Wakai Atsushi, Hijioka Yasuaki, Yokozawa Masayuki, Watanabe Manabu, Sakurai Gen
Institute for Agro-Environmental Sciences, National Agriculture and Food Research Organization, Tsukuba, Ibaraki, Japan.
Center for Climate Change Adaptation, National Institute for Environmental Studies, Tsukuba, Ibaraki, Japan.
PLoS One. 2025 Feb 25;20(2):e0318335. doi: 10.1371/journal.pone.0318335. eCollection 2025.
The future increase of large-scale weather disasters resulting from the increased frequency of extreme weather events caused by climate change is a matter of concern. Predicting future flood damage through statistical analysis requires accurate modeling of the relationship between historical precipitation and flood damage. An analysis that considers precipitation as a time series may be appropriate for this purpose. Functional data analysis was applied to model the relationship between historical daily precipitation and daily flood damage for river basins in the Kanto and Koshin regions of Japan. Flood damage statistics from the national government and 1-km grid past precipitation data from the National Agriculture and Food Research Organization were used. The models obtained through the functional data analysis were more accurate than those derived from the simple linear regression without considering the time series of precipitation. The new models were also about four times more accurate in estimating the annual sum of flood damage, compared to the flood damage of each flood event. The accuracy of prediction was higher in recent years than in earlier years of the study period (1993-2020). The results showed that the influence of precipitation on flood damage was more apparent in recent years. This findings may imply that the progress of the river development project and the resulting improvement of the structures along the river have indirectly affected levels of flood damage associated with levels of precipitation.
气候变化导致极端天气事件频率增加，由此引发的大规模气象灾害的未来增长令人担忧。通过统计分析预测未来洪水损失需要对历史降水量与洪水损失之间的关系进行准确建模。将降水量视为时间序列的分析可能适用于此目的。运用函数数据分析对日本关东和甲信地区流域的历史日降水量与日洪水损失之间的关系进行建模。使用了国家政府的洪水损失统计数据以及国家农业和食品研究组织提供的1公里网格过去降水量数据。通过函数数据分析获得的模型比不考虑降水量时间序列的简单线性回归模型更准确。与每次洪水事件的洪水损失相比，新模型在估计年度洪水损失总和方面的准确性也高出约四倍。在研究期（1993 - 2020年）内，近年来预测的准确性高于早期。结果表明，近年来降水量对洪水损失的影响更为明显。这一发现可能意味着河流开发项目的进展以及由此带来的沿河结构改善间接影响了与降水量水平相关的洪水损失水平。
