A Study of Determining Criteria for Special Weather Report on Heavy Rain Based on Regional Rainfall and Damage Characteristics

The global increase in temperature due to climate change is affecting regional hydrological cycles, resulting in an increase in the frequency and intensity of extreme rainfall in certain areas. South Korea is also experiencing more frequent localized heavy rainfall within short periods due to global warming. In response, the Korea Meteorological Administration (KMA) operates a two-stage heavy rain advisory/warning system to minimize damage during heavy rainfall and effectively respond to expected rainfall. The heavy rain advisory is issued when cumulative rainfall of 60mm (110mm) is expected over 3 hours (12 hours), while the heavy rain warning is issued when cumulative rainfall exceeds 90mm (180mm) over the same periods. However, the impact of heavy rainfall varies by region, depending on local characteristics such as infrastructure, geographical location, and topography. Currently, the heavy rain advisories/warnings apply uniform standards nationwide, which do not take into account regional characteristics.
Therefore, to enhance the effectiveness of the heavy rain advisory/warning system, it is essential to establish region-specific criteria that reflect regional rainfall characteristics and local factors (such as damage, topography, and soil). To this end, the KMA is conducting research to develop differentiated heavy rain advisory/warning standards for each region. This study analyzes regional damage characteristics related to rainfall by utilizing recent meteorological and damage data, considering the increasing spatial concentration of heavy rainfall and its intensity. Using cluster analysis techniques, the study identifies regions with similar characteristics in terms of rainfall frequency, damage frequency, and topographical features, and aims to provide supporting data for setting region-specific heavy rain advisory/warning standards. The data used in this study include rainfall observation data (3-hour and 12-hour cumulative) from 712 KMA AWS stations from 2016 to 2022, as well as heavy rainfall damage data from the Ministry of the Interior and Safety's Disaster Yearbook and the National Disaster Management System (NDMS). The K-means clustering method, which assigns data to the closest cluster center, was employed to determine the optimal number of clusters.