Have you ever heard about “weather wanderer”? It means people who trust Korean weather information provided by foreign countries instead of the KMA (Korea Meteorological Administration). Due to the recent errors of the KMA, the number of Korean weather wanderers, who do not trust the KMA, are increasing. 

 The KMA predicted a heat wave for this summer, not a rainy season; but on the contrary, there was an all-time long rainy season. Many people blamed the KMA for the massive damage, including urban inundations and loss of lives. In addition, several media sources reported that the KMA made a mistake in predicting the path of typhoon Bavi, but the U.S. and Europe correctly predicted the path. While Korean distrust of the KMA is increasing, rather than unconditional criticism and neglect, we need to know exactly why the KMA has made so many errors in weather forecasting. Despite the constant errors, Korea's weather forecasting system, which is still in a period of transition, is essential in our lives and a problem we must solve. The Argus looked at the problems, causes and solutions for the Korean weather forecasting system based on the three elements of weather forecasting. 

Problem

Reducing climate accuracy

 Korea is originally a country with narrow land and complex terrain, so it is difficult to forecast the weather accurately. On average, the probability of the weather forecast being right is about 85 percent. Therefore, more than 40 days are incorrectly forecasted in a year (365 days). The problem is that even 85 percent of the hits are hard to get because of Korea's climate conditions. In fact, according to the Board of Audit and Inspection in Korea, in 3,228 (62 percent) of 5,193 cases of predicted rain by the KMA for five years (2012-2016) it actually rained; and in 1965 cases (38 percent) it did not rain. In addition, in cases where the KMA predicted that it would not rain, in 1,808 cases there was rain. The average hit rate of precipitation is only 46 percent. 

 A recent example of the KMA's weather inaccuracy is the rainy season of this summer that lasted 54 days. The KMA forecasted rain of 50 to 100 mm in Gwangju, Korea from Aug. 7-8th. In fact, however, there was a lot of flooding damage in Gwangju during Aug. 7-8th, with a heavy rain of 516mm. If the accuracy of the weather forecast by the KMA decreases, the damage caused by the failure to prepare for heat waves, heavy rain, and heavy snow is significant. According to the central Disaster and Safety Countermeasures Headquarters and the Ministry of Interior and Safety, the rainy season actually left 50 people dead or missing, and caused more than 6,000 people to become homeless.

Lack of Numerical Weather Prediction Model

 According to the Board of Audit and Inspection in Korea's 2019 audit results report (the actual status of the project to develop numerical forecast models), the Numerical Weather Prediction Model covers about 40 % of the tasks for accurate weather forecasts. This is an important part compared to other factors such as quality of weather observation data (32%) and capacity of forecasters (28%). Korea used a model developed in Japan before 2010, and used from 2010 UM (U.K. model) modified to suit the Korean weather situation.

 In the meantime, from 2011 to 2019, about 70 people and a budget of about 80 billion won were invested to develop a Korean Integrated Model (KIM). The use of foreign models is based on the disadvantages of spending on user fees, not reflecting the weather characteristics of our country, and the inconvenience of updating them. All parts of the model were made with Korea's own technology, which was significant and expected to improve the accuracy of the forecast. The model went through trial operation from spring of 2019 and was introduced in the forecast from 2020. 

However, after the introduction of KIM, the accuracy of the forecast of precipitation has decreased. It is being used in parallel with UM to supplement KIM, which was in its early days, and the accuracy of rainfall forecast was 67% from April to July 13 this year, down from last year. 

Inadequate utilization of meteorological equipment

 Weather prediction requires weather satellites that collect data and supercomputers that simulate based on this data. But these pieces of weather equipment have some problems. Before the 2000s, Korea received data every 30 minutes from the Japanese weather satellite Multi-Functional Transport Satellite (MTSAT). This caused a series of errors in weather forecasts in Korea if Japan's weather forecasts were not accurate. Then, in the 2000s, it began using the COMS (Communication, Ocean & Meteorological Satellite) otherwise known as the Chollian satellite. COMS launched in 2010 and had better analytical capabilities than Japan's satellite. However, the 2017 Board of Audit and Inspection in Korea pointed out that the KMA failed to utilize COMS. This is because it launched COMS but failed to develop software to analyze the satellite data.

