Data and Code for: Persistent Multi-Scale Consistency in Best-Track Intensity Evolution and Rapid Intensification in Atlantic Tropical Cyclones (1851–2024)

This repository contains the data and code required to reproduce the persistence diagnostic described in the manuscript: "Persistent Multi-Scale Consistency in Best-Track Intensity Evolution and Rapid Intensification in Atlantic Tropical Cyclones (1851–2024)" The analysis is based on the Atlantic HURDAT2 best-track dataset (1851–2024), publicly available from the National Hurricane Center (https://www.nhc.noaa.gov/data/). The repository provides:- a processed storm time-series dataset derived from HURDAT2- a reference Python implementation of the persistence diagnostic- example datasets for validation and testing The persistence diagnostic is constructed using:- causal rolling means of maximum sustained wind speed (3, 5, and 9 advisory windows)- directional agreement across timescales- a forward-only activation rule based on sustained multi-scale consistency Storm-level persistence is defined as the total duration of active time steps and is used to evaluate statistical association with rapid intensification (RI). All computations are performed sequentially in time (forward-only), using only current and prior observations. The materials provided here are sufficient to reproduce all results presented in the manuscript from the public HURDAT2 dataset. This repository contains a minimal, transparent implementation of the diagnostic and does not rely on external data sources or unpublished components beyond those described in the paper.