There are only 5 sunny days in the two months of July and August? The National Weather Service’s answer is… .

“oh! Why is it raining like this?” Housewife A (42) was surprised to see the monthly forecast provided by Microsoft ( MS

) to find out the weather this summer . When he entered July, it was raining all over the 28th except for the 7th, 20th and 26th. August did more. Except for the 17th and 31st, it rained on 29 days. Restrictions due to the new coronavirus infection (Corona 19) have also been lifted, and the first person who entered high school was planning to go on a short domestic trip with his family for the time being to prepare for the college entrance exam, but if the actual weather is the case, the trip is virtually impossible. seemed Mr. A said, “I saw an article that it was hot and rainy this summer, but I didn’t know it was this much.” The ‘weather forecast calendar’, which says that it rains on all but three or four days in July and August, including Seoul, has recently spread around online communities. Netizens who were trying to plan a summer vacation like Mr. A came across this and responded that they were absurd.

MS “Long-term predictions of more than 10 days may not be accurate”

We asked Microsoft, which provided the information, for an official position on how it forecasts the weather.

After confirming the headquarters, Microsoft Korea said, “Microsoft is generating accurate short-term weather forecasts by combining multiple observational data sets and proprietary algorithms.” can,” he said. According to Microsoft’s explanation, short-term forecasts of about 10 days are accurate, but forecasts for each day in July and August are likely to be inaccurate.

Then, it was announced that the data necessary for weather forecasting in Korea is provided by ‘K Weather’, a private weather company, and added, “Microsoft is constantly striving to provide accurate weather forecasts.”

“Globally, there is no daily forecast in 2-3 months”

The Korea Meteorological Administration is also in the position that daily forecasts over a long period of time are virtually meaningless.

An official from the Korea Meteorological Administration explained, “There is no country in the world that makes daily forecasts for two to three months in the future.”

There are several reasons why forecast uncertainty increases as the distance from the reference date for predicting weather increases. First, there is a possibility that the initial value for forecasting is wrong because there is an error that the observation equipment cannot measure. If this initial value is input, the difference from the actual weather becomes very large if the minute error at the time of measurement reaches a long period of time. An official from the Korea Meteorological Administration likened it to “a similar principle to the fact that when archers aim at a target and there is a slight difference in angle, the gap increases when an actual arrow hits the target.”

Therefore, in order to accurately measure the base value for weather forecasting, it is necessary to continuously perform corrections by reflecting additional information that occurs every moment. In the end, it means that the first measured value actually fits within three days, at the most, within ten days.

An official from the Korea Meteorological Administration said, “It is unknown how Microsoft uses the base value for long-term forecasts of more than 10 days and applies a numerical model, but the uncertainty of long-term forecasts will be the same as the Korea Meteorological Administration.” The Korea Meteorological Administration explains that although it is possible to mechanically produce a predicted value after several years rather than a month by entering the base value, it is impossible to know whether it will actually match. This is because, like a kind of 메이저사이트‘butterfly effect’, even a small variable can change the weather greatly.

Korea Meteorological Administration “Long-term outlook with probability divided into 3 quartiles”

If so, is the long-term forecast such as one month or three months announced by the Korea Meteorological Administration meaningless?

According to the ‘Understanding long-term forecast’ data provided by the Korea Meteorological Administration to the Hankook Ilbo, the Korea Meteorological Administration classifies forecasts into short-term (3 days), medium-term (10 days), and long-term forecasts of more than one month. Short-term forecast, which announces forecasts of 12 meteorological elements, such as temperature, precipitation, and sky conditions, within 3 hours at 3-hour intervals, and mid-term, which is divided into morning and afternoon from the base date to the next 10 days and announced twice daily (every day for 8 to 10 days) Forecasts are called weather forecasts, but one month later is called climate forecasts. I heard the analogy that the weather is a person’s mood that changes depending on the mood, and the climate is like a person’s personality that does not change greatly.

An official from the Korea Meteorological Administration said, “There is a social demand for weather forecasts, so information is provided for about 10 days, but beyond a month, there is no basis for determining the weather by day.”

Therefore, the Korea Meteorological Administration performs several climate prediction models to predict the future probabilistically, taking into account the fact that the range of fluctuation increases over time when predicting the long-term with the current weather. For example, if 25 of 50 models predict that the temperature will be higher than normal, 15 will be similar to normal, and 10 will be lower than normal, there are 50%, 30%, and 20% probabilities, respectively.

The Korea Meteorological Administration divides the climate values ​​of the past 30 years into three categories, lower than normal, less/similar to, higher, or more → forecasts the distribution of predicted values ​​compared with past climate values ​​→ announces forecasts based on the probability of occurrence for each of the three quartiles are doing

Specifically, a long-term forecast is predicted through the process of ① live situation analysis ② prediction model analysis ③ impact analysis. In the real-time analysis stage, analysis of long-term trends such as global warming, recent temperature, and precipitation, as well as analysis of short-period trends such as El Niño, movements of major air masses that affect Korea, and oscillations in the Arctic and the North Atlantic Ocean are conducted.

After data analysis such as long- and short-term trends, domestic and foreign model analysis, and climate impact surveys were completed, climate forecasting expert meetings (four times a year) and East Asia expert meetings (twice a year) were held, followed by forecasting meetings. Afterwards, forecasts are produced and published.

According to the three-month forecast announced last month after going through this process, there is a 40% chance that the July temperature will be similar to or higher than normal (24-25.2 degrees), and the precipitation will also be similar to or higher than normal (245.9-308.2 mm). 40% each.

Leave a Reply

Your email address will not be published. Required fields are marked *