Planetary-scale answers, unlocked.
A Hands-On Guide for Working with Large-Scale Spatial Data. Learn more.
At 22:33 on October 22, 2026, a magnitude 7.0 earthquake struck Baqên County in Nagqu, northern Tibet. At least 16 people died, China’s state news agency Xinhua reported. CNN, CBS News and Al Jazeera carried the report. Early reports, such as NBC News, said 32 people were injured. China’s Ministry of Emergency Management later gave 54, Outlook India reported, and the Press Trust of India reported 76.
At least 339 homes and 20 livestock sheds collapsed across Nagqu, its media office said, Outlook India reported. The US Geological Survey (USGS) says most homes in the area are built of adobe or unreinforced brick laid in mud. USGS says such homes collapse easily in strong shaking.
Xinhua photos carried by the Associated Press show families whose homes were unsafe sleeping in tents, some at a primary school. Officials sent 3,239 tents and 1,000 prefabricated homes, ECNS reported.
What we found: in the October 22 quake’s strong shaking, open maps show no building within 1 km of 40 of the 42 settlements. That counts mapped buildings; homes may stand there unmapped. Strong shaking here means intensity VII or more, as modeled by USGS. The open map we used is Overture Maps, which draws on OpenStreetMap and other open sources.
A damage estimate built from Overture Maps would count no buildings within 1 km of those 40 settlements. No coverage we found has reported this gap. Finding it takes matching every mapped settlement and building in the area to its modeled USGS shaking level.
Key takeaways
USGS estimates that 6,348 people lived where the shaking reached intensity VII or more. Of them, 2,652 lived where it reached intensity VIII or more. USGS calls intensity VII “very strong” shaking, with considerable damage to poorly built buildings.
The estimate comes from a global population model. It does not count herders who move with their animals. USGS’s headline estimate, about 110,000 people, counts weaker shaking too, from intensity V up.
In January 2025, a quake of similar size at Tingri, also in Tibet, killed 126 people, PBS NewsHour and Al Jazeera reported.
The idea we tested: the October 22 quake killed fewer people than the Tingri quake because far fewer people and buildings lay in its strong shaking, not because its shaking was weaker. The idea comes from coverage that calls the area remote, as in AccuWeather’s headline. It also comes from the USGS Prompt Assessment of Global Earthquakes for Response (PAGER) model. PAGER estimates deaths from the people at each shaking level and how weak their homes are.
Before looking at the results, we wrote down what would count against the idea. One: the October 22 strong shaking held half or more as many people, buildings or mapped places (schools, clinics, shops) as Tingri’s. Two: its area was less than half of Tingri’s.
USGS estimates 6,348 people in the strong shaking on October 22, and 37,641 at Tingri. 6,348 is 17% of 37,641. Verdict: supports. Limit: both are model estimates, and the population model does not count herders.
Overture Maps shows 70 buildings in the strong shaking on October 22, and 3,782 at Tingri. But the two areas are not mapped alike. At Tingri, 75 of 142 settlements had no building mapped within 1 km. On October 22, 40 of 42 had none. Verdict: can’t decide.
Overture Maps lists 2 places, such as schools, clinics and shops, in the strong shaking on October 22, and 13 at Tingri. 2 is 15% of 13. Verdict: supports, but 2 and 13 are too few to rely on.
The strong shaking covered 4,441 km² on October 22 and 4,862 km² at Tingri. 4,441 is 91% of 4,862, not less than half, so by the rule we set in advance this supports the idea.
Intensity VIII or more covered 940 km² on October 22, 48% of Tingri’s 1,946 km². Only Tingri reached intensity IX, over 455 km². By these models, the most violent shaking covered a smaller area this time. We did not set this as a test in advance, but it counts against the idea that the shaking was not weaker.
The idea holds in part. The October 22 strong shaking covered 4,441 km², 91% of Tingri’s 4,862 km², but held 6,348 people, 17% of Tingri’s 37,641. Its intensity VIII area was 940 km², 48% of Tingri’s 1,946 km². So weaker peak shaking may also explain part of the lower toll. These tests cannot say how much each cause counts.
Counts of collapsed homes by village, from rescue teams or satellite images, would settle it; so far only a total of 339 for Nagqu is reported. For the families in Baqên County’s villages, the toll may still rise as rescuers reach them. Today, an estimate built from open building maps would count no buildings within 1 km of 40 of the 42 settlements in the strong shaking.
One Wherobots job loaded each USGS ShakeMap grid as a table. The grid holds the modeled shaking in cells about 1 km across. The job gave every Overture building, place and settlement the shaking level of its cell, with a spatial join in structured query language (SQL). It then added up the area at each shaking level and read the USGS population estimates. The final run took about 3 minutes.
We followed the USGS PAGER method (Jaiswal and Wald, 2010), which counts the people at each shaking level. Like PAGER, we round the modeled intensity to the nearest whole level, so any modeled value of 6.5 or more counts as intensity VII. An earlier run started intensity VII at 7.0. It gave the same verdicts, and we switched to PAGER’s rule before writing.
The shaking maps and the population are modeled. USGS updates ShakeMaps as data arrives; we used version 3 for October 22. Settlements are Overture localities, which include villages and township and county seats. “No building within 1 km” means none mapped; buildings may stand there unmapped. Deaths are as reported by Chinese state media, and the toll may rise.
A magnitude 7.0 earthquake struck Baqên County in Nagqu, northern Tibet, at 22:33 local time on October 22, 2026. At least 16 people died, China’s state news agency Xinhua reported. Injury counts rose from 32 to 76 as reports came in.
Almost none near the settlements. Strong shaking, intensity VII or more as modeled by the US Geological Survey (USGS), reached 42 settlements: villages and township and county seats. Overture Maps, an open map that draws on OpenStreetMap and other open sources, shows no building within 1 km of 40 of them. In the strong shaking of the 2025 Tingri quake, it shows no building within 1 km of 75 of 142 settlements.
Intensity measures how strongly the ground shook at a place, on the Modified Mercalli scale from I to XII. USGS calls intensity VII “very strong” shaking: damage is slight in well-built buildings but considerable in poorly built ones. USGS says most homes in this area are adobe or unreinforced brick laid in mud. In this article, strong shaking means intensity VII or more.
Partly because far fewer people lived in the strong shaking: 6,348 against 37,641, USGS estimates. The strong shaking covered 4,441 km², 91% of the 4,862 km² at Tingri. But intensity VIII or more covered 940 km², about half of the 1,946 km² it covered at Tingri. That smaller area of violent shaking may also have lowered the toll.
One Wherobots job loaded the USGS shaking maps for both earthquakes. It matched each building, mapped place (such as a school, clinic or shop) and settlement in Overture Maps to its shaking level. It then read the USGS population estimates. Three of the four tests supported the idea that fewer people lay in the strong shaking; the building test could not decide. The smaller area of the most violent shaking counts against the idea that the shaking was not weaker.
Super Typhoon Champi: Guam Recovers, Okinawa and Taiwan Hit
98% of Okinawa's power cuts on Oct 23 were in Ishigaki and Taketomi, where a forecast put 96%+ of buildings in 120 km/h winds. Guam's grid is 99% back.
share this article
Awesome that you’d like to share our articles. Where would you like to share it to:
