CAREER: Temporal Clustering of Hydrometeorological Extremes

本CAREER提案的主要目标是检验极端水文气象事件是否及时聚集。时间聚类是指事件在时间上同时发生的趋势，极端事件的发生会影响后续事件发生的可能性。提议的工作分为三个阶段，研究的动机是以下主要问题：极端水文气象事件是否表现出时间聚类，如果是，哪些关键的物理过程解释了这种行为？总循环模型（GCMs）的输出能再现观察到的群集行为吗？是否有可能利用时间聚类来改进水文气象极端事件的预报？基本假设是极端水文气象事件表现出受气候过程控制的时间聚类。美国大陆将成为重点，因为该地区受到大量自然灾害的困扰，产生了广泛的社会经济影响。该地区最具破坏性的灾害包括：洪水和强降雨、极端高温和低温以及热带和热带外风暴。将开发Cox回归模型来检验极端事件对气候过程的依赖性。这些方法建立在PI广泛使用的分析工具和数据集的基础上。这项研究在理解极端事件的频率和原因方面迈出了全面的一步；这有更广泛的影响。

The main goal of this CAREER proposal is to examine whether extreme hydrometeorological events cluster in time. Temporal clustering refers to the tendency of events to occur together in time, with the occurrence of an extreme event affecting the likelihood of a subsequent event. The proposed work is delineated in three phases, with research motivated by the following main questions: Do extreme hydrometeorological events exhibit temporal clustering, and, if so, which key physical processes explain this behavior? Can outputs from General Circulation Models (GCMs) reproduce observed clustered behavior? Is it possible to take advantage of temporal clustering to improve the forecasting of hydrometeorological extreme events? The underlying hypothesis is that extreme hydrometeorological events exhibit temporal clustering that is controlled by climate processes. The continental United States will be the focus as this area is plagued by a large array of natural hazards yielding extensive socio-economic impacts. Some of the most damaging hazards in this general area will be included: flooding and heavy rainfall, high and low temperature extremes, and tropical and extra-tropical storms. Cox regression models will be developed to examine the dependence of extreme events on climate processes. The methodologies build on analysis tools and data sets that the PI has used extensively. This research represents a comprehensive step forward in the understanding of the frequency and causes of extreme events; this has substantial broader impacts.