Earthquake Tweets

Note: The design of this repo is to be easily replicable to other collections local to Epic-Analytics

Process:

1. Run epic_utils/parallel_get/get_geo.rb for the earthquake collection, like this:

   ruby get_geo "Earthquake" 24

   This will create 24 .json files of the form tweets_<process>.json.

   Next, put these files together with: cat *.json > ../../earthquakes/raw-geo-earthquake-tweets.jsonl

2. Use the GeoTagged Earthquake Tweet Processor.ipynb Jupyter Notebook to process this massive file of tweets. Most importantly, this Notebook generates the csv required for the timeline (d3)

3. Now use tippecanoe to tile objects:

    tippecanoe -Pf -o epic-earthquake-tweets.mbtiles earthquake-tweets.geojsonl
   

   With some trial and error, this one seems to work okay for the full 2M+ file:

    tippecanoe -Pf -Z1 -z16 -B5 -m4 -as -o earthquake-tweets.mbtiles earthquake-tweets.geojsonl
   

4. Upload the tileset to Mapbox:

    ~/upload-tiles.js earthquake-tweets.mbtiles
   

5. The docs folder in this repository contains the webpage to view these tweets interactively.

TODO:

1. Link extraction
2. Other filters
3. Timeline of tweets currently in-view.
4. Better coloring/sizing and data-driven styling to use more visual channels
5. A different basemap that doesn't wash out at large scale.