Around noon on 8 October 2026, a fire started in the Nömrög Strictly Protected Area in Khalkhgol sum, Dornod Province, Mongolia, about 27 km from the China–Mongolia border. The vehicle that caught fire belonged to Mobicom, a Mongolian mobile operator and a consolidated subsidiary 98.8% owned by KDDI1. Mobicom acknowledged that the vehicle fire spread to the grassland and said the group was there to hand over communication works for border posts and the border port and to survey local conditions2. Mongolian media reported that the convoy entered the reserve to fish for taimen; the purpose of the entry is still under police investigation3. The fire was fully extinguished on the morning of 16 October4. On 20 October the Mongolian Ministry of Environment and Climate Change put the preliminary burned area at 528,765.5 ha, of which 468,965.7 ha lies inside the reserve, 79.8% of the reserve5.

I analysed this fire with public satellite data, first published on 12 October and updated for more than a week afterwards: Nömrög (Numrug) Grassland Fire in Dornod, Mongolia: Satellite Monitoring and Weather. “My post” below refers to it.

Note

Updates:

  • 10-22: added the TV Asahi and Fuji TV reports from 10-21 onward and KDDI’s written response.

Timeline

Beijing time (UTC+8) unless noted; posts on X are in JST.

TimeEventSource
10-06Fibre opening ceremony at the Sümber border port; the border authority issues border-zone permits for 10-06 to 10-08 to 13 people and 5 vehiclesMongolian General Authority for Border Protection statement6
10-08 11:41–11:51Himawari-9 and GK-2A, two geostationary satellites, first detect warming in the fire pixelMy post
10-08 11:55Mongolia’s National Emergency Management Agency receives the reportNEMA7
10-08 12:15A fire lookout tower of the Bailang Forestry Bureau, Hinggan League, spots a fire inside Mongolia about 40 km from the borderHinggan Forestry and Grassland8
From 10-09Hinggan League posts 530 people along 131 km of border; around 13:00 fire appears on the border line of East Ujimqin BannerHinggan Forestry and Grassland, Xinhua9
10-11More than 2000 people deployed in Inner Mongolia; Hinggan League starts burning firebreaks along the border from 12:40Xinhua, Hinggan Forestry and Grassland
10-12 00:55Mongolian side announces the fire is containedMNB10
10-12Mobicom issues a statementikon.mn2
10-12My post is publishedThis blog
10-12ParsToday Japanese and People’s Daily Online Japanese post in Japanese on X, the earliest Japanese-language contentX
10-12 19:00China’s National Meteorological Centre and three other agencies issue the Global Hazardous Weather Monitoring and Forecast Bulletin: FY-4C estimate of about 5050 km² burned, with a three-day forecast of wind, humidity and fire dangerNMC11
10-13More than 3400 people deployed on the Chinese side; the fire has not crossed the borderXinhua12
10-14People’s Daily Japanese posts a video on X; two Mongolian users post in Japanese, one of them mentioning the Japanese embassy in Mongolia, naming Mobicom in Japanese for the first timeX
10-14 20:58NTV airs an interview with Mobicom CEO Yasuhiro Ochi, who apologises to the Mongolian peopleNTV МЭДЭЭ13
10-15SCMP reports that the vehicle may have belonged to MobicomSCMP
10-16 07:00Mongolian side announces the fire fully extinguishedNEMA4
10-17 11:30 JSTSong Wenzhou posts in Japanese on X, 110,000 views; Kabashima Mariko relays the SCMP report, 12,000 views. First spread among accounts based in JapanX1415
10-18 to 10-19Posts by ordinary Japanese users on X increase, several asking why mainstream media are not reportingX16
10-19Xinhua Japanese publishes an article; livedoor, d menu and au Web Portal republish it from the same dayXinhua Japanese17
10-19I check NHK, Kyodo, Jiji, Nikkei, Asahi, Yomiuri, Mainichi and Sankei: no reports; nothing in KDDI’s newsroomMy own check
10-20 17:12Mongolian Ministry of Environment and Climate Change publishes the preliminary burned areaMontsame5
10-20 18:40 JSTlivedoor posts the Xinhua Japanese article on X, 17,879 repostsX18
10-20 22:07 JSTKyodo publishes the first report by Japanese mainstream media, 12 days 9 hours after the fire started; KDDI responds for the first time in that coverage, confirming that a Japanese executive was in the reserve at the time; Sankei republishes at 22:20Kyodo19
10-21Alterna, NEWS Post Seven and others follow120
10-21 16:12 JSTTV Asahi (ANN) runs a report headlined 「KDDIが謝罪　中国メディア『日本人が乗った車が原因』」; Fuji TV (FNN) follows at 16:41. These are the earliest reports by a Japanese TV network I could findANN21, FNN22
10-21KDDI answers a written inquiry from Alterna and says it received Mobicom’s report on 11 OctoberAlterna23
10-22 00:24 JSTSecond ANN report, with remarks from a Mongolian reporter and Mongolia’s deputy prime minister; ANN’s YouTube title marks it as that night’s Hōdō StationANN24
10-22 08:30 JSTStill no reports found on Google News from NHK, Nikkei, Asahi, Yomiuri, Mainichi, Jiji, TBS or Nippon TVMy own check

