A news screen at a railway station in Seoul shows a map of the epicenter of the 3.4-magnitude quake. (Jung Yeon-Je/AFP/Getty Images)

A 3.5-magnitude earthquake was detected in northern North Korea on Saturday afternoon, near the nation's known nuclear test site, according to the U.S. Geological Survey. 

But experts said seismic data suggested that it was probably not caused by a new nuclear test, although it could have been a delayed geological reaction to the last test nearly three weeks ago. There were no reports of radiation around the site. 

"This event occurred in the area of the previous North Korean Nuclear tests," the USGS said on its website. "We cannot conclusively confirm at this time the nature (natural or human-made) of the event."

The earthquake-monitoring agency in neighboring China initially said it suspected that the North Korean quake, which occurred at 3:59 p.m. local time, was caused by an explosion, though the magnitude was significantly lower than a previous nuclear test this month. The USGS estimated the depth of Saturday's quake to be five kilometers (three miles).

But South Korea's meteorological agency said the incident appeared to be a natural quake.

An official from the agency, speaking to the Associated Press on the condition of anonymity citing office rules, said an analysis of seismic waves and the lack of sound waves clearly showed that the quake wasn't caused by an artificial explosion.

Nuclear proliferation watchdog CTBTO said the quake was unlikely to have been man-made. 

North Korea's last nuclear test — its sixth to date — was detected as a 6.3-magnitude earthquake Sept. 3 and was followed by a 4.1-magnitude quake that experts said could have been a tunnel collapsing after the nuclear explosion.

CTBTO Executive Secretary Lassina Zerbo tweeted seismographic graphs from Saturday's quake alongside those from the second quake on Sept. 3, with both data patterns looking broadly similar.

"Indication event natural and same epicentral distance," he tweeted.

He also said there had been a smaller quake nearly four hours earlier Saturday. 

"Two #Seismic Events! 0829UTC & much smaller @ 0443UTC unlikely Man-made! Similar to "collapse" event 8.5 mins after DPRK6!," he tweeted.

Daryl G. Kimball, executive director of the Arms Control Association, said he had spoken to Zerbo.

"While the data analysis is preliminary, the early indications strongly suggest that the smaller events today (which are in a seismically inactive area) are likely further geologic disturbances created by the Sept. 3 nuclear test explosion," he said. 

"In other words, not another nuclear test but related to the high-yield Sept. 3 nuclear weapon test explosion."

In Tokyo, a Japanese government official told reporters: "We are gathering information but we haven't heard of any emergency meeting among senior government officials," Kyodo reported.

A meteorological agency official in Russia's Far East also told Kyodo that there were no abnormal radiation levels after the incident