These notebooks work through scraping avalanche.ca for historical avalanche rating data by region and elevation, and then run some analyses to identify patterns that emerge.
- web_scrape_avalanche.ipynb: Scrapes avalanche.ca and structures data by region for analysis
- investigate.ipynb: Runs analysis on the data
- region_boundaries.ipynb: Defines the avalanche region boudaries for use in investigate.ipynb
The code is set up to scrape and analyze regions across Western Canada with a sufficient time series of ratings.
Above-treeline rating time series can be clustered to see which regions are most similar at these elevations, with interesting regional patterns emerging.
Each region can be characterized according to its mean avalanche rating, variance of rating through time, and gradient of rating across elevation (e.g. how much higher is the above-treeline rating vs the below-treeline rating?). For example the lower mainland has a relatively low rating on average, but with high variance in time and across elevation (e.g. above-treeline ratings are much higher than below-treeline ratings).
In the space of rating gradient across elevation vs mean rating, interesting patterns appear! Note the high-mean-high-gradient northwest coast, high-mean-low-gradient cariboo/columbia/purcells.
Outside of Python 3.6 and some standard scientific python packages (e.g. numpy, pandas, scikit-learn), you will need Chromedriver installed for webscraping.