This Place is Dangerous for Flood Damage This Summer?!...AI Risk Map Shows

[Financial News] An analysis has shown that major cities like Seoul and Busan have high flood damage risks.
On the 29th, a research team from POSTECH (Pohang University of Science and Technology) and Kyungpook National University used artificial intelligence (AI) to predict flood risks by region and created a 'flood risk map' for the entire country. The results showed that the risk due to heavy rain was high in major cities such as the Seoul metropolitan area and the Gyeongnam region centered around Busan.
This map was created by the POSTECH research team using AI. First, they analyzed flood damage data recorded by the Ministry of the Interior and Safety for each district nationwide over the past 20 years (2002-2021). Based on this, they subdivided four key elements determining flood risk: 'hazard' (how much rain falls), 'exposure' (population and facilities exposed to risk), 'vulnerability' (the degree of susceptibility to damage), and 'response capability' (how well one can cope), and taught these to AI.
Among various AI models, the 'XGBoost' and 'Random Forest' models predicted flood damage with high accuracy of over 77%. Interestingly, the two models each identified different elements as the most important. XGBoost analyzed the 'impervious surface ratio' (the proportion of surfaces that rainwater cannot penetrate) as the biggest risk factor, while Random Forest identified 'river area' as the most significant. Nevertheless, both AI models evaluated major cities like Seoul and Incheon as 'high flood risk areas.' This shows that these areas are more vulnerable to damage due to high population density, extensive concrete surfaces, and concentrated buildings and infrastructure around rivers.
The most significant achievement of this study is that it has become possible to numerically evaluate 'prediction uncertainty' regarding flood risk. Regions commonly assessed as dangerous by multiple AI models can be prioritized in disaster prevention policies, while areas where model evaluations differ can be classified as needing further investigation. This is explained to provide practical help in devising effective flood measures with limited budgets.
Furthermore, the research team also proposed practical solutions. Since 'impervious surface ratio' and 'river area' were identified as major risk factors through AI analysis, they emphasized the need for nature-friendly urban development policies, such as securing green spaces where rainwater can naturally be absorbed into the ground and restricting development around rivers, to reduce flood damage.
The first author of the paper, Eunmi Lee from POSTECH, stated, "We were able to precisely analyze environmental changes and actual damage data using AI," and expressed hope that "it will also help in preparing practical flood response strategies." 
This research was conducted with support from the National Research Foundation of Korea's Science and Engineering Research Support Program and the Hyundai Motor Chung Mong-Koo Foundation, and was recently published in the environmental science journal 'Journal of Environmental Management.'jiany@fnnews.com Jiyan Yeon, Reporter