---
title: Who Decides the First Snow?
url: https://oosioo.com/en/p/%EC%B2%AB%EB%88%88%EC%9D%80-%EB%88%84%EA%B0%80-%EC%A0%95%ED%95%98%EB%82%98
date: 2026-11-04T12:54:00+00:00
author: SYSOP
summary: Seoul's first snow is not decided by whether it snows somewhere in Seoul. It is recorded only when an observer at the Seoul Weather Observatory in Songwol-dong, Jongno-gu, sees snow falling with their own eyes. Sleet on Bukhansan or snow pellets in Gangnam don't count unless they reach Songwol-dong. Matching press reports of KMA announcements for twelve winters, 2014 to 2025, the earliest and latest first snows were 30 days apart. After Kyoto switched to machine judging in 2020, its normal first snow moved six days earlier, and in the United States snow under 0.1 inch is only a "trace". Under the same sky, the first snow date depends on who measures it and with what. An interactive lets you judge "Is this first snow?" yourself. This year's first snow was seen on Seoraksan on 29 October, but as of 4 November Seoul's first snow has not come. The twenty-ninth piece in A Page for Today.
---
# Who Decides the First Snow?

![On a pale grey background, the left side reads "A Page for Today · 29" and the title "Who Decides the First Snow?", with the line "Seoul's first snow only counts when an observer on a hill in Songwol-dong sees it fall with their own eyes" beneath it. Lower left, a slip of lined paper reads "Seoul Weather Observatory · Log / Weather: snow  Method: by eye / Depth: 0 cm — still first snow". On the right, snowflakes drift beyond a white window frame and a thin line of snow sits on the rooftops, with a black label under the window showing an eye icon and the words "Human eyes"](/uploads/76942baefe5fcd41.webp =720)

This time, the word for *A Page for Today* is "first snow".

This year's first news of snow came on 29 October. At around 11:40 that morning, snow began to fall around Daecheongbong and Socheongbong, the high peaks of Seoraksan. Nothing settled, and it came nine days later than last year. The same morning, Seoul recorded its first frost of the autumn.

In Korea the first snow is more than a weather report. People promise to meet "on the day of the first snow". Couples wait for it; songs and TV dramas are built around it. So the announcement travels fast, and people check the window.

But the announcement came from the Seoraksan National Park office, not from the Korea Meteorological Administration (KMA). Snow on a mountain does not make "first snow in Seoul". As I write on 4 November, there is still no word of Seoul's first snow; in a normal year it comes around 20 November. The same words, "first snow", are used, but the first snow people wait for in Seoul is a separate thing. So whose snow does Seoul's first snow have to be, and where must it be seen?

## Seoul's first snow falls in Songwol-dong

The answer is in a smaller place than you might think. Seoul's official first snow is decided at the Seoul Weather Observatory, on a hill behind Gyeonghuigung Palace in Songwol-dong, Jongno-gu. It does not matter where in Seoul it snows. Snow has to be observed in this observatory's yard.

The observatory has a long history. According to KMA, it opened on 1 October 1907 as the Seoul branch of the Incheon weather station, at the foot of Madusan in today's Jongno area. It moved to Nagwon-dong in 1913, then to a new building in Songwol-dong in November 1932, and began observing from its current site on 1 January 1933. The main building, the rain-gauge room and the plot of trees planted for seasonal observation became a registered national cultural heritage in 2014. In 2017 the World Meteorological Organization (WMO) named it, together with the Busan observatory, a "centennial observing station": one that has kept records the same way for more than a hundred years.

Seoul's temperature, snow depth and cherry blossom dates are all decided here too. On the morning of 27 November 2024, 16.5 cm of snow lay in Seoul and the news called it "the deepest November snow since records began in 1907". Those 16.5 cm were measured in this observatory's yard.

## It only counts if a person sees it

The way first snow is decided is more surprising still. A person decides it, not a machine. According to a 2021 report in the *Seoul Shinmun*, it only counts as Seoul's first snow if a KMA staff member at the Seoul Weather Observatory sees it falling with their own eyes. KMA said the same about the first snow of 4 December 2025: first snow is announced when an observer confirms that snow is falling. In weather terms this is *mokcheuk* (目測), observation by eye.

That leads to days like this. Around 10 November 2021, sleet fell first on Bukhansan, Gwanaksan and Dobongsan. People in Seoul posted photos and celebrated the first snow. But KMA said Seoul's first snow had not come yet, because nothing had been observed in Songwol-dong. Only after snow was seen there did it become first snow, and Seoul's first snow that year was recorded as 10 November.

