Tropycal

Tropycal is a Python package intended to simplify the process of retrieving and analyzing tropical cyclone data, both for past storms and in real time, and is geared towards the research and operational meteorology sectors.

Tropycal can read in HURDAT2 and IBTrACS reanalysis data and operational National Hurricane Center (NHC) Best Track data and conform them to the same format, which can be used to perform climatological, seasonal and individual storm analyses. For each individual storm, operational NHC and model forecasts, aircraft reconnaissance data, rainfall data, and any associated tornado activity can be retrieved and plotted.

The latest version of Tropycal is v1.2.1.

Installation

Conda

The currently recommended method of installation is via conda:

conda install -c conda-forge tropycal

Pip

Installation is also available via pip:

pip install tropycal

From source

Tropycal can also be installed from source by cloning the GitHub repository:

git clone https://github.com/tropycal/tropycal
cd tropycal
python setup.py install

Dependencies

matplotlib >= 2.2.2
numpy >= 1.14.3
scipy >= 1.1.0
pandas >= 1.3.0
xarray >= 0.10.7
networkx >= 2.0.0
requests >= 2.22.0
pyshp >= 2.1

To fully leverage tropycal's plotting capabilities, it is strongly recommended to have cartopy >= 0.17.0 installed.

Documentation

For full documentation and examples, please refer to Tropycal Documentation.

As of v0.3, the documentation is up-to-date following a bug that started with v0.2.5 where the documentation was not updated with each release.

Sample Usage

As an example, read in the North Atlantic HURDAT2 reanalysis dataset, excluding Best Track (current year's storms):

import tropycal.tracks as tracks

basin = tracks.TrackDataset(basin='north_atlantic')

Individual Storm Analysis

Individual storms can be retrieved from the dataset by calling the "get_storm" function, which returns an instance of a Storm object. This can be done by either entering a tuple containing the storm name and year, or by the standard tropical cyclone ID (e.g., AL012019).

Let's retrieve an instance of Hurricane Michael from 2018:

storm = basin.get_storm(('michael',2018))

This instance of Storm contains several methods that return the storm data back in different data types. The following examples will show how to retrieve 3 different data types.

Retrieve Michael's data in different data formats:

storm.to_dict()
storm.to_xarray()
storm.to_dataframe()

Visualize Michael's observed track with the plot function:

Note that you can pass various arguments to the plot function, such as customizing the map and track aspects. The only cartopy projection currently offered is PlateCarree. Read through the documentation for more customization options.

storm.plot()

If this storm was ever in NHC's area of responsibility, you can retrieve operational forecast data for this event provided it is available. Forecast discussions date back to 1992, and forecast tracks date back to 1950.

Retrieve a single forecast discussion for Michael - both of these methods will yield an identical result:

#Method 1: Specify date closest to desired discussion
disco = storm.get_nhc_discussion(forecast=dt.datetime(2018,10,7,0))
print(disco['text'])

#Method 2: Specify forecast discussion ID
disco = storm.get_nhc_discussion(forecast=2)
print(disco['text'])

NHC also archives forecast tracks, albeit in a different format than the official advisory data, so the operational forecast IDs here differ from the discussion IDs. As such, the forecast cone is not directly retrieved from NHC, but is generated using an algorithm that yields a cone closely resembling the official NHC cone.

Let's plot Michael's second forecast cone:

storm.plot_nhc_forecast(forecast=2)

Now let's look at the 12th forecast for Michael.

Note that the observed track here differs from the HURDAT2 track plotted previously! This is because this plot displays the operationally analyzed location and intensity, rather than the post-storm analysis data. This is done to account for differences between HURDAT2 and operational data.