 Also, before the 1990s, the KMA did not have supercomputers. The KMA predicted the weather with a mainframe computer, which was mostly used by banks and conglomerates. Supercomputers were also introduced in Korea in the 1990s, but in 2010 they did not perform well. So two new supercomputers were introduced in 2011 and used until 2016.  At the end of 2015, the KMA brought in supercomputer No. 4 for 55 billion won. Despite the tens of billion won needed for purchase, maintenance and processing costs, the KMA has introduced a new supercomputer which costs about 40 billion won every five years, since 2000 for weather accuracy. Despite this huge investment in meteorological equipment, problems arise due to poor utilization of these technological resources.

Cause

Geographical characteristics and abnormal climate of Korea

 The Korean Peninsula is located in the mid-latitude region, where the energy imbalance is large due to the temperature difference between cold air in the polar regions and hot air in the equator. Moreover, since it is on the border between the continent and the ocean, it is more volatile due to the influence of both sides. In addition, it is more difficult to predict the weather on the Korean Peninsula because the Korean Peninsula is bordered on three sides by the sea; and there are many mountain ranges. From the beginning, Korea has limitations in improving the accuracy of weather forecasts. 

 South Korea's extreme climate change is adding complexity to this. According to the “Korea Climate Change Evaluation Report 2020” published by the Ministry of Environment and the Korea Meteorological Administration in July, the average surface temperature of Korea rose 1.8 degrees Celsius from 1912 to 2017. This is more than double the world's average increase (0.85 degrees). According to the report, if the current greenhouse gas emission trend continues, the temperature of the Korean Peninsula in the late 21st century will be 2.9-4.7 degrees higher than the current level. The reason Korea's climate change is especially severe is because the mid-latitude is often abnormal due to global warming. As greenhouse gas emissions have increased, Arctic temperatures have risen, and the jet stream has weakened, making it easier for cold air in the north to descend to the mid-latitude. This phenomenon is the cause of severe downpours and the longer-than-expected rainy season this summer.The more extreme this climate phenomenon becomes, the more difficult it is to predict the flow of the atmosphere, resulting in weather errors.

Poor numerical forecast model in Korea

 While many developed countries have their own weather forecast models, Korea did not have a local weather forecast until recently. So, the KMA used the Japanese model from 2002 to 2010 and the British model from 2010 to 2019. But there are many problems in terms of geography. Unlike Korea, which has a lot of climate change, the British climate is almost the same as the West Coast oceanic climate. In addition, the west is important in weather conditions as air moves from west to east because of the rotation of the earth. But west of England is the sea and west of Korea is a continent. In addition, it failed to accurately reflect Korea's topography and weather characteristics as it was a foreign technology, and there were also fatal disadvantages of not being able to upgrade the forecast model as required.

 Regarding the reduced accuracy of weather forecasting after the introduction of KIM, a KMA official said, “The KIM has not reduced the accuracy of weather forecasting. It's not a weather model problem; it's a forecasters' decision.  In addition, KIM was introduced earlier this year, so it is too early to judge performance. At least a year of trial and error is needed.” Also, the KIAPS (Korea Institute of Atmospheric Prediction Systems) project group that developed KIM has been disbanded, and the KMA is managing the KIM currently. This has caused problems with the update technology. KIM is an independent and creative technology that Korean experts of KIAPS have worked on for nine years. Due to the complex characteristics of numerical forecast models, it is not easy for non-developers to know and handle the system better than developers.

Rapid development of meteorological equipment and lack of experts in equipment utilization

 First of all, a large national budget is needed to operate supercomputers, which are weather equipment. To install a very large supercomputer, new buildings needs to be built, and maintenance costs are expensive. Also, supercomputers have improved their performance by a thousand times over the last ten years due to their rapid development. Even if a country spends a lot of money to buy a supercomputer, it will lose a lot of money in a few years because its value will be very low. 

 However, the fundamental cause of the weather forecast inaccuracy is that weather equipment is developing so fast and experts are not using the equipment properly. For example, it is forecasters who complete weather forecasts by interpreting weather data calculated by supercomputers. Park E-hyeong, a spokesman of KMA, said, "There are times when errors in weather forecast data and actual conditions occur. It is the role of forecasters to reduce and compensate for these errors. South Korea's forecasters make weather forecasts first by determining whether inaccuracies could cause human casualties or not."

 KMA forecasters, who play such an important role, work in poor conditions. Day and night shift work causes fatigue. Even if the forecast is slightly wrong, there are many complaints. In fact, according to a survey submitted by KMA, 57 percent of all KMA employees said they did not want to take forecasting tasks, which is a key responsibility of the organization. KMA employees answered that forecasting tasks cause physical and mental stress. Shriveled by criticism, forecasters tend to make conservative predictions (safe rather than accurate ones), and then the public is likely to take them as wrong forecasts again. In addition, forecasters must have expertise accumulated over a long period, but it is difficult to attain this experience due to a cyclical work system that changes positions and regions every two years.    

Solution

Data exchange and AI development with neighboring countries

 The difficulty of weather forecasting due to Korea's geographical characteristics can be supplemented by actively accumulating Korean climate data. Korea needs to collect more detailed data because of the slow accumulation of data among advanced countries. To achieve this goal, cooperation in weather data with neighboring countries is also a way. Currently, South Korea, China and Japan are exchanging and studying data such as fine dust and earthquakes under a project called the Joint Research Cooperative Program (JRCP). As such, we need diverse international cooperation to exchange weather data with neighboring countries and study to increase accuracy.

 Also, a bigger factor that undermines weather accuracy is climate change. As climate change is getting worse, weather data from the past are limited in predicting abnormal weather conditions that could change rapidly in the future. To solve this problem, we need to build an AI system that can predict the future. For example, a research team led by Professor Ham Yoo-geun of the Department of Oceanography at Chonnam National University has developed a technology that can predict El Nino up to 18 months in advance by using AI. El Nino is an abnormal climate phenomenon in the eastern and central parts of the Pacific Ocean that causes heat waves and floods, with seawater temperatures rising more than 0.5 degrees above an average year. The AI, developed by Professor Ham’s research team, helps prepare for various weather disasters caused by El Nino by diagnosing in advance the occurrence and strength of El Nino. In addition to El Nino, we need to introduce AI technology that can predict future climate changes so that we can accurately predict abnormal weather conditions to prepare for related damage. 

Update of Korean Numerical Forecasting Models 

 KIM which has been test-run since April this year, has no significant difference in performance compared to UM and overseas numerical models. But this is only when it was first introduced. Data needs to be updated periodically to improve the performance of numerical forecast models. 

 The KMA will also set up a separate research team to deal with the KIM, including experts from the business group that developed the KIM, to upgrade it in the future. Just as the KMA updated its physical processes and epidemiological models to the KIM in June, it will have to make continuous updates. According to the Korea Meteorological Administration, the KIM will be supplemented by various tests and upgrades in the future and will reach the current UM level within two years. "The KIM has been going through four seasons in Korea for a year and is in the process of collecting Korean-tailored data. We are updating it to improve performance," a KMA official said.

Developing human resources and technology that can utilize weather equipment 

 The nation's artificial satellite Chollian 1 ended its weather observation mission as of April this year. It is necessary to develop satellites for weather observation to replace it. However, it is more urgent to establish a system that can make good use of current equipment than to develop weather equipment that requires a long-term investment. First of all, the KMA announced in December that it would introduce a professional civil service system to give preferential treatment to professional forecasters, thereby eliminating circular placement problems and increasing the accuracy of forecasts. As a result, about 10 professional government officials who are exclusively in charge of forecasting work are currently working as forecasters. 

 In addition, the agency is developing an AI weather forecast assistant called 'Alpha-weather' to assist forecasters. The KMA plans to complete the development of ‘Alpha-Weather’, by next year, and to create an advanced version, which can provide region-specific weather forecasts by 2024. Ultimately, the KMA's goal is to develop an ‘Alpha-weather’ that can provide customized weather information for each individual citizen by 2027. The completion of such "Alpha-weather" technology is expected to improve the working conditions of forecasters and increase the accuracy of forecasts.

 Because the KMA is losing the trust of the public, Park E-hyung, a spokesman of the KMA, said, “The KMA's main goal is to prevent casualties caused by weather disasters. Therefore, the KMA makes a forecast as soon as it finds any possibility of human casualties. We hope you observe the weather agency's warnings carefully. We can bear the criticism that your clothes are wet from the rain because of KMA errors. However, the KMA cannot guile and feels a huge responsibility if the agency's incorrect forecast causes human casualties.” 

 As the KMA said, weather forecasts are directly related to the safety of the people. Therefore, the KMA should work hard to pinpoint and resolve the problems and causes that cause forecast errors, and we hope that the Korean people will listen carefully to the voice of the Korean Meteorological Administration. The Argus hopes that in the near future, Koreans will see their own weather service with a loving eye.

By Choi Yea-jin
yeajinchoi@hufs.ac.kr