Empirical likelihood methods and statistical applications in climate studies

点击翻译按钮获取中文摘要

My proposal consists of two projects, the first of which is concerned with statistical methodology and the second is concerned with statistical applications in climatology. My first project focuses on the empirical likelihood method which enables us to make inference on population characteristics from data. It does not require elaborate models and takes full advantage of the inexpensive and fast computing power available today. For example, one may use this method to compute an interval estimate for the true average height of a population without knowing the height distribution of the population. The empirical likelihood method is a modern method of inference introduced by Professor Art Owen at Stanford University in the late 80's. It has now evolved into an important and powerful tool for statisticians. I have already made contributions to the empirical likelihood theory. My focus now is on the problem of handling a technical constraint which affects the accuracy of the empirical likelihood method. This is an important problem which has attracted the attention of prominent researchers including Professor Jiahua Chen at UBC and Professor Owen himself. I hope to find a better solution to the problem. My second project is concerned with methods for historical temperature reconstruction which is important for understanding global warming. We only have temperature records from the time of the industrial revolution (roughly 1850) and onwards. To appreciate the warming in recent decades, we need a long-term historical perspective which requires reliable estimation/reconstruction of global temperature before 1850. My project will study a statistical model for estimating the historical temperature using proxy data such as tree ring and ice core data and other climate variables. This is an important problem directly related to research on global warming. It is of interest to statisticians and climatologists as well as the government.