Why hold on so tightly to one place? KMA's reason is "climatological and statistical consistency". Seoul is big. The top of Bukhansan, the apartment blocks of Gangnam and the alleys of Jongno have different temperatures. If snow anywhere counted, first snow would come earlier and earlier as more places were watched, not because the climate changed but because there were more eyes. Only by watching one spot in the same way can you set the first snow of 1973 next to that of 2025.

The observers have explained why they still trust human eyes. In a profile of the observatory published by the Seoul city government, they said sensors do detect snow and rain and report them automatically, but it is hard for a machine to tell each change apart, as big flakes turn into snow pellets or hail turns into sleet.

## What counts as snow

So what does an observer have to see to write down "snow"? Several kinds of white things fall from the sky.

KMA's Open MET Data Portal defines a "snow day" as a day on which any of the following is observed: snow, snow showers, snow grains, snowstorm, sleet showers, sleet or snow pellets. It is based on days when an observer saw the phenomenon directly. Each one has a definition.

| Phenomenon | KMA definition |
|---|---|
| Snow | Precipitation of ice crystals shaped like needles, columns or plates |
| Sleet | Rain and snow falling together |
| Snow pellets | Opaque white ice particles (2–5 mm) |
| Snow grains | Very small white ice particles (under 1 mm) |
| Snowstorm | Settled snow blown high into the air at the same time as snow is falling |

Hail is not on the list. It is counted separately as "hail days". Settled snow that is merely blown about by the wind is not on the list either. A snowstorm is only a snowstorm when new snow is also falling.

I should be honest about one thing. I could not find KMA's written rule saying whether first snow uses exactly this list, for example whether a day with only sleet counts. Reports only say "when snow is seen falling". One thing is clear, though: snow does not have to settle. First snow is not "the day it lay" but "the day it was seen falling". That is why a first snow can have a depth of 0 cm.

You can try judging for yourself below. "Twelve first snows" shows Seoul's first snow dates from 2014 to 2025, "Is this first snow?" gives eight situations, and "Rules by country" compares Seoul, Japan and the United States.

```map /embed/first-snow-2026-en.html 4:5
Twelve first snows: a dot chart of Seoul first snow dates from 2014 to 2025 against the normal of 20 November / Is this first snow?: judge eight situations, then check against the Seoul rule / Rules by country: how Seoul, Japan and the United States decide first snow
```

## Twelve first snows

I tried to collect Seoul's first snow dates year by year. KMA's raw records need an access request, so I could not open them this time. Instead I gathered press reports of KMA's announcements. They almost always add "N days earlier or later than normal, and than last year". That let me work back from each year's "than last year" to the previous year's date. From 2014 to 2025, twelve years fitted together this way. Only 2016 rests on a single report.

![A dot chart titled "Seoul's first snow, twelve winters". The horizontal axis runs from 2014 to 2025 and the vertical axis shows dates, with a thick blue line in the middle at the normal of 20 November. Years earlier than normal are blue dots above the line; later years are red dots below. 2014: 14 Nov, 2015: 25 Nov, 2016: 26 Nov, 2017: 17 Nov, 2018: 24 Nov, 2019: 15 Nov, 2020: 10 Dec, 2021: 10 Nov, 2022: 29 Nov, 2023: 17 Nov, 2024: 26 Nov, 2025: 4 Dec. A box on the right reads "Earliest and latest: 10 Nov 2021 ↔ 10 Dec 2020, 30 days apart within these twelve years" and "Records since 1973: earliest 23 Oct 1981, latest 16 Dec 1984, next latest 15 Dec 1975, October first snow only three times: 1980, 1981, 1997"](/uploads/0be5cf2269f4c6b0.webp)

The normal first snow in Seoul (the 1991–2020 average) is 20 November. Yet hardly any of the twelve years came close to 20 November. 2021 came on 10 November, ten days early; the year before, 2020, came on 10 December, twenty days late. The two are exactly a month apart. The 2020 first snow fell on the same date as in 1982 and 1988, the third-latest since 1973. The latest was 1984 (16 December), followed by 1975 (15 December). The earliest was 1981, when snow came on 23 October. Seoul has had an October first snow only three times: 1980, 1981 and 1997.

Of the twelve years, seven were later than normal and five earlier. On average they were about three days late. But twelve years are not enough to call a trend. Some years, like 2018, saw 8.8 cm settle with the first snow (the most since 1981); in 2024, 16.5 cm lay the morning after the first snow. A late first snow does not mean a winter with little snow.

