A familiar scene has played out around the country, much to the chagrin of the Japan Meteorological Agency.

After an early earthquake warning is issued, startled people brace themselves for the “big one.” They feel a slight wobble, or even nothing at all, and then go about their business as usual.

The agency has improved the technology for issuing accurate and timely warnings for quakes, but the limitations of the system are still clear.

Since April, the agency has issued several early earthquake warnings for regions where the intensity of the quakes was projected at lower 5 on the Japanese scale of 7. The shaking of a lower 5 earthquake makes it difficult to walk.

But almost all the warnings issued since April have proved wrong.

For example, an alert was sent at 1:13 p.m. on May 19 during a news program aired by public broadcaster Japan Broadcasting Corp. (NHK).

“An earthquake emergency warning has just been issued,” an announcement said during the show. “Please remain alert against strong shaking.”

However, the quake’s maximum seismic intensity, which was recorded in Takayama, Gifu Prefecture, was 4.

Aside from the quake in question, six warnings were issued between April 20 and May 19 for tremors originating in areas along the boundary of Gifu and Nagano prefectures in central Japan and in the Kanto region.

But none of them was strong enough to hit the lower 5 threshold.

The most embarrassing ones were issued on May 11 and May 19.

The former was for a quake off Ibaraki Prefecture that was later estimated at an intensity of just 3.

The May 19 quake in Gifu Prefecture was predicted to be up to upper 5, but, like the Ibaraki tremor, the actual intensity was 2 levels lower.

Such discrepancies appear inevitable under the current system.

When an earthquake occurs, small seismic waves first spread from the focus, followed by stronger ones.

The initial waves detected at around 170 monitoring spots are immediately analyzed to gauge the scope of the quake. An early warning advisory is issued if it is estimated at lower 5.

One big problem for the emergency alert system, which has been available for the public since 2007, is that the data collected in the initial stages tends to be rather inaccurate.

As time passes by, additional data becomes available that could be used for more reliable early warnings.

But if the agency waits to collect that information, the “early” warning could be issued after the targeted regions are already rocking from the stronger seismic waves.

When the agency reviews the accuracy of its warnings, a discrepancy in intensity between the projection and the actual figure within 1 is considered a “margin of error.”

If the difference is 2 or more, it is called a “swing out.”

The alerts given in May were categorized as swing outs. And they raised concerns that the agency could be viewed as “the boy who cried wolf” if the trend continues.

At a news conference in May, Yasuo Sekita, director-general of the agency, called for understanding from the public concerning the technological challenges involved in instantly projecting the intensity of an earthquake.

“It is basically inevitable,” he said about missing the mark.

The agency has incorporated technological advances in the early warning system. The latest technology introduced in March 2018 significantly improved its track record.

Of 193 emergency alerts issued by then, 55 quakes, or 28 percent, registered an intensity of 3, a level felt by most people, or lower.

After the introduction of the latest technology, only two alerts, or 6 percent of the 30 in total, proved to be under that level.

Despite the early warning system’s checkered record, citizens do not appear annoyed by the false alarms.

“The share of people who prefer a warning issued promptly--even if it proves wrong--has remained at around 60-70 percent until recently,” said Hiromichi Nakamori, professor of disaster information studies at Nihon University.

He has periodically conducted surveys on public attitudes toward the early warning system.

The agency once issued a “wrong” alert for a broad area of the Kanto region in January 2018, but in Tokyo, even a level 1 quake was not felt.

An online survey held on the heels of that blunder found that 65.5 percent of respondents said the mistake “could not be helped” or was “not particularly a cause for concern.”

The survey also found that a combined 85.5 percent said the agency should promptly provide an explanation to the public through TV, radio and email for the mistaken assessment.

“Ending in a failure cannot be helped,” Nakamori said. “But the agency should make sure that it always briefs the public on the mechanism of the emergency alert system to gain their understanding.”