In 1923, the worst loss of US Navy ships in peacetime occurred on the foggy night of 8 September when a squadron of 14 destroyers, on a speed run down the California coast, turned left in the wrong place and plowed into the rocks of Point Honda. Seven of the warships were wrecked, and 23 sailors died despite heroic efforts by the ships' crews and rescuers on shore.
Human Error or Tsunami Effects?
In the aftermath, the ships' captains were formally assigned the blame for "navigational complacency." The flagship's navigators did not fully trust the innovative radio beacons that had recently been installed along the coast. They overruled the radio readings in favor of old-fashioned dead reckoning, which had placed the ship several miles north and farther offshore from where it actually was. The other ships obediently followed when they might have known better.
Naturally no one wants to think that they can be so badly mistaken. Years later, an article in the Naval Weekly claimed that "there is in the mind of every naval officer who was navigating that fateful night, a great belief that natural causes beyond their control contributed to the disaster. Countless runs down the Pacific coast under identical conditions have been made by naval vessels always with safety." During the official inquiry, some argued that the currents along the coast had been disturbed by tsunami waves from the Great Kanto earthquake a week earlier. Today some people still wonder.
The Great Kanto earthquake struck in the subduction zone off central Japan just before noon on 1 September local time. Today it is assigned a moment magnitude of 7.9, just short of "great" size, but the word "great" in its name is retained by long custom. And for good reason: the cities of Tokyo and Yokohama lost more than half of their housing stock from the shock and ensuing fires, and nearly 150,000 lives were lost. Since 1960, Japan has observed the first of September as Disaster Prevention Day.
The earthquake resulted from a rupture beneath Tokyo Bay and the Kanto Plain to its west. A tsunami as high as 10 meters struck the local coasts, but the greatest share of damage was due to shaking and fire, made worse by a typhoon that swept through the region. (Indeed, September and October are typhoon season in Japan.)
The fault that ruptured lies between the Philippine Sea plate and the overlying North America (Okhotsk) plate. However, the Eurasia and Pacific plates are not far away, and the tectonic picture of this part of Japan is very complex. Studies over the years, including a re-evaluation of the evidence published in 2006, have modeled the earthquake not as a classic megathrust event, like the March 2011 Tohoku quake, but as an oblique event that mixed thrust (convergence) and lateral (sideways) motion. Moreover, the modeled slip surface is deep below the land area of Japan rather than at shallow depths near the Sagami Trench, where the rupture might have lifted the seafloor to spawn tsunamis.
Knowing what we know about the 1923 earthquake, and tsunamis in general, today, I think it's possible to have a better-informed opinion about the facts of the Point Honda disaster 90 years later. Here's mine.
What Are the Facts?
We do know, as we did in 1923, that earthquakes can affect areas well beyond their epicenters through long-period ocean waves. There are two questions to consider: were there signs of tsunamis recorded in California, and was it plausible for tsunami effects to be significant at the time of the Point Honda disaster?
The pointhondamemorial.org site collects a great deal of documentation around the incident and its aftermath. It quotes the court of inquiry's finding "that no unusual current conditions existed, but that the set to the north and east was caused by bad steering, together with a certain amount of current which, while not explicitly laid down in the Sailing Directions, may be expected at any time in any direction and should be guarded against by the careful navigator." What was the state of the sea that day?
The site quotes several different accounts of current conditions. According to the San Francisco naval operations center, a ship going north the same night reported "a strong south and inshore current from Arguello to Piedras Blanca. Coast and Geodetic Survey reports slight variation in tide." The captain of the liner Cuba, wrecked on San Miguel Island before dawn on 8 September, told a reporter "that a mysterious change in the normal ocean current pushed him southward and east of his estimated position coming up en route to San Francisco." Both of those reports have the current going the wrong way to have put the destroyers north of their expected location.
So much for currents. What about waves? Today we have a good idea how fast Pacific tsunamis propagate. The earthquake was at 11:58 a.m. local time on 1 September, which is 7:58 p.m. on 31 August in California. In 2011, the tsunami from Japan arrived at California in about 10 hours, and disturbances monitored by tide gauges continued for days afterward. A tsunami from the Kanto mainshock, then, would have arrived early in the morning of 1 September in California, a full week before the Point Honda disaster.
In its 4 September issue the New York Times published reports of high waves in California, including swells that topped the Los Angeles Harbor breakwater and reached unusual heights in Santa Barbara. The timing is not clear, but the waves appear to have arrived on 2 or 3 September. "Naval observers said no storm of any size had been reported anywhere on the Pacific," the Times stated, "and they thought the huge swells due to reaction on this coast from the Tokio catastrophe." Today, I cannot give that report any weight. I can call it just conceivable that tsunamis caused by aftershocks in Japan combined with reflected tsunami waves in the Pacific to amplify distant storm swell. After a large tsunami, there can certainly be swirls or eddies lasting for days, especially in the rugged continental borderland off Southern California, but persistent changes in regional currents are not part of the post-tsunami repertoire for the simpler Central California coast.
On 13 September the Times summarized the previous week's developments:
The precise nature of the changes off the California coast are still unknown to those whose business is to record, chart and report such disturbances, but there have been at least two positive indications of a change in the nature of the Pacific in this locality.
First, there has been noted a definite and strong inshore set of the ocean current at that point.
Second, there have been unusual and dangerous waves, at time reaching points ten and twelve feet in excess of the highest normal waves. The area of disturbance appears from the present reports of Point Arguello to beyond the American-Mexican border line.
Whether the oceanic changes on this side are the direct result of the upheaval in Japan, or whether they are due to a secondary disturbance in the bed of the ocean a comparatively few miles off the California coast, is yet unknown.
The change in current was most likely just one of those things. Every year in California, the ocean is a little different. There are cyclic changes in the North Pacific that last for decades. And today we have good maps of the seafloor. A submarine disturbance big enough to alter the coastal currents would have raised an unmistakable landslide tsunami. So I can't conclude from a single week's evidence that there was any "change in the nature of the Pacific" at all.
Finally, the problem with the destroyer squadron was that they were too far north, not just too close to shore. None of the evidence bears on that, except two reports of currents that would have put the ships too far south instead.
Revise the Verdict?
The Point Honda disaster was a total tragedy, and a scandal besides. The court of inquiry, while praising the heroism of the crews in crisis, recommended that 11 officers be court-martialed. But those courts-martial ended with only two convictions.
I believe that the judge may have had doubts about the state of the sea and no good science to rely on; also, there was genuine confusion during the voyage for which the lead navigator, set on his speed run and his dead reckoning, was the prime culprit. As one of the captains later said, "I kept track of the navigation as well as possible but we had nothing to go on but dead reckoning. We couldn't take soundings and were forbidden to ask for radio bearings."
Today I think the verdict was right: there was no "act of God" in the form of tidal waves from Japan. Even if there had been, the navigators should not have discounted the radio beacon system. If in doubt, they should have slowed down and worked it out. The only good outcome of the Point Honda disaster was that the evidence of human errors behind the accident led to improvements to the navigation system.
Sometimes the military does formal reviews of historical actions to correct injustices. If that were done today, it would be interesting for the relevant specialists at NOAA to model the state of the atmosphere and ocean in September 1923 from the old records. Then an expert in seismo-forensics could model the tsunami, too, using today's techniques.