Seoul's first snow has not come yet this year. If it arrives before 10 November, it will be the earliest of these twelve years; if it waits past 10 December, the latest.

Here is a curious detail. One first-snow table circulating online gives 1984 as 18 December and 1975 as 21 December. That table says it was compiled from records of "newly settled snow". Change the rule from "the day it fell" to "the day it lay", and the same years' first snow slips back by two to six days. Change what counts as first snow, and the date changes with it.

## How other countries decide

Deciding first snow by human eye was not only Korea's way. The Japan Meteorological Agency also long relied on observatory staff looking at the sky to call the first snow (*hatsuyuki*). Japan groups August to the following July into one "cold year" and calls the first snow observed in that span the first snow. Sleet counts.

In 2020 that changed. According to a notice from the Kyoto Local Meteorological Office, from 3 February 2020 rain, sleet and snow have been classified automatically by machine. The office explained that automating what used to be visual observation reports allows "objective and continuous observation". According to a 2026 article by the Japanese private forecaster Weathernews, observatories other than Tokyo and Osaka no longer have people looking for snow. They judge it from temperature, humidity and precipitation sensors, using the fact that at the same temperature, the drier the air, the more likely precipitation falls as snow.

The United States takes a different path altogether. The US National Weather Service (NWS) records snow by "measured", not "seen". If 0.1 inch (about 2.5 mm) or more settles, it is "measurable snow"; if less, it is a "trace", marked T on the record. What American media count each year is usually the "first measurable snow". A 0 cm snowfall that would be recorded as first snow in Seoul is only a trace in the United States.

![Under the title "Three ways to decide the first snow", three cards sit side by side. First, "Seoul — one site: Seoul Weather Observatory, Songwol-dong, Human eyes": an observer must see it falling with the naked eye, it does not have to settle, snow on Bukhansan or in Gangnam doesn't count unless it reaches Songwol-dong, KMA's snow-day list includes sleet and snow pellets. Below: "One of 24 sites where KMA still observes by eye (as of Aug 2024)". Second, "Japan's local observatories — Kyoto and others, from 3 Feb 2020, Machines": rain, sleet and snow told apart automatically from temperature, humidity and precipitation sensors, sleet counts as first snow, a year runs August to July, Tokyo and Osaka still look by eye. Below: "Kyoto's normal first snow: 15 Dec → 9 Dec. A machine that never looks away made it six days earlier". Third, "United States — National Weather Service, A ruler": 0.1 inch (about 2.5 mm) or more on the ground is measurable snow, less than that is a trace (T), the media usually count the first measurable snow. Below: "A day that would be first snow in Seoul is recorded here only as a trace"](/uploads/72d9e5c3cb6c13d1.webp)

## When machines watch, first snow comes earlier

The Kyoto notice has one more passage worth noticing. Once judging switched to machines, first snow came earlier. A machine watches the sky without a break. It catches flurries lasting a few minutes that a person might miss. So the Japan Meteorological Agency did not simply stitch old and new records together. Using surviving 10-second and 1-minute data, it went back to 2009 and recalculated when a machine would have called first snow. It measured the gap between person and machine and corrected the old normals. Kyoto's normal first snow moved from 15 December to 9 December, six days earlier.

The weather did not change. Kyoto's sky is the same. The eyes watching it changed, and the first snow date moved by six days. It is the mirror image of why Seoul clings to one spot in Songwol-dong.

This autumn, something similar happened on Mount Fuji. On 30 October the Japan Meteorological Agency announced Fuji's first snowcap of the season after confirming snow lying near the summit, 28 days later than normal. Since observations began in 1894, staff had made this call by looking up at the mountain with their own eyes. This year, for the first time, weather radar and an automatic judging system read the snow data and decided. Snow capping a summit and snow falling in an observatory yard are different things, but the shift in who does the seeing, from people to machines, is the same.

Korea, too, has fewer places where a person looks. KMA's Open MET Data Portal lists "sites observed by eye (as of August 2024)": 24 of them, including Seoul, Incheon, Suwon, Busan, Daegu, Gwangju, Daejeon and Jeju. For periods without visual observation, snow days are shown as "–", with a note: "Snow may have occurred during periods without observation." If nobody is looking, snow can fall and leave no record.

## If AI decided the first snow

One day Seoul's first snow may be called by a machine instead of a person. It is not hard to imagine with today's technology. AI could watch CCTV feeds across the city and catch the moment snowflakes appear on screen. Sensors don't doze off at night.

