针对具有复杂数据结构的极端欧洲冰雹事件的统计模型

该项目将开发新的统计方法，以整合多种数据来源，从而在气候变暖背景下评估大冰雹事件的风险。贝叶斯方法和统计分析等技术将解决缺失数据问题，并评估冰雹风险随时间的变化情况。申请者应具备数学、统计学以及R语言编程方面的背景。

Job description: Details: Background: Large hail is a rare but impactful weather hazard, capable of causing extensive damage to homes, vehicles, crops and infrastructure. European hail observations come from a diverse range of sources including weather radar, hail pads, insurance claims data and citizen science, differing greatly in spatial coverage, quality and record length. This project will develop new statistical methods to combine these diverse data sources, producing a more robust picture of the distribution of large hail events and reducing uncertainty in our understanding of changing atmospheric perils under a warming climate. A central aim will be to quantify how the risk of severe hail may evolve over the coming years as the climate warms. The student will develop methods to account for missing or sparse observations, and distinguish changing risk from changing observation practices. This project will benefit from partnership with Liberty Mutual, a global insurer with extensive expertise in atmospheric perils. PhD project: A warming climate may be associated with extreme atmospheric perils, one being large hail events, a rare but highly damaging weather hazard. This project will quantify how the risk of severe hail may evolve over the coming years, utilising all available data on this peril effectively. The project will draw on statistical techniques from extreme value analysis, where theoretically-motivated approaches allow for extrapolation and estimation in a distribution’s tails. Model-building here requires tackling challenging data integration problems, leading to missing data problems. Data may be measured at different spatial locations or resolutions, with structural gaps occurring when they are combined. There may be varying levels of quality associated with each data input. Data may also be unrepresentative due to some locations being easier to access than others. To address these, the project will investigate using Bayesian methods to model extreme events that simultaneously deal with missing values within a unified framework. Statistical or machine learning methods will also be explored to deal with Structured Missingness arising from integrating data sources with incomplete overlap. The primary supervisor (Prof ［会员专属信息］) is an expert in missing data methods with the co-supervisor (Dr Simpson) an expert in extreme value modelling, external partners are Dr Rhodes and Dr Hernandez at Liberty Mutual. Liberty’s involvement grounds the project in real-world relevance facilitating a direct route to practical impact. As a global insurer with substantial exposure to hail risk, partnership with Liberty will provide an industry perspective on how hail risk is currently managed, and a view on the relevance of this research within the exposure management and catastrophe modelling space. Guidance provided will ensure the research aligns with industry-relevant questions. Regular contact with researchers at Liberty also gives the student the opportunity to build a professional network. Applicant Profile: Students with a strong background in mathematics and statistics, who have an interest in developing new statistical methodology with applications in climate science. Experience with coding in R will also be useful.

资助说明： 该项目属于 UNRISK CDT 项目的一部分，该项目提供 15-18 项由全额资助的 NERC 学生奖学金，涵盖完整的大学学费；按标准 UKRI 标准发放的个人津贴；6000 英镑的个人研究与培训费用；5000 英镑（每人）的群体级培训费用；以及用于特殊项目的“灵活基金”。 国际申请者需承担申请学生签证及国际健康附加费相关的费用（IHS） 针对英国及国际申请者开放申请。根据 UKRI 规定，国际申请者的录取名额有限。 关于 UNRISK 的更多信息可访问其网站：[会员链接]