Degree

Doctor of Philosophy (PhD)

Department

Geography and Anthropology

Document Type

Dissertation

Abstract

The dissertation focuses on western region of Southwest Pacific Ocean (SWPO)

basin (135E - 180, and 5S - 35S) tropical cyclone (TC) climatology using observed

and modeled data. The classification-based machine learning approach

identifies the synoptic geophysical and aerosol environment favorable or unfavorable

for TC intensification and intensity change prior to landfall incorporating

observational and satellite data. A multiple poisson regression model with varying

temporal monthly lags was used to build a relationship between the number of

monthly TC days with basin wide average dust aerosol optical depth (AOD), sea

surface temperature (SST), and upper ocean temperature (UOT). This idea was

expanded by building a prediction model of TC count to see how changes in one

unit of dust, SST, and UOT can contribute to changes in monthly TC days. A

decision tree and random forest classifier was used to discriminate tropical depression

(TD) and tropical storm (TS) events and examined their classification ability

of unseen data. The goal was to derive a robust model that can balance correct

and incorrect classifications and provide higher prediction accuracy. Classification

decisions are determined by training selected classifiers using variables assigned

to hundreds of storm event samples and identified the most influencing predictors

in the classification decision. Mean composite maps of the most important geophysicaland aerosol variables during classification decisions using each set of TD

and TS case was developed to facilitate geophysical comparisons during different

environments. The spatiotemporal climatology of influential variables is important

to better understand the TC climatology. The study used hexagonal tessellation

and geographically weighted regression to spatially model the TC minimum central

pressure, SST, 1000 mb relative humidity, and sea salt AOD relationship to better

understand the spatio-temporal TC climatology over the space. This research further

developed a classification and prediction model for whether a TC will intensity

or weaken just before making landfall using a random forest classifier, geophysical,

and aerosol data including physical observations of TCs 24-hours prior to landfall.

Initial intensity, sea skin temperature (SkT), and longitude identified as the most

important variables for the classification decision for the mainland and island landfall

cases. The predicted intensity prior to landfall should lead to a higher success

rate of informed decisions along the coast which will alleviate coastal Australia

and SWPO islands TC related risk.

Committee Chair

Trepanier, Jill

DOI

10.31390/gradschool_dissertations.5310

Download

Share

COinS