But that would raise questions harder than "who is looking".

First, whose camera? Use every camera in Seoul and the ones on the slopes of Bukhansan will always see snow first. Seoul's first snow would come earlier every year. Use only one camera in Songwol-dong, and there is less reason to bring in AI at all.

Second, how do you join it to the old records? Like Kyoto, you would have to measure the gap between people and machines and recalculate the normals. Otherwise nobody could tell whether a headline saying "first snow is coming earlier" reflects climate change or a change in method.

Third, what do you teach it to call snow? The line between sleet and rain, or between snow pellets and hail, is blurry even for people. It is exactly the point where observers said machines struggle. AI learns from labels that people attach. Who attached those labels, and by what rule, would end up deciding the first snow date.

Whatever the technology, first snow remains a question of definition. Behind the sentence "the first snow has come" there is always an agreement about where, who, and what was seen as snow.

## Deciding the first snow

Back to the opening question: who decides the first snow? In Seoul, the observer at the observatory on a hill in Songwol-dong, Jongno-gu. The moment the observer looks at the sky and writes "snow", the first snow has come to Seoul, whether or not it snowed outside my window.

That might sound a little disappointing. But it is why the first snow of 23 October 1981 and that of 10 December 2020 can stand in one line and be compared. Both were seen from the same place, in the same way, across more than a century. That is what an official record is for. Something seen steadily in one way is more useful than something seen by many.

Still, first snow is not only a record. Everyone has their own first snow, regardless of the official one: a flake that stuck to the bus window on the way to work and melted, sleet on a mountain, the moment a child shouted "It's snowing!". None of these make it into any record, but for that person, that was the day of the first snow.

When the observer in Songwol-dong writes down this year's first snow, I probably won't be looking out of the window. But if something white drifts past my own window before that day comes, I plan to call it my first snow.

*A Page for Today* is a series in which I open my idea notebook to a random page and lay that day's thoughts on the line I find there. This time, the single word "first snow" became the twenty-ninth piece. If you spot a wrong date or rule, please let me know and I will correct it.

## Sources

- KMA Seoul Metropolitan Office, Seoul Weather Observatory (history, heritage listing, WMO centennial station) https://www.kma.go.kr/metropolitan/html/info/business06.jsp
- KMA Open MET Data Portal, "snow days" description and sites observed by eye (as of August 2024) https://data.kma.go.kr/climate/snw/selectSnwChart.do?pgmNo=697
- Seoul Metropolitan Government, *Seoul Sarang*, "Seoul's first snow has to fall here: the Seoul Weather Observatory" https://love.seoul.go.kr/articles/3101
- *Seoul Shinmun* (9 Nov 2021), on KMA staff in Jongno having to see snow with their own eyes https://seoul.co.kr/news/society/2021/11/09/20211109500047
- YTN (10 Nov 2021) https://www.ytn.co.kr/_ln/0108_202111100844479756
- *Segye Ilbo* (17 Nov 2023) https://www.segye.com/newsView/20231117508221
- *Financial News* (10 Dec 2020) https://www.fnnews.com/news/202012100626080681 · (4 Dec 2025) https://www.fnnews.com/news/202512041919329651
- *Kyunghyang Shinmun* (25 Nov 2015) https://www.khan.co.kr/article/201511251738591
- *Dailian* (24 Nov 2018) https://m.dailian.co.kr/news/view/753944
- *Weekly Kyunghyang* (27 Nov 2024), from Seoul's first snow to a "snow bomb" https://weekly.khan.co.kr/article/202411271430011
- Kyoto Local Meteorological Office, notice on changes to normals following the automation of visual observation reports (2020) https://www.data.jma.go.jp/kyoto/2_data/kisetsu/new_heinenchi.pdf
- Weathernews (Oct 2026) https://weathernews.jp/news/202610/220041/
- Aju Business Daily (29 Oct 2026), first frost in Seoul, first snow on Seoraksan (Korean) https://www.ajunews.com/view/20261029133122380 · Kangwon Ilbo (29 Oct 2026) https://www.kwnews.co.kr/article/20261029500707
- YTN (1 Nov 2026), Mount Fuji's first snowcap confirmed by radar for the first time (Korean) https://www.ytn.co.kr/_ln/0104_202611011537299346
- US National Weather Service, Snow Measurement Guidelines https://www.weather.gov/media/ffc/snow_measurement_guidelines.pdf · NWS Jackson, KY, Snowfall Climatology https://www.weather.gov/jkl/snowfallclimatology