The delay in Japanese mainstream media

This post makes one point: from the fire on 8 October to Kyodo’s first brief late on 20 October, Japanese mainstream media were silent for 12 days and 9 hours25.

Expecting coverage on the day of the fire would be unreasonable, but there should have been coverage three or four days after it started at the latest. Material was available throughout these 12 days. The Chinese side began deploying on the 9th and Xinhua had a story on the 11th. The Mongolian side announced containment on the 12th, and Mobicom issued a statement the same day acknowledging that its vehicle fire spread to the grassland. On the 14th Mobicom’s Japanese CEO apologised on Mongolian television. On the 15th SCMP named Mobicom and KDDI. On the 19th Xinhua even published a Japanese-language article, and KDDI’s own au Web Portal republished it the same day. As of the 19th, NHK, Kyodo, Jiji, Nikkei, Asahi, Yomiuri, Mainichi and Sankei had published nothing. In Kyodo’s brief on the 20th, the account of how the fire started also cites Xinhua.I also found no reports from Western wire services and major papers such as Reuters, AP, AFP or the BBC when searching Google and Bing as of the 21st. This post discusses only Japanese media because the vehicle belonged to a subsidiary of a listed Japanese company, the CEO who apologised is Japanese, and demand from readers in Japan was visible on X from the 17th.

I know Japan has the kisha club system. Dependence on primary sources explains part of this: KDDI’s newsroom had published nothing by the 20th, the police had not released findings, and reporters had no official Japanese statement to quote. That explains the absence of investigative reporting, but it does not explain the absence of even a relayed report. The Japanese CEO of a subsidiary 98.8% owned by a listed Japanese company apologising publicly on foreign television is news in itself, and it does not require waiting for a KDDI announcement.

From the 21st the coverage moved into follow-ups. That afternoon TV Asahi and Fuji TV ran reports, both headlined on 「KDDIが謝罪」 (KDDI apologises); the account of how the fire started was still taken from Xinhua, and ANN’s headline says outright 「中国メディア『日本人が乗った車が原因』」 (Chinese media: a car carrying Japanese caused it). KDDI’s apology was not issued as an announcement. It was a written response to media inquiries, and the version Alterna printed includes the line 「なお、モビコムからは10月11日に報告を受けています。」 (We received a report from Mobicom on 11 October). KDDI therefore knew by the third day after the fire started, and nine days passed before its first public response on the 20th. ANN’s second report, just after midnight on the 22nd, is the first piece of Japanese mainstream coverage to include interviews from the Mongolian side: a Mongolian reporter at the reserve, Suren Delger, says 「国民の怒りは頂点に達しています。」 (public anger has reached its peak), and Deputy Prime Minister Nomtoibayar says 「謝っただけでは許されない」 (an apology alone is not enough)24. As of 08:30 JST on the 22nd, Google News still shows no reports from NHK, Nikkei, Asahi, Yomiuri, Mainichi, Jiji, TBS or Nippon TV.

Consequences of the delay

I focus on mainstream media rather than individuals because the reach differs. Mainstream media have television, national newspapers, wire services feeding regional papers, and distribution through portals such as Yahoo! News. Once Kyodo’s brief went out on the 20th, Sankei republished it 13 minutes later, and regional papers such as the Shimotsuke Shimbun and the Oita Godo Shimbun carried it the next day. The reach of one wire brief and that of posts on X differ by orders of magnitude.

