Long-Term Trend for Environmental Extremes with Changepoint Detection
Extreme environmental events like avalanches, blizzards, heat waves, hurricanes, wildfires, etc., often make profound impacts on human lives, economies, and ecosystems. As a result of climate change, extreme weather and climate events are projected to become more frequent and intense, thereby increasing its impact on many areas of life (USGCRP, 2018). Quantifying long-term linear trends in the environmental extremes can be a useful indicator to track and better understand the Earth’s changing climate. However, many environmental time series data contain inhomogeneous shifts in their means induced by various undocumented outside factors such as instrument changes, location changes, or regime shifts in local climate. Those mean changes, if ignored, can substantially affect the long-term trend estimates for the data. In this dissertation proposal, we propose using a genetic algorithm to estimate the number and times of changepoints in the environmental extremes as a data homogenization procedure before quantifying their long-term linear trends. We will illustrate our methodology by analyzing two different environmental extremes: monthly maximum coastal sea levels and monthly maximum snow depth. Our main objective is to better understand how the extreme coastal sea levels and extreme snow depth have been changing over time via the rigorous estimation of the long-term linear trends and return levels with consideration of the changepoints.