douglasdaly/hurricanes

An analysis of hurricane data over time.

Hurricane Data Analysis

An analysis of historical tropical cyclone data from Weather Underground and global climate data from NASA and NOAA.

See the associated blog post for more.

---

Prerequisites

For the cartopy package you'll need to have installed the GEOS and PROJ libraries. On linux the packages are:

• libproj-dev
• proj-data
• proj-bin
• libgeos-dev

Setup

Configure your options in the Makefile, specifically the REQUIREMENTS_ENGINE variable (either pip or pipenv). Also, if your default system python is not Python 3 you'll want to change the PYTHON variable to match the command for Python 3 (likely python3 on *nix systems).

To install the requirements needed for the project:

$ make requirements

Data

Raw Data

Once you've configured your environment, you can download the raw data using:

$ make get_data

Though this will likely fail on the storm data (Weather Underground will probably cut you off after too many requests). To continue updating storm data, periodically run:

$ make continue_get_data

Until you see the output:

Pulling storm data... DONE

The NASA and NOAA data will download before the Weather Underground data and should not be a problem. The tool get_data is utilized for all of this. See that tool for further information.

Processed Data

Once you've downloaded the raw data files you can run:

$ make process_data

To process the raw data and save local copies of the (lightly) processed data to local .pkl files for further research. The tool process_data is utilized for this. See that tool for further information.

Feature Data

Once the data has been processed you can then generate some of the more interesting and further-processed feature data with the command:

$ make generate_features

The tool generate_features is utilized to do this. See that tool for further information.

Note: This could be time-consuming depending upon your system. There's an interpolation calculation on the NOAA temperature data which is fairly heavy computationally (the code is designed to take advantage of multi-core systems).

Media

Once you've run all the preceding commands in the Data section, you can generate various media files for the project:

$ make generate_media

The tool generate_media is utilized to do this. See that tool for further information.

Note: This may require additional dependencies (like pngcrush, optipng and zopfli) if certain compression features are utilized. You may also want to have ImageMagick installed as that can generate nicer GIF images.

License

This project's code is licensed under the MIT license. See the LICENSE file for more information.

Data for this project was acquired from the Weather Underground website, NASA and NOAA.