What the Japanese public actually saw during these 12 days shows up when posts on X are counted by day. The table below counts posts about this fire filtered with lang:ja, by JST day, reposts included26.

Date (JST)PostsOf which reposts
10-06 to 10-1140
10-12149
10-1333
10-141812
10-1542
10-1621
10-17424403
10-182,4392,333
10-197,4237,152
10-2042,59541,410
10-21 (through the 21:00 hour)72,68170,653

Hourly Japanese-language posts on X about this fire, 2026-10-06 to the 21:00 hour of 10-21 (JST), linear scale above and log scale below; the numbers mark key events in the timeline

Hourly Japanese-language posts on X about this fire, 2026-10-06 to the 21:00 hour of 10-21 (JST), linear scale above and log scale below; the numbers mark key events in the timeline

Hourly Japanese-language posts on X about this fire, 2026-10-06 to the 21:00 hour of 10-21 (JST), reposts included. Linear scale above, log scale below. 1 fire starts; 2 NMC bulletin, my post published; 3 Mobicom CEO apologises; 4 posts by Song Wenzhou and Kabashima Mariko; 5 Xinhua Japanese article; 6 livedoor posts on X; 7 Kyodo brief. Counts: Grok; query in the footnote

  • 10-06 to 10-11: 4 posts, all unrelated false matches. In the first four days after the fire started, the 8th to the 11th, there were zero relevant Japanese-language posts.
  • 10-12 to 10-16: 41 posts in total. Of the 13 that are not reposts, about 5 are from accounts based in Japan, all individuals relaying Chinese news, with few views; 3 are from Japanese-language accounts of foreign media. The first to name Mobicom in Japanese were two Mongolian users on the 14th. Within these 13, the posts from the 15th and 16th were not checked one by one: the counts for those two days come from the counts endpoint, and only the first page of search results was read for original posts.
  • 10-17: 424 posts. One Japanese post by Song Wenzhou reached 110,000 views, the first spread among accounts based in Japan.
  • 10-18: 2439 posts, five to six hundred per hour in the evening, mostly reposts of one personal post about Mobicom executives fishing27. Ordinary users began asking the same question: why are Japanese media not reporting this.
  • 10-19: 7423 posts, holding at 500 to 660 per hour from the evening.
  • 10-20: 42595 posts. Rising from 15:00: 3225 at 18:00, 6137 at 19:00, a peak of 7050 at 20:00, then 6313 at 21:00, 6124 at 22:00 and 5362 at 23:00. The most reposted was livedoor’s post at 18:40, carrying Xinhua’s Japanese wire story from the 19th, which accounts for 27.5% of all reposts on the 19th and 20th. Kyodo’s brief went out at 22:07, 3 hours 27 minutes after the livedoor post; by then the hourly post count on X had been falling from its peak for two hours. After it went out, one user wrote 「やっと日本の通信社で日本発の記事」 (“finally, an article from a Japanese wire service, originating in Japan”); another quoted the livedoor post, complained that it took 12 days after the fire for Japanese media to report it, and noted that it was a republished Xinhua story28.
  • 10-21: 72,681 posts through the 21:00 hour, more than the 52,926 from the 6th to the 20th combined. 3709 at 00:00, falling each hour to 1093 at 04:00, the day’s low; rising from 05:00, with 4340 and 4893 at 07:00 and 08:00. From 09:00, apart from the day’s high of 5167 at 12:00, each hour had 3146 to 4243 posts. Which posts were reposted in each hour was not counted.

People did want to know. The interest was visible on X from the 17th and was in the thousands by the 19th, and mainstream media came more than a day later still. By the time Kyodo published, the hourly post count on X had passed its peak, and the next day’s total exceeded the previous 15 days combined. For an incident caused by a subsidiary of a listed Japanese company, the Japanese public’s primary sources were Chinese and Mongolian reports, translated and republished on X by individuals and portal sites. Mainstream media arrived last.

I have started to doubt whether journalistic professionalism and investigative reporting as a trade still exist in practice in Japan. I am somewhat disappointed.

Individuals

Apart from the media, I also checked whether anyone had done their own analysis.

