TTM-preprocessing

Tools for producing geodata from travel time matrix text / csv files. See this backend for serving the data.

Overview

Features

• Generate travel time isochrone polygons in geojson format from TTM .txt / .csv files and YKR grid geodata
• Produce a minimal geojson file of the YKR grid
• Simplify geodataframes and geojsons, gzip everything -> lighter computation, minimal file sizes

How the isochrones are made

isochrones are produced for each YKR grid cell / travel mode combination. The data is classified into isochrones based on breakpoints (values of travel time in minutes). For example, if using breakpoints of [15, 30, 60], the result geojsons will have three (multi)polygons:

• Area with traveltime <= 15
• Area with 15 < traveltime <= 30
• Area with 30 < traveltime <= 60

File sizes

File sizes are kept as small as possible by:

At the geodataframe level:

• Removing the unneeded information (such as extra columns, crs info)
• Simplifying the needed information (such as coordinate precision)

At the file level:

• Removing all whitespace from the geojson objects written by geopandas
• Gzipping everything

You should expect one processed matrix (all isochrones for all travel modes) to take up about 2.5GB of disk space. This will of course vary based on the config used to generate the isochrones.

Tools and dependencies

Python

Make a virtual env:

python -m venv venv 

and activate it:

source venv/bin/activate

Install dependencies:

pip install -r requirements.txt

Bash / gzip

Bash and gzip are necessary for compressing the files. If you are on a unix-based system, you probably have both and everything should just work. If you are on windows, the easiest thing to do is probably to run a linux container. So, you'll need docker.

Input data

You'll need:

• All of the .txt / .csv files that make up a travel time matrix
• The corresponding YKR grid

The data is available for example here.

Processing the data

Catchmets

Optionally edit preprocessing/config.py to include the travel modes and breakpoints you wish to use.

Generate the isochrones with:

python preprocessing/isochrones.py \
  -g <directory/with/ykr-grid> \
  -m <directory/with/matrix/files> \
  -w <directory/to/write/isochrones/to> \
  -y <year/to/tag/output/files/with>

The amount of time this takes depends on your hardware, but be prepared for at least a few hours. Progress is logged to the terminal.

Grid

Generate the grid with:

python preprocessing/grid.py \
  -r <path/to/ykr-grid> \
  -w <directory/to/write/grid/to>

Compression

Unix-based:

Make the compress script executable and run it. Give the path to the directory containing the files you want to compress as the only argument.

./compress <path/to/data_directory>

Windows:

I do not know what the best way to do anything on windows would be, but this should get you a workable environment:

docker run -it --rm -v ./:/home ubuntu:latest
cd home

After that, follow the unix-based instructions.