Making appointments, planning, going out, one of the things people care most about is the weather. The current weather is important right now, but what's important to us is the weather after hours or days. In the morning of the 22nd, it rained around 10mm per hour in Seoul this morning. But there must have been many people who were upset in the morning. This is because there was no rain forecast in Seoul until yesterday evening. Not only in Seoul, but also in some central areas where heavy rains were issued, rains were stronger than expected, and it wasn't until 7:00 in the morning. Even in Gangwon-do, Yeongwol and Wonju were given heatwave warnings and changed to heavy rain warnings. At 8 am today, heat waves and heavy rain warnings were issued to these areas at the same time.

So was the last weekend. Last Sunday (19th), it rained a lot in the morning in Seoul. The rain, which had been falling hard, began to show a lull in around 10 am. The Korea Meteorological Administration predicted that there will be a lot of rain again in Seoul from around 3 pm. However, the rain that had been coming from 3 o'clock did not come, and the forecast from the Korea Meteorological Administration, which said that it would rain from the evening, also missed. The Meteorological Administration predicted a very strong rain of around 50mm per hour in Seoul at dawn the next day (the 20th), but in the rest of the day it rained around 5mm in the rest of the region, except that it fell 6~7mm per hour at Gangdong and Songpa. There was a big difference from the forecast. At this point, it is natural that dissatisfaction erupts.

Currently, you can receive a lot of weather information through various platforms other than the website of the Korea Meteorological Administration. The information is also based on forecasts produced by the Korea Meteorological Administration. Even if the platform is diversified, the weather is bound to depend on the forecast of the National Weather Service. Currently, the forecasting accuracy of the Korea Meteorological Administration is 62% (12 hour forecast l Critical Success Index (CSI) level as of June 2020.) The longer the forecasting time, the lower the forecasting accuracy. As will, how much of that you have no real improvement expected future yen citizens desired level of weather forecast is available'll learn first from the system meteorological Agency forecast.


● Weather forecast

undergo three processes greatly when I do this meteorological weather forecast. first, We need observations to know the current situation of the Earth.Because meteorological phenomena are very large global scales, we need to know not only the surroundings of the Korean Peninsula but also the situation around the world. Predict the future weather conditions through the numerical forecast model, which is a computer program, predicts the future weather using various physical equations that govern natural phenomena. Report and make final forecasts for the public. The forecaster does not just read the results of the numerical forecasting model. However, no matter how sophisticated the previous two processes are, the observations and models are not 100% accurate. In the end, the weather forecast must be equipped with three beats evenly, such as the precision of observation, the performance of the model, and the capabilities of the forecaster, but even with close and accurate observations, there are some observation errors, and the numerical forecast model is also new. like can not accurately predict the future situation. Finally forecasters also be an error occurs in predicting the results because people weather misinformation of meteorology is to generate deohaejyeo are inevitable limitations of each process.


● Korean model reader is why we need Should?

April 28, KMA Korean numerical forecast model (KIM, Korean Integrated model) introduced the. apseodo said numerical weather prediction models inde program to predict the weather, and our proprietary technology from one program to the heat It is made of. Previously, Korea was using a British model (UM), but because it is an overseas technology, it did not accurately reflect Korea's topography and weather characteristics, and there was a fatal disadvantage that it was impossible to upgrade the forecast model, which is essential for upgrading. On the other hand, if we have our own technology, we can easily upgrade the lacking part, and have the advantage of utilizing our weather characteristics. Because of this necessity, KIM was developed to spend 100 billion won over the past 10 years. Currently, it shows 98.9% of the performance of the UK model (UM). With the development of KIM, we have become the world's ninth-largest country with its own numerical forecasting model, and currently its performance ranks sixth in the world. (Hereinafter Korean model-KIM l British model-UM)


Currently, the Korea Meteorological Administration is using KIM and UM in parallel. Because KIM is not perfect and forecasters need time to adjust to KIM, they cannot trust and use KIM alone. Precipitation prediction accuracy itself shows 2~4%p higher performance than UM. (See picture)


What is encouraging is that we have developed and are using our own technology, which used Japan's numerical forecasting model just over 20 years ago. For reference, Japan's numerical forecasting model has been developed for 56 years, and our model has been in business for four months. The UK has also developed a model for 50 years, and we have reached a similar level in a short time.


● Grade drop after introduction?… What is the problem

Then, has the weather forecast changed since the introduction of KIM? According to the Meteorological Agency data obtained by SBS, the forecasting accuracy of the Meteorological Agency fell from April to early July. Based on the 3-day forecast, 2019 was 68%, but this year it fell 1%p to 67%. (Refer to the picture)
* For accuracy, POD (Probability of Detection) was used.

Of course, it did not introduce a better model than the existing UM, but obviously, there are cases in which KIM predicts the weather on the Korean Peninsula better than UM. Even on rainy days before the rainy season, the KIM forecast was more accurate than UM. Everything is consequential, but if the strengths of the two models were synergistic, the accuracy of precipitation forecasts should have increased. This is a part of the forecaster's lack of competence. The forecast building also has its own difficulties. This is because there is too little time to understand the characteristics of KIM, and there is insufficient data on where to trust KIM over UM. Everyone will have their own position, but it is a pity that the accuracy has decreased in the position of people who have to feel the weather.


● Can the forecast accuracy be improved?

First of all, there is ample potential for improvement in the observed data values. Currently, Korea has not received all the observations from all over the world delivered by the World Meteorological Organization (WMO). Meanwhile, observations from the World Meteorological Organization were first delivered to Japan and China. The Korea Meteorological Administration aims to improve these areas by the end of this year to provide quality data. If the goal is achieved, more data than previously used data can be used for the numerical forecast model. Even if you don't use all the observations in the model, it can be helpful enough to screen better data.

The second is a numerical forecast model. This is the area with the greatest improvement. Numerical forecast models grow based on experience. If the model's predicted results differ from the actual observations, the forecaster gives feedback and upgrades to improve this. Currently, KIM has been upgraded once in June and will upgrade once more from October to November of this year. Also, by 2022, if Supercomputer 5 is put in and can handle more extensive data than it is now, the synergy is expected to shine.

Finally, it is the forecaster's ability. Korea has been using UM since 2010. For this reason, current forecasters are well aware of the characteristics of UM. What's wrong with UM, what parts are good, etc. However, there is not enough control of KIM to be used in the future. It is not easy to expect forecasters to take the lead in recognizing the characteristics of KIM at a time when the KIM was put in and not a railroad in the rainy season. However, the sooner the forecasters adapt to the model, the sooner it will have a direct impact on the overall forecast given by the National Weather Service. And in the end, this will show the credibility and satisfaction that citizens feel at the Korea Meteorological Administration.

<Reference>
[2020.07.19 8 News] Korean forecast system'KIM' with 100 billion won, accuracy?