RABTracks

Author: Stella Bourdin (Formerly Uni. Oxford, now LMD-IPSL, Ecole Polytechnique), with help from Leo Saffin and Kevin Hodges (Uni. Reading), and Yushan Han (UCDavis), and advice from many members of the Huracán project.

This code and data are distributed under the GPL-3.0 license. This does not cover the IBTrACs data, whose license is not indicated on the website.

A paper is being written to be submitted to ESSD describing the construction process of the present database.

Obtaining the RABTracks dataset

The RABTracks dataset is publicly available through this link. Download the rabtracks.zip archive.

The archive contains one file per basin. Each file has been compressed using the LZMA algorithm. They can be decompressed using the unxz or xz --decompress command.

Data description

Coordinates:

• source is the origin of the track data, either IBTrACS, or a (re)-analysis.
• record is an index for each TC point across all sources.

Data variables:

• track_id is the unique TC identifier from IBTrACS.
• time is the time of the corresponding record.
• lon and lat describe the position of the point in track_id at time for each source. lon_pres & lat_pres and lon_wind & lat_wind describe respectively the position of the pressure minimum and wind maximum associated with the TC center.
• pres and wind10 describe the intensity (in terms of minimum SLP and maximum surface wind speed, respectively) of the point in track_id at time for each source. NB : this variable is empty for the IBTrACS source, where there are several wind attributes, and it is up to the user to decide which one they may use.
• vo850 describes the intensity in terms of relative vorticity at 850hPa.
• translation_speed and azimuth have been computed along the tracks; they are in m/s and degrees, respectively. The *_roll counterparts have been computed applying a 24h rolling mean to the lon and lat data.
• radius_max_wind is the radius of maximum wind, computed as the distance between the pressure minimum and the wind maximum, where these position were available (for TRACK data). The radius_max_wind_roll counterpart as been smoothed with a 24h rolling mean.
• lpsarea is the Low Pressure System area as computed by SyCLoPS.
• vtu, vtl and b are the Cyclone Phase Space (CPS, Hart 2003) parameters. They are currently only provided for some of the TRACK sources.
• vtu_roll, vtl_roll and b_roll are the 24h rolling means of the CPS parameters.
• WCSI means Warm Core Symmetric Intensifying. It is True when the cyclone has a Symmetric Warm Core as per the CPS parameters (|B|<15 & VTL>0), and when it is intensifying at the same time (intensification being defined from the 850hPa vorticity)
• CCA means Cold Core Assymetric, and is True when the cyclone has an Assymetric Cold Core as per the CPS parameters (|B|>15 & VTL<0).
• label is the SyCLoPS (adjusted) label for Low Pressure System category. It is only provided for SyCLoPS sources.
• is_tc is a composite flag for identifying the TC stage. For IBTrACS, it is True when nature=='TS', for SyCLoPS, it is True when short_label=='TC', and for TRACK it is True when WCSI is True (see ESSD paper for more details).
• is_etc is a composite flag for identifying the TC stage. For IBTrACS, it is True when nature=='ET', for SyCLoPS, it is True when short_label=='EX', and for TRACK it is True when CCA is True (see ESSD paper for more details).
• ET flags the onset and offset of Extra-Tropical Transition. ET=+1 corresponds to the onset of transition, and corresponds to the last timestep when is_tc is True for a given track_id and source. ET=-1 corresponds to the offset of transition, and corresponds to the first timestep when is_etc is True for a given track_id and source (see ESSD paper for more details).
• All other attributes along the dimension record only come from IBTrACS, so please refer to the IBTrACS Column Documentation.

Examples of how to read/use

You can check the notebooks in paper_figs to see the code used to generate the figures for the ESSD paper.

Building RABTracks yourself

If you want to run the code yourself, (1) clone this repository, and then (2) download the input data: it is available through the same link, but you will need to download the input.zip archive.

Decompress the zip file into your RABTracks folder, and decompress all the compressed .xz files using the unxz or xz --decompress command. If necessary, move the ibtracs and RA_tracks folders to the first level, and create the RABTracks folder to store the output. Your folder structure should be similar to the one below. From there, you can run 2_build_RABTracks.ipynb (and modify it if you like) to build the dataset. Note that you might need to install a python environment with the necessary packages.

.
├── ibtracs -> Contains the input files from IBTrACS
│   ├── ibtracs_EP.pkl
│   ├── ibtracs_EP.pkl.xz
│   ├── ibtracs_NA.pkl
│   ├── ibtracs_NA.pkl.xz
│   ├── ...
├── RA_tracks -> Contains the input files from the different reanalyses/trackers
│   ├── SyCLoPS-ERA5.pkl
│   ├── SyCLoPS-ERA5.pkl.xz
│   ├── ...
│   ├── TRACK_CPS-ERA5.pkl
│   ├── TRACK_CPS-ERA5.pkl.xz
│   ├── ...
│   ├── TRACK-ECMWF-OP-AN.pkl
│   ├── TRACK-ECMWF-OP-AN.pkl.xz
│   ├── ...
│   ├── TRACK_WCSI-ERA5.pkl
│   ├── TRACK_WCSI-ERA5.pkl.xz
│   ├── ...
├── RABTracks -> Will store the output
│   ├── Create this folder if necessary!
├── paper_figs -> Script to create the figures (examples of how to read/use the data)
│   ├── ...
├── LICENSE
├── rabtracks.py
├── 1_input_data.ipynb -> Was used to create the input from raw data (not published)
├── 2_build_RABTracks.ipynb -> Use to create RABTracks from the input
├── 2_update_public.txt
└── README.md

References

• SyCLoPS: Han & Ullrich, JGR: The System for Classification of Low-Pressure Systems (SyCLoPS): An All-In-One Objective Framework for Large-Scale Data Sets