A Multivariate Framework for Hurricane Forecast Assessment

• Activity Number: 87
• Type: Invited
• Date/Time: Sunday, August 3, 2014 : 8:30 PM to 10:30 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Zachary Weller*+ and Jennifer A. Hoeting
• Companies: and Colorado State University
• Keywords: forecasting ; multivariate ; nonparametric ; hurricanes ; climate ; STATMOS

Abstract

Due to the large social and economic impacts of tropical cyclones (TCs), accurately forecasting these weather phenomenon is of great interest to weather and climate scientists. TC characteristics that are often of interest and produced by different models include forecasted tracks (location), maximum wind speed, radii of wind speeds, minimum sea level pressure, rainfall, and storm surge. Traditional tropical cyclone forecast assessment techniques focus on univariate analyses of these characteristics separately. In this work, we develop and explore multivariate methods for hurricane forecast assessment via permutation tests. The methods developed will assist climate scientists in the improvement of hurricane forecasting models and could be extended to assess other multivariate forecasting systems.