Theoretical developments for the analysis and modeling of extreme value data have tended to focus on limiting cases and assumptions of independence. However, massive datasets from models and sensors, space-time dimensionality, complex dependence structures, long-memory, long-range and low frequency processes all motivate the need for sophisticated methods for correlated and finite data that follow complex processes. The importance of extremes has been rapidly growing in areas ranging from climate change and critical infrastructures to insurance and financial markets. Here we briefly discuss the state-of-the-art and key gaps, through the case of rainfall extremes under climate change. Preliminary analysis suggests new directions and points to research areas that deserve further attention.
展开▼
