Exploring nonstationary characteristics of distributions of annual rainfall indices in Taiwan using GAMLSS

Abstract

Global climate change would induce changes of the rate of hydrologic cycles and intensify the frequency and intensity of hydrologic extreme events Changes in rainfall characteristics lead to the hypothesis of stationary in traditional hydrologic analysis no longer applicable Nonstationary analysis of rainfall characteristics thus becomes one of the important issues in water resources management In this study the generalized additive models for location scale and shape (GAMLSS) is adopted for the nonstationary analysis of the annual rainfall indices in Taiwan Ten annual rainfall indices are used to detect alterations of the magnitude duration and extreme of annual rainfall regime at eight rainfall stations in the north central south and east regions of Taiwan The results indicate that 55% of the annual rainfall indices are nonstationary and the characteristics of regional similarities are not found in most of the annual rainfall indices Trends of some annual rainfall indices are observed at different stations Annual 1-day maximum rainfall and annual 2-day maximum rainfall have upward trend in the past 30 years at Taipei Hengchun and Dawu stations Annual maximum consecutive rain days has downward changes in the south and east regions Daily rainfall intensity has rising trend in the past 40 years at most stations except for Sun Moon Lake and Chengkung stations Annual dry days has increasing trend at Taipei Hengchun and Dawu stations while the downward trend in recent years has been observed at Yilan Taichung and Chengkung stations in recent years

Date of Award	2019
Original language	English
Supervisor	Jenq-Tzong Shiau (Supervisor)