Evolution Characteristics of Ensemble Member Forecasts Through Model-Based Clustering

• Activity Number: 633
• Type: Contributed
• Date/Time: Thursday, August 4, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Prabhani Kuruppumullage Don*+ and Francesca Chiaromonte and Jenni Evans
• Companies: Penn State University and Penn State University and
• Keywords: Model based clustering ; Cyclones

Abstract

Tropical and extratropical cyclones provide considerable risks to mankind; especially the populations near the coasts, and have been studied since at least the time of Kublai Khan. The use of operational numerical models has greatly improved accuracy in forecasting their evolution. To investigate the structure and dynamic behavior of Typhoon Sinlaku (2008), we apply model-based point and path clustering methodologies to the European Centre for Medium-Range Weather Forecasts (ECMWF) 51-member ensemble forecasts. In particular, we focus on path clustering of the evolution of Typhoon Sinlaku in two spaces: the physical space (2-dimensional space spanned by latitude and longitude) and the CPS space (3-dimensional space that defines the evolving structure of a storm). A regression mixture framework is used for clustering, where units are curves (paths) in the physical or CPS space. Traditional and resampling based diagnostics point to a mixture of five cubic polynomial components. Path clustering in physical space creates a meaningful partition of ensemble members related to the evolution of the storm. Interpretations are less straightforward but still informative in the CPS space.