In South Korea, the snowy season spans from October to April, and the annual average snowfall varies significantly depending on specific regions, latitudes, and elevations, ranging from 0 to 260 cm. The average annual snowfall in South Korea is 25.1 cm. Despite of the relatively shallow snowfall depth, over the past decade, South Korea has experienced approximately 120 million dollars in damages attributed to snow-related incidents. In this study, the DPSIR (Driver-Pressure-State-Impact-Response) framework was employed to consider the meteorological and socioeconomic factors to calculate the snow damage vulnerability. A total of 17 indicators were taken into account to comprehend meteorological conditions, socioeconomic factors, and historical damage records from 1994 to 2020. However, due to the limited availability of meteorological observatories and changes in greenhouse design standards, accurately estimating the snow damage amount poses challenges. Therefore, based on the vulnerability, the risk levels were classified into four categories and estimated snow damage generated by the categorized models was compared with those of the model constructed using the entire dataset. The categorized models offer improved estimation results, as the meteorological and socioeconomic characteristics within each category differ and should be addressed separately in modeling. Among the categorized models, the Green zone exhibited the best results, primarily because it did not include outlier snow damage incidents. The developed model in this study could be utilized to mitigate the impact of heavy snowfall and prioritize snow removal regions.

在韩国，雪季从 10 月持续到 4 月，年平均降雪量因特定地区、纬度和海拔高度而异，从 0 到 260 厘米不等。韩国的年平均降雪量为 25.1 厘米。尽管降雪深度相对较浅，但在过去十年中，韩国因降雪相关事件而遭受了约 1.2 亿美元的损失。在本研究中，采用 DPSIR （Driver-Pressure-State-Impact-Response） 框架来考虑气象和社会经济因素来计算雪灾脆弱性。总共考虑了 17 个指标来理解 1994 年至 2020 年的气象条件、社会经济因素和历史损失记录。然而，由于气象观测站的可用性有限和温室设计标准的变化，准确估计雪灾量带来了挑战。因此，根据脆弱性，将风险等级分为四类，并将分类模型产生的估计雪害与使用整个数据集构建的模型产生的估计雪害进行比较。分类模型提供了改进的估计结果，因为每个类别中的气象和社会经济特征不同，应在建模中单独解决。在分类模型中，绿色区域表现出最佳结果，主要是因为它不包括异常值雪灾事件。本研究中开发的模型可用于减轻大雪的影响并确定除雪区域的优先级。