The Detection Methods Of Chinese Extreme Temperature Events

In recent decades, extreme climate events caused serious influence on the national economy and environment, so it attracts growing interests. The Assessment Report of the United Nations Intergovernmental Panel on Climate Change (IPCC) refers to the effects of extreme temperature events. Research on extreme temperature focuses on the analysis of its definition. Ma Zhuguo et al added a revised value as the threshold value of the extreme temperature which based on January and July average temperature in the 1951-2003 year. Andrew et al (1999) defined that if the maximum temperature is greater than or equal to a given threshold value of temperature within 2-5 days, then we can say it is an extreme event. In addition, analysis of the extreme temperature events using different indices is also a major research direction. The main indices include cold night, warm night, cold day, warm day, the index of continuous heat wave, the index of continuous cold wave, frost index, and extreme temperature index and so on. In addition, the trend of the extreme temperature is also studied. The main feature of global climate is becoming warm. Existing studies indicated that the lowest temperature rise significantly in most regions as the temperature of the global surface land increase, and the lowest temperature changes faster than the maximum temperature. Therefore it demonstrated an asymmetry of warming day and warming night. In the recent decades, the maximum temperature increased slightly, and the minimum temperature increased significantly. In the last 40-50 years, the extreme minimum temperature and average minimum temperature of some regions increased significantly, especially in winter of the north in Chine.This dissertation discussed the robust mean value method and median value method, the robust mean and median detection method of sliding window to detect the extreme high and low temperature events. We compared the improved methods with traditional methods.Chapter 2 described the definition and property of the cumulant and the estimation algorithm of three-order invariant and four-order invariant. Meanwhile the Cornish-Fisher expansion was introduced. Chapter 3 gave a robust mean detection method and a robust median detection method. We actually calculated the maximum temperature of the summer and winter over the past 50 years in China. The advantages of the improved methods were illustrated from the perspective of time and space.Chapter 4 defined a robust mean detection method and a robust median detection method of sliding window similarly. The improved method is better than the other methods through numerical simulation. At last we used new detection method to calculate the maximum temperature of the summer and winter over the past 50 years in China, and analyzed the changing trend of extreme high and low temperature events.