to be achieved. First, it was necessary to locate the object geographically and position it relative to the battleships in the field of view. The second objective was to determine the object's height and size. To achieve these objectives, a least-square photogrammetric triangulation determined the position and attitude of the camera used by the Japanese.

Two aerial stereo photographs taken by a U.S. Army Air Corps mapping camera on 26 November 1941 and known ground points served as a control base for triangulation. A measurement of the object's pixel position on the digital photograph established its location in the harbor. Several objects with known locations were also measured to establish the validity of the triangulation resection.

The geographic coordinates of the object were calculated as 21 degrees 21 minutes and 4.2 seconds North latitude, by 157 degrees 57 minutes and 12.5 seconds West longitude. The minimum height and length dimensions of the object were thus determined by the same techniques and calculated to be 1.25 meters high and 18.2 meters long.

Phase V—Comparison of Dimensions

Comparing the physical dimensions of the Type-A-class midget submarine with the photogrammetric measurements, the vertical height and length measurements of the object coincide with those derived from the analysis.

Phase VI—1998 and 1999 Forensic Engineering Analyses

In order to establish a methodology to conduct forensic analysis of the shock-wave effects of the underwater explosion phenomena associated with the torpedo attack, the research team postulated a chronology of images observed and an activity scenario.

History documented that the first attack wave occurred at 0758. Based on previous rudimentary shadow analysis, the team verified that the combat photo was taken between 0801-0803, thus establishing the latest time at about 0803. In the absence of both a huge water spout associated with torpedo detonations against the side of the ship and torpedo bombers flying over the battleships, the forensics team concluded that the attack group had completed its attack run and observed the following (See Figure 4):

1. The distance between the splash points of the aerial torpedoes aimed at (wakes 2 and 4) the West Virginia is 450 meters. Assuming the average torpedo speed was 40 knots, the running time is estimated to have been about 22 seconds.

2. Huge water spouts observed from another photo taken during the attack measure about as high as two ships' lengths. A water mass falling from this altitude would take an estimated 30 seconds, with approximately another 30 seconds for the dense mist to dissipate partially.

3. In Figure 4, a water spout has formed next to the port hull plating of the West Virginia. In the absence of any “Kate” or “Val” bombers over the battleships, the hypothesis of torpedoes being fired from a submarine becomes viable. In the photo, the first torpedo has exploded, and the second has just hit the Oklahoma. Since the midget torpedoes were Type 97, which contained a 1,000-pound warhead, and the air-drop torpedoes were Type 91, which had a 452-pound warhead, the explosive energy to be considered in the underwater explosion analysis ranges from 452 to 1,000 pounds.

4. White-water disturbance surrounds the suspect submarine.

5. Three triangle-shaped water sprays, south and in line with the submarine, exhibit a typical spray pattern associated with contra-rotating propellers striking the approaching surface concussion waves.

6. The concussion waves were propagated radially from the West Virginia. The distance between each wave ring appears to be substantially large, migrating with significant momentum. (Reticule measurements indicate each concussion wave ring was about 2 to 3 feet high).

7. With regard to the West Virginia image in Figures 4 and 5 and in the explanation of the water spout in item 3 above, the forensics team performed an analysis of the relationship of the torpedo detonation and its generated cavitation region.

During an underwater explosion event, the chemical reaction of the solid explosive yields gaseous products and exerts high explosive pressure on the surrounding water. The disturbance (compression shock wave) near the charge propagates radially at 3 to 5 times the speed of sound. The surrounding water is compressed and attains an extremely high radial velocity. The compression shock wave reflecting from the free surface results in a tensile reflected wave, causing a region of water to cavitate, thus forming a bulk cavitation zone. The water refraction near the surface wave has a vertical velocity. The velocity of a water particle is the vector sum of the velocities produced by the associated pressure. Thus, the midget submarine is within the cavitation zone and was buffeted by a vertical motion, exposing the conning tower and its contra-rotating propellers.

The magnitude of the vertical velocity and the motion history of the water particles behind the shock front is dependent on its location relative to the charge, the free-surface effect, and the bottom reflection, which was compounded by explosive gas pulsation, especially in shallow water. In general, such underwater explosion energy will lift an object vertically in the cavitation region.

