Modeling Hurricanes using Principle Component Analysis in conjunction with Non-Response Analysis

TL;DR

This paper combines principal component analysis and non-response analysis to model hurricane precursors, revealing early indicators like pressure and temperature fluctuations that precede storm formation.

Contribution

It introduces a novel application of PCA and non-response analysis to identify early hurricane indicators and the conical relationships among variables.

Findings

Detection of pressure and temperature patterns days before storms

High correlation between storm conditions and buoy data pre-storm

Identification of conical relationships among variables

Abstract

This paper demonstrates how principle component analysis can be used to determine the distinct factors that house the terms that explain the variance among the co-dependent variables and how non-response analysis can be applied to model the non-functional relationship that exist in a dynamic system. Moreover, the analysis indicates that there are pumping actions or ebb and flow between the pressure and the water temperature readings near the surface of the water days before a tropical storm forms in the Atlantic Basin and that there is a high correlation between storm conditions and buoy conditions three-four days before a storm forms. Further analysis shows that that the relationship among the variables is conical.