Projected variations in the regional clustering of precipitation stations around Chicago

ABSTRACT: Precipitation frequency analyses are typically performed for regions or groups of neighboring gauges that represent similar topographic and climatic characteristics. For past precipitation frequency analyses, regions are defined using observed datasets. However, for frequency analysis of projected precipitation, climate change can influence the grouping of sites, such as homogeneous regions of extreme precipitation. Therefore, this study investigates the effect of region definition in the regional frequency analysis (RFA) of extreme precipitation in climate change scenarios. Specifically, we use a statistically downscaled climate modeling-based dataset for Chicago, Illinois, USA and 8 climate change cases (4 models with 2 future climate scenarios). The cases were developed using the asynchronous regional regression model, which focuses on accurately resolving the tails of the probability distributions of precipitation data. For the 40 stations around Chicago, the clustering of precipitation stations varies. The precipitation characteristics, such as the averages of the monthly maximum and annual precipitation and the L‑moments of the annual maximum daily precipitation, vary significantly over different time periods and regional clusters. Furthermore, the number of stations that exhibit heterogeneity in terms of their clusters is lower when changes in the clustering of the climate regions are considered than when these changes are ignored. The results of this work illustrate the need to consider changes in the regional clustering of precipitation stations in RFA, which is particularly useful for designing water-related infrastructure in response to climate change.