Kazuki Sakikawa, Hidetaka Chikamori, Ryoji Kudo
Received 1 February, 2026
Accepted 2 July, 2026
Published online 18 September, 2026
Kazuki Sakikawa1), Hidetaka Chikamori2), Ryoji Kudo2)
1) Institute for Rural Engineering, National Agriculture and Food Research Organization, Japan
2) Graduate School of Environmental, Life, Natural Science and Technology, Okayama University, Japan
Flood and sediment disasters caused by heavy rainfall have intensified under climate change, necessitating reliable bias correction of simulated precipitation for flood risk assessment. We integrated the Meta-statistical Extreme Value (MEV) framework into the Quantile Mapping (QM) method for annual maxima of daily precipitation and evaluated its performance across Japan. The MEV framework estimates the non-exceedance probability of annual maxima using all daily precipitation samples, thereby reducing estimation variability. The MEV-based bias correction reduces errors relative to uncorrected values and exhibits lower variability than QM applied to the annual maximum series method. Its performance depends on the limited within-year sample size, the goodness-of-fit of the daily precipitation distribution, and regional climatic characteristics; thus, its applicability should be carefully considered. Specifically, because the MEV framework partitions daily precipitation by year, the limited annual sample size induces substantial year-to-year variability, particularly in the upper tail. Distributions that better represent the upper tail tend to reproduce this variability, leading to larger correction errors and increased correction variability. Moreover, in regions with frequent heavy rainfall, the MEV-based bias correction becomes less accurate and yields more variable estimates, depending on the applied daily precipitation distribution.
Copyright (c) 2026 The Author(s) CC-BY 4.0
