Reducing the Dynamic State Index to its main information using Principal Component Analysis

TL;DR

This paper investigates whether the Dynamic State Index (DSI), a key atmospheric indicator, can be simplified using Principal Component Analysis without losing essential information, and how this reduction varies with spatial scale.

Contribution

The study demonstrates that the DSI can be effectively reduced to three key terms using PCA, simplifying calculations while retaining main information.

Findings

Three of six DSI terms are sufficient for accurate calculations

Reduction depends on spatial scale

PCA effectively identifies main information in DSI

Abstract

The Dynamic State Index is a scalar quantity designed to identify atmospheric developments such as fronts, hurricanes or specific weather pattern. The DSI is defined as Jacobian-determinant of three constitutive quantities that characterize three-dimensional fluid flows: the Bernoulli stream function, the potential vorticity (PV) and the potential temperature. Here, we tackle the questions (i) if the mathematical formulation of the DSI can be reduced, while keeping the main information, and (ii) does the reduction of the DSI depend on the spatial scale? Applying principle component analysis we find that three of six DSI terms that sum up to the Jacobi-determinant are sufficient for future DSI calculations.