Statistical and Model-Driven Analyses of North Atlantic Tropical Cyclone Activity
Open Access
- Author:
- Kozar, Michael Eric
- Graduate Program:
- Meteorology
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- April 11, 2011
- Committee Members:
- Michael Mann, Thesis Advisor/Co-Advisor
Michael Mann, Thesis Advisor/Co-Advisor - Keywords:
- Climate
Atlantic
Tropical Cyclones
Hurricanes
Landfall - Abstract:
- Tropical cyclones (TCs) are among the most destructive and deadly natural phenomena on the planet. Using the historical Atlantic TC record, this thesis first examines the empirical relationships between climate state variables and Atlantic tropical cyclone (TC) counts. State variables considered as predictors include indices of the El Niño/Southern Oscillation (ENSO) and Northern Atlantic Oscillation (NAO), and both “local” and “relative” measures of main development region (MDR) sea surface temperature (SST). In addition, the considered predictors include indices measuring the so-called “Atlantic Meridional Mode” (AMM) and the West African monsoon. Using these predictors in forward stepwise Poisson regression, this thesis examines the relationships between TC counts and climate state variables. As a further extension on past studies, both basin-wide named storm counts and cluster analysis time series representing distinct flavors of TCs, are also modeled. A wide variety of cross validation metrics reveal that total TC counts may be more skillfully modeled than the cluster series, and the most skillful models most commonly share three predictors: the MDR SST index, an index of ENSO, and the NAO index. The observed record of Atlantic TCs is relatively short however, and is subject to potential biases owing to lack of observation platforms such as aircraft reconnaissance and satellite imagery in earlier decades. Studies of long-term trends in TC activity are thus hindered by the limitations and uncertainty within the historical data. Therefore, this thesis also examines TC activity over a longer time frame using results from a long-term simulation of the NCAR CSM 1.4 coupled ocean-atmosphere climate model. The model has been forced with estimated natural and anthropogenic factors over the past millennium. Atmospheric variables from the model simulation are used to drive a recently developed downscaling relationship that simulates TC genesis and tracking over the course of the 1000-year simulation. This downscaling process generates a realistic long-term TC track dataset over an extended period of time, free of the observational record's many restrictions and biases. The realistic track data was used to perform an objective analysis of long-term trends in Atlantic TCs and TC landfalls and the potential underlying climate drivers. Analysis of TC event time series reveal that counts of landfalling TCs and even landfalling hurricanes (i.e. the subset of relatively strong TCs) track relatively well with the total basin-wide TC activity on multidecadal and longer timescales. Statistical models driven with relevant climate predictors derived from the model fail to explain as much variance as those which have been developed and applied to modern historical TC counts, but they do demonstrate significant statistical skill over the long-term.