ABSTRACT In many places, magnitudes and frequencies of floods are expected to increase due to climate change. To understand these changes better, trend analyses of historical data are helpful. However, traditional trend analyses do not address issues related to shifts in the relative contributions of rainfall versus snowmelt floods, or in the frequency of a particular flood type. We present a novel approach for quantifying such trends in time series of floods using a fuzzy decision tree for event classification and applied it to maximal annual and seasonal floods in 27 alpine catchments for the period 1980–2014. Trends in flood types were studied with Sen’s slope and double mass curves. Our results reveal a decreasing number of rain-on-snow and an increasing number of short rainfall events in all catchments, with flash floods increasing in smaller catchments. Overall, the results demonstrate the value of incorporating a fuzzy flood-type classification into flood trend analyses.

中文翻译：

---

高寒流域洪水类型趋势分析

摘要 在许多地方，由于气候变化，预计洪水的规模和频率会增加。为了更好地了解这些变化，历史数据的趋势分析很有帮助。然而，传统的趋势分析并没有解决与降雨与融雪洪水的相对贡献的变化或特定洪水类型的频率变化相关的问题。我们提出了一种使用模糊决策树进行事件分类的量化洪水时间序列趋势的新方法，并将其应用于 1980 年至 2014 年期间 27 个高山集水区的最大年度和季节性洪水。用森的斜率和双质量曲线研究了洪水类型的趋势。我们的结果显示，所有集水区的雨雪次数减少，短时降雨事件数量增加，随着较小集水区的山洪暴发增加。总的来说，结果证明了将模糊洪水类型分类纳入洪水趋势分析的价值。