On Google, restricted to 8 to 19 October, I searched the Japanese keywords モンゴル, 草原火災, ドルノド, ヌムルグ, KDDI, モビコム, 衛星画像 and ひまわり; on X I searched period by period with lang:ja and date filters. Under 衛星画像 (satellite imagery) and ひまわり (Himawari), the only result was the Japanese version of my own post. Every other post was a relay: of foreign media before the 17th, and of SCMP, Xinhua and Kyodo after it, plus some screenshots of US remote sensing products. I also had Claude, ChatGPT, Gemini, DeepSeek and Grok each search; the results were the same.

Japan has no firewall, its networks are fast, and access to US open datasets is very quick. My post used only free public data and public methods; anyone interested in this field could have produced it in the same 12 days. If my post is worth 80 points, I would have hoped for at least a 60- or 70-point piece from within Japan. Still, whether individuals do this affects thousands or tens of thousands of people; whether mainstream media do it affects society as a whole. Why no individual analysis appeared, I do not know.

China

China has its own problems with transparency; that needs no elaboration. For this event, though, a few facts are comparable:

  • About 30 minutes after the fire started, a fire lookout tower in Hinggan League had spotted it; deployment along the border began the next day.
  • Xinhua published on the 11th, CCTV and People’s Daily followed from the 13th, and Southern Weekly had an on-site report on the 18th.
  • On the 12th, the fourth day of the fire, the National Meteorological Centre publicly released FY-4C fire monitoring and a three-day weather and fire danger forecast, signed by four agencies.
  • At least one organisation in China achieved similar results with similar techniques; I will not go into details.

On the Chinese side there was a wire story on day 3 and a published official monitoring product on day 4; on the Japanese side, the first report, relaying other outlets, came on day 12.

Where my yardstick comes from

I grew up reading the work of Western journalists and investigative reporters, as well as Chinese investigative reporters and Qdaily. They could state clearly when, where and what happened, and they laid out causes and consequences as a chain as far as they could.

I have also subscribed to The Wall Street Journal, The Atlantic, Foreign Affairs and The Economist. These shaped my standard for judging whether an analysis holds up. That is the yardstick, and this post measures Japanese mainstream media against it.

That is all.

  1. Alterna, モンゴルの自然保護区で大規模火災、出火原因は日系企業 (Japanese), 2026-10-21. Ownership figure from KDDI’s annual securities report. ↩︎ ↩︎

  2. ikon.mn, full text of Mobicom’s statement (Mongolian), 2026-10-12. ↩︎ ↩︎

  3. nutag.mn, Dornod police working group begins investigating the cause of the fire (Mongolian), 2026-10-17. ↩︎

  4. Mongolia National Emergency Management Agency, fully extinguished at 10-16 07:00 (Mongolian). ↩︎ ↩︎

  5. Montsame, Massive Wildfire Damage in Numrug Reserve to Be Assessed, 2026-10-20. ↩︎ ↩︎

  6. Eguur.mn, full text of the General Authority for Border Protection statement (Mongolian), 2026-10-14. ↩︎

  7. Mongolia National Emergency Management Agency, daily report for 10-12 (Mongolian). ↩︎

  8. Hinggan Forestry and Grassland, 鏖战七天七夜 坚守北疆防线 (Chinese), 2026-10-16. ↩︎

  9. Xinhua, 蒙古国草原火蔓延至我边境 2000余人正在堵截 (Chinese), 2026-10-11. ↩︎

  10. MNB, grassland fire in Khalkhgol sum contained (Mongolian), 2026-10-12. ↩︎

  11. National Meteorological Centre, 全球灾害性天气监测预报快讯 (Chinese), 2026-10-12 19:00, 《中蒙边境火情实况及未来三天天气及风险预报》. The link is a section page; select the 2026/10/13 03:00 issue in the list on the right. ↩︎

  12. Xinhua, 3400 余人布防堵截 (Chinese), 2026-10-13. ↩︎

  13. NTV МЭДЭЭ, interview with the Mobicom CEO, 2026-10-14. ↩︎

  14. Song Wenzhou (@sohbunshu), 迷惑日本人、モンゴル草原に山火事を起こす, 2026-10-17. View and engagement counts as checked on 10-21; the same applies below. ↩︎

