California scientists have been surprised by the severity of damage done in Mexico City by a relatively distant earthquake and are expressing concern over its meaning for a major quake expected to strike this area in the next 20 years.

The death toll from Thursday's quake, centered more than 200 miles from Mexico City, is at least 2,000, compared to the approximately 450 who died in the San Francisco earthquake of 1906, California's worst. Scientists said analogous soil problems suggest that populous seaside cities such as Long Beach could be seriously damaged even by a quake centered dozens of miles away.

"The most significant thing about the Mexican quake is the extent of the damage at such a long distance from the epicenter," said T. Toppozada, a seismologist with the California conservation department.

"A lot of us were surprised at that much damage at such a long distance," said Thomas H. Heaton, a U.S. Geological Survey seismologist based at the California Institute of Technology.

Mexico City was built on a dry lake bed, whose soft clay vibrates like jello when jolted. Previous quakes and tremors apparently left buildings there vulnerable to substantial damage in the latest quake, which measured 7.8 on the Richter scale.

American scientists were meeting Thursday in Washington to discuss a somewhat similar earthquake side effect to which California is prone: "Soil liquefaction" can suddenly turn pockets of water-saturated sand into liquid during a tremblor. Caltech scientist George Housner told the group the public is largely unaware of the phenomenon, although it caused severe damage during a 1964 quake in Niigata, Japan, and during the 1906 San Francisco quake.

Among the California areas vulnerable to soil liquefaction is Long Beach, population 381,700, 20 miles south of here. Much of the city sits on relatively loose soil deposited by streams and construction filling. A 1933 quake there killed 120 people, the highest earthquake death toll in southern California.

A map by the conservation department shows several other parts of southern California with extensive dirt fill where liquefaction could be an important factor.

Heaton said weather is also a factor. Areas that experience a few seasons of heavy rain, as southern California has in recent years, will have a high water table and be more prone to soil liquefaction than are dry areas.

Riley Chung, director of the geotechnical engineering and hazard mitigation division at the National Academy of Sciences, said the scientists discussed possible responses to the hazard. These include abandoning structures in dangerous areas, adding deep pilings or digging relief wells to drain water.

A 1981 study by the Federal Emergency Management Agency concluded that there is a better than 50 percent chance of an earthquake of magnitude 8 or more along the San Andreas fault, 35 miles north of Los Angeles, in the next 20 to 30 years. The study estimated that fatalities from such a quake could range from 3,000 to 14,000, depending on the time of day.

That would be the highest single death toll on American soil since the Civil War.

In recent years, scientists here have tried to pinpoint the likeliest center of such a quake. Thursday's Mexican quake probably will accelerate that effort: Its epicenter was predicted by a Caltech scientist and tends to support the use of a unique method of historical geology in determining quakes' locations.

Victoria Le Fevre, a Caltech graduate student in seismology, created controversy with her article in the May issue of the Journal of Geophysical Research speculating on the likelihood of a major quake along the Mexican coast about 200 miles northwest of Acapulco. Le Fevre used the "gap theory," which argues that the most likely place for seismic activity on a fault is at those points where the fault has been most quiet for the longest period.

Le Fevre said today that she did not predict a quake, because she did not suggest a time or magnitude for seismic activity there. But some scientists objected to her pinpointing the place, arguing that it might escape a major earthquake if the Cocos geologic plate under that area of the Pacific were to slide smoothly rather than violently under the North American continental plate.

Caltech seismologist Kerry Sieh, also using the gap theory, has become an earthquake archeologist, digging into portions of the San Andreas fault for evidence of past quakes that might suggest historical patterns. Sieh last year reported that excavations near Palmdale, on the San Andreas fault north of here, show quakes recurring every 145 years on the average, the last occuring 128 years ago.

Sieh calculated a 50 to 90 percent chance of a severe earthquake, between magnitudes 7.5 and 8.5, in the next 50 years along a 200-mile stretch of the fault from north of Los Angeles to the Salton Sea in the Imperial Valley. In anticipation, researchers have littered the landscape at Parkfield, a point on the San Andreas fault identified as a gap, with seismic instruments.

Scientists and earthquake planners here said Los Angeles was far ahead of Mexico in improving building construction -- in Heaton's view the principal factor in limiting earthquake casualties. A 1976 quake in Tangshan, China, the most serious in modern times, killed 242,000 people in a city where most of the buildings were unreinforced brick and masonry. Los Angeles has 8,000 such structures, erected before new building codes were passed. A new city ordinance requires owners to install special braces on such buildings or tear them down.

Paul Flores, director of the government-funded southern California Earthquake Preparedness Project, said he was familiar with his Mexican counterparts' work and felt they had been stymied by political pressures. "Mexico City has been working very hard to upgrade its building codes," he said, "but the codes are one thing and having the codes applied and the whole construction process is something else."