Copyright © 1992, 1999, Kevin T. Kilty, AllRights Reserved
"What is 10 inches high in some places, 4500 square miles and worries the hell out of laymen and professionals alike?" asked a popular science magazine in 1976. The answer in 1976 was "The Palmdale Bulge." With the advantage of hindsight we know the real answer to have been "Studies of the Palmdale Bulge," for now we wonder if the bulge ever existed at all.
Employees of the USGS found the bulge in 1975 by comparing some old surveying work (done in 1959?) with some recent work. The bulge appeared sometime between 1960 and 1974. Since a similar bulge foreshadowed the San Fernando earthquake of 1971, and this one was occuring on a section of the San Andreas fault that produced the 1857 earthquake, geologists saw it as an ominous feature.
"There is no sense of alarm" one California official said, "but we are treating the Palmdale Bulge as a threat."
In 1976 President Ford authorized 2.6 million dollars for earthquake prediction research, 2 million of it for the Palmdale Bulge. A more detailed study was needed to better define the feature, but most researchers felt the most rigorous study would not alter the gross pattern of vertical movement.
However, by 1977 the areal extent of the bulge was fixed at 32,400 square miles, which was 8 times the original estimate, and it was found to be 12 inches high in places. Its behaviour was also erratic. One portion of the bulge had dropped 7 inches between 1973 and 1977. Thus, the bulge was now a rapidly changing feature in addition to being ominous.
In January 1978 300 Scientists fanned out for an extraordinary geological survey. They conducted a 1.4 million dollar program to study the bulge over a 3 month period and get a stop action picture.
Yet additional studies could not clarify the bulge or its meaning. In December 1979 Science News reported that one group of researchers finds the bulge expanding rapidly while another group of researchers claims it may never have risen at all. One section of earth from Palmdale to Pasedena fell 16cm between 1974 and 1976, and Very Long Baseline Interferometry (VLBI) showed that a section of the bulge expanded north-south 20 cm in only 3 years. Barry Raliegh of the USGS says "It seems likely that it means...a large earthquake sooner than earlier observations indicated." However, two researchers at UCLA, David D. Jackson and W. Lee, found something strange about the bulge.
They had taken the survey line beginning at San Pedro and ending near Palmdale, and extracted long wavelength features using a trend surface fit. They produced short wavelength features as the difference between the observations and the long wavelength features. One would not expect the long- and short wavelength features to behave the same over time. Yet, on this particular survey line they did, which implied a systematic error in the data. Most likely, said Jackson and Lee, corrections for thermal expansion of the rod were made in error. Jackson and Lee also suggested another source of error which a group at the USGS eventually concluded could account for nearly all the bulge. This second source of error arose because the survey lines rose from sea level to over 1000m. Thus each level siting was generally away from the ground surface in one direction and toward the ground surface in the other. Each observation thus sited across the temperature gradient in the air and was systematically refracted.
It wasn't as though refraction caught everyone by surprise. Studies had shown refraction to be a negligible concern, but these also had been done in mild climates or on relatively level ground. The USGS group showed that these earlier results did not apply to conditions at the Palmdale Bulge.
Over the next few years bulge doubters grew steadily in numbers and completely overwhelmed the believers. From 1976 to 1979 there were 14 articles in scientific journals that mentioned the Palmdale Bulge specifically, and an unknown number of other articles like "Aseismic uplift in Southern California" that dealt with the bulge without mentioning it by name. From 1980 through 1987 there were only 5. There are no citations to it in Georef after 1987. A summary article on earthquake prediction published in 1987 didn't even mention the Palmdale Bulge. In most respects, therefore, the Palmdale Bulge fit exactly the typical instance of pathological science as Langmuir articulated it. Ironically the last reference that I find was an abstract for a talk by R.O. Castle, the original discoverer of the bulge. In his talk he maintained that evidence against the bulge could not withstand scientific scrutiny and was, as he put it, "modern sophistry."
How did so much of the geophysical community come to accept the existence of the bulge? Geophysicists are a hopeful lot, but there were many warning signs about it. There was, for instance, no clear theory of why it rose in the first place. While I agree with R.O. Castle that demanding an explanatory theory in advance is inconsistent with "scientific method," but to articulate no theoretical reason for believing in observations near the level of noise certainly stretches the scientific method. Next the rapidly changing shape of the bulge certainly made it mechanically suspicious(1); and, finally, certain behavior like uplift magnitude depending on length of survey siting suggested internal inconsistency.
Probably the most important fact about the bulge was that its existence rested upon tiny residuals teased from an enormous number of observations. The San Pedro to Palmdale survey line consisted of approximately 25000 separate levelings. A systematic error of only 10 microns per leveling, which was proved to be possible in the relevant setting, could explain the entire apparent uplift. Jackson thinks a typical systematic error was probably only one-fifth this, but nevertheless, the Palmdale Bulge was a feature perilously close to the noise level.
1. In my Ph.D. dissertation I made no judgements about the reality of the bulge, but I suggested that mechanically it had no apparent connection with long-term tectonic processes.