  15. Kabashima Mariko (@KNHjyohokyoku), モンゴル大規模草原火災、日本企業の車両が出火原因か, 2026-10-17 11:30 JST. ↩︎

  16. 藍白もなか (@aishiro_monaka), post on 2026-10-19, 16,000 views. Link to be added. ↩︎

  17. Xinhua Japanese, モンゴル草原火災、日系企業の車両が原因 自然保護区へ進入, 2026-10-19; republished the same day by livedoor, d menu and au Web Portal. Link to be replaced with the specific article. ↩︎

  18. livedoor News (@livedoornews), 2026-10-20 18:40 JST, 17,879 reposts, 2,561 quotes, 46,365 likes; links to livedoor’s republication of Xinhua’s 10-19 Ulaanbaatar wire story. Link to be added. ↩︎

  19. Kyodo, モンゴルで大規模草原火災　日系企業の車が原因か, 2026-10-20 22:07 JST; republished by Sankei at 22:20. ↩︎

  20. NEWS Post Seven, 「東京都の2倍超」が焼失した大規模火災はなぜ起きたのかKDDIに取材, 2026-10-21. ↩︎

  21. TV Asahi (ANN), モンゴル草原で大規模火災　KDDIが謝罪　中国メディア「日本人が乗った車が原因」, 2026-10-21 16:12 JST. ↩︎

  22. FNN Prime Online, モンゴル草原で大規模火災 出火元はKDDI子会社の車両か、KDDIが謝罪 東京都の2.4倍焼失、中国にも延焼迫る, 2026-10-21 16:41 JST. ↩︎

  23. Alterna, ［続報］「モンゴル草原火災を極めて重く受け止める」と親会社のKDDIがコメント, 2026-10-21. ↩︎

  24. TV Asahi (ANN), 「国民の怒りは頂点に」日系企業の車から出火か…モンゴル自然保護区で大規模火災, 2026-10-22 00:24 JST. ↩︎ ↩︎

  25. The English-language daily The Japan Times published KDDI subsidiary faces criticism and questions over Mongolian wildfire by staff writer Tomoko Otake at 12:51 JST on 10-20, about 9 hours before Kyodo. It is an English-language outlet read mainly by foreign residents of Japan and readers abroad; the counts and discussion in this post cover only Japanese-language media aimed at readers in Japan. ↩︎

  26. Counts by Grok, 2026-10-21. The initial query was モンゴル 火災 lang:ja, 33,276 posts. To check recall it was expanded to combinations of place names and fire terms, joined with a company-name branch, with job-hunting noise excluded. The query actually used is (((モンゴル OR ドルノド OR ヌムルグ OR ハルハゴル OR ノムルグ OR ノムログ OR ハルフゴル OR ハルハ川) (火災 OR 草原火災 OR 山火事 OR 野火 OR 延焼 OR 焼失 OR 火事)) OR ((モビコム OR Mobicom OR KDDI) (モンゴル OR 保護区 OR 出火 OR 火災 OR 草原) -就活 -インターン -28卒 -選考 -面接 -火災保険 -CTEE -サムスン)) lang:ja, giving 52,926 posts for 10-06 to 10-20, of which 51,325 are reposts; of a sample of 90 from the added posts, 86 were on topic. The table uses the expanded numbers. 10-21 was added with the same query through the 21:00 hour, for a cumulative 125,607 posts since 10-06, of which 121,978 are reposts; the most recent hours may still change slightly as posts are deleted. Reposts were counted separately with is:retweet; JST days are split at 15:00 UTC. 10-06 to 10-14 were counted post by post from search; 10-15 to 10-21 use X’s official hourly counts endpoint, not checked post by post, and that endpoint only covers the last 7 days. The ranking of reposted accounts was built by picking candidates from a sample of 300 and counting each; an account outside the sample ranking higher cannot be ruled out. Hourly data: mongolia_fire_D_hourly_jst.csv. ↩︎

  27. @kaminoyume27, post on 2026-10-18, 6,672 reposts. It uses 出火 and 草原 but not 火災, which is the direct reason the query was expanded. Link to be added. ↩︎

  28. @nichinichibijou, 2026-10-20 19:49 JST, quoting the livedoor post, 2,292 reposts. Link to be added. ↩︎