Hurricane Helper is the data processor for The Wall Street Journal's Hurricane Tracker. This script is an opinionated parser of National Hurricane Center storm data, written in Python. For every named storm tracked by the center in the Atlantic and Eastern Pacific oceans, Hurricane Helper creates a GeoJSON FeatureCollection and saves it as a .json file in geojson/ with the name of the storm. The FeatureCollection contains:
- A polygon feature of the forecast cone with the following properties:
- storm The storm name in title case
- cat The storm intensity
- fcstpd The forecast period in hours (always "120", 5 days)
- Linestring features representing the historical and forecast track segments, each with the following properties, attributable to the first point of the linestring:
- storm The storm name in title case
- cat The storm intensity
- datetime ISO 8601 datetime in UTC/GMT,
- wind wind speed in miles per hour,
- pressure pressure in millibars (or
nullifsourceisforecast), - source "historical" or "forecast"
- current boolean true for the first forecast point, listed as "current center location" by the NHC
- Point features, in oldest-to-newest order, each with the same properties as the linestring segments.
All storms are also added to one big FeatureCollection and saved to geojson/currentGeoJSON.json.
Since the goal of this project is to show all current hurricanes, if you wish to display a single storm, you can filter on the storm name property.
For polygons, we use the category provided by the NHC:
- HU Hurricane
- MH Major hurricane (Saffir-Simpson scale category 3 and above)
For points and linestrings, we replace HU or MH with H1-H5 depending on the Saffir-Simpson scale.
For less intense wind speed measurements, we've recently observed the following categories:
- TS Tropical storm
- STS Subtropical storm
- STD Subtropical depression
- DB Tropical disturbance
- LO Remnant low
Other categories may be reported.
- Forecasts are produced in "local" time, while historical positions are recorded in UTC/GMT. This standardizes all times to UTC/GMT.
- Small, unimportant tropical disturbances are included in the official RSS feed. This outputs only named storms.
- Sometimes the forecast cones are not in the shapefiles. This suppresses the output until all features are present for each storm.
- Wind speeds are reported in knots. This converts to miles per hour using the correct precision.
- Hurricane numbers aren't specified. This puts each hurricane on the 1-5 scale.
- Nonexistent values are given as
-9999.0. This changes those toNone(Python)/null(JavaScript). - Storms disappear after they are no longer tracked. This saves the last data for each storm to a file for that storm.
The last historical point and the first forecast point have the same datetime but different locations. This may be because the first forecast point contains some uncertainty.
sh setup.sh to create a Python virtual environment and install requirements
sh updateData.sh provides a shell script template for processing the files with a cron job.