Extreme Value Theory at the Interfaces: Outliers, Extremal Dependence, Compound & Clustering Events
Convener:
Alok Samantaray
|
Co-conveners:
Meriem Krouma,
Leonardo Olivetti,
Jordan Richards,
Sebastian Engelke
We welcome contributions spanning theory, methodology, and applications, with a particular focus on, but not limited to:
Advancing EVT methods
• Tail-copulas, Pareto processes, spectral measures, χ/χ̄ diagnostics
• Extreme quantile regression, hybrid EVT- quantile regression models, non-stationary tail models
• EVT-constrained ML or hybrid physics–ML models
Applying EVT to extremes in the real world
• Compound, connected, and cascading extremes (drought–flood sequences, heatwave clusters, storm surges, etc.)
• Spatial and temporal clustering of events using object-tracking or spell detection
• EVT-aware downscaling, bias correction, and model evaluation
Bridging EVT with services
• Synthetic event generation and scenario design for stress-testing systems
• Tail-focused calibration/validation and skill scores
• Uncertainty quantification relevant to hazards, risk, and exposure
Additional topics of interest include, but are not limited to:
• EVT-based calibration and validation of extremes
• Tail-focused verification methods (extremal scoring rules, return-level skill, reliability in the tails)
• Applications to extremes in reanalyses, climate models, and hydrology services
We particularly encourage contributions that bridge EVT with climate, hydrology, and infrastructure risk applications, including decision-relevant uncertainty quantification and studies of compound or cascading extremes. Submissions may include new methodological developments, open datasets and tools, or real-world case studies.