Extrapolations of the bulk cavitation zone calculated from selected charge size and attack geometry indicated that the midget submarine was within the cavitation region. The 40-foot shallow draft of Pearl Harbor, with its bottom reflection potential and the gas bubble pulsation would have exaggerated the cavitation range, causing the submarine to be launched upward by the explosion.

As the submarine broached near the surface, it increased speed suddenly and headed into the oncoming waves caused by the concussion. Water crashed into the propeller disk, forming three distinct rooster-tail sprays synchronous with each passing wave. The third smallest rooster tail had dissipated somewhat by the time the photo was taken. This underwater explosion analysis has strengthened evidence of the existence of a midget submarine through a positive correlation between the motion history of the midget submarine and the observed surface disturbance generated.

Summary of Combined Analyses

Photographic evidence indicates that a Type-A-class midget submarine penetrated Pearl Harbor on 7 December 1941. She positioned herself in the Southeast Loch and waited to launch her torpedoes at the capital ships on Battleship Row. These ships were identified as the USS Oklahoma and the USS West Virginia. When the aerial strike started, the midget fired her first torpedo at the West Virginia, then turned to port, fired her second torpedo, and increased speed. With her second torpedo heading toward the Oklahoma, the midget began her run toward the harbor entrance to escape. During this time, the submarine was struck by a series of concussion waves struck the submarine; because of a loss of trim owing to the release of her torpedoes, the effects of her speed, and the oncoming concussions, the midget broached the surface as she was buffeted by cavitation effects. She pitched vertically in the cavitation field generated by the underwater explosion, which exposed her rudder and contra-rotating propellers and caused rooster tail sprays to form. An aerial photograph captured the entire event.

Figure 6: An artist's conception by coauthor Hsu depicts a midget submarine scoring a torpedo hit on the West Virginia at a depth of 10-15 feet beneath the water surface, or below the ship's armor. Torpedo wake numbers 5 and 6 coincide with the calculations of the research team, which contends that the West Virginia and the Oklahoma were attacked from below.

Six torpedo wakes are evident in the photo. Four appear partially obliterated by the concentric concussion waves emanating from the recent strikes on the battleships. The remaining two are thinner and affected less by the wave action. These two wakes terminate in the midst of the water disturbance where the suspect object is located. Two of the broader aerial torpedo wakes also terminate at these ships, with the third broader wake aligned in the direction of the California. The broader torpedo wakes could indicate greater wave action degradation, thus making them older than the narrower wakes. The two narrower wakes converge at another area of water disturbance nearly in line with the aerial torpedo entry points but nearer to Battleship Row. At the point of convergence, a thin, dark object can be seen projecting above the surface of the disturbed water surrounding it. Slightly to the rear of this object, three spray splashes appear, with the largest nearest the object and the smallest the farthest away. Each spray plume is aligned with the concentric concussion waves that have just reached the area.

The observed torpedo track (see Figure 4, Track No. 5) indicated alignment with the water plume on the port side of the West Virginia. Because of the water's optical reflection and parallax effect, the research team determined that the torpedo was at a depth of about 10-15 feet beneath the water surface. This alignment triangulates the impact point to be beneath the ship's armor but where the vertical plane intersects with the observed plume (see Figure 6).

The final disposition of this midget submarine is still unknown. Photographic and forensic engineering evidence, however, indicates that she entered Pearl Harbor successfully on 7 December 1941 and positioned herself in the Southeast Loch. When the aerial attack against Battleship Row was under way, she fired both of her torpedoes against the West Virginia and the Oklahoma.

The authors all are marine forensic and imaging specialists. Commander Rodgaard is a senior image analyst for Autometric, Inc.; Mr. Hsu is a technical director for Techmatics, Inc., and is a member of the Marine Forensic Panel (SD-7) of the Society of Naval Architects and Marine Engineers; Dr. Lucas is a staff member of CL Imagery Exploitation Services, and Captain Biache is an independent photographic imagery specialist.

More Pearl Harbor Resources

Subscribe to Naval History Now!

Back to 1999
Back to Naval History

back to top

Members Only  Privacy Statement  Contact Us
All items on this page copyright © 2006 U.S. Naval Institute, Annapolis, Md., unless otherwise indicated.
Any technical problems or questions should be directed to Webmaster@usni.org.