Avalanche Forecasting in the Spearhead Range

Eventually, I would like for this project to be a self-contained Python script which can be run in a short time every morning, and automatically produce maps for a given day's forecast. Given the time constraints of the project and of my own knowledge of Python, combined with the difficulty of parsing a convoluted XML file into useable data, this project will be done "by hand" in ArcMap for a given day as a proof of concept.

 

In the methodology of this proof of concept, we will begin by looking at the available data. The bases of this project are the CAC avalanche forecast and topography. The CAC forecast is available as an XML file, so a future self-contained version of this project will have to use Python to parse the pertinent avalanche size, slope angle, and likelihood information. Topographical information is available in the form of Digital Elevation Models from the Canadian government via GeoGratis. The project also utilizes shapefiles containing roads, streams, rivers, lakes, glaciers, and mountain peaks from GeoGratis.

 

To begin, this project produces a base map on which the hazard rasters will be displayed. This base map is a hillshaded DEM, overlaid with roads, lakes, rivers, and streams. The base map also includes points to represent glaciers and peaks, as reference points for map users. An automated version of this project could simply rely on a single base map file rather than stepping through the process of creating the base map each time.

 

The second component of this project is the actual analysis portion. It is essentially a multi-criteria analysis, considering slope angle, elevation relative to treeline, and the CAC's forecasted avalanche size, avalanche likelihood, and slope aspect for each avalanche type listed in the forecast. (Storm slabs, wind slabs, and deep persistent slabs). 

 

I found that considering avalanche size and likelihood as independent variables often produced a map with an unreasonably high danger rating. I considered several options to combat this problem, and decided that the simplest way would be to combine expected avalanche size and likelihood into one single danger rating. In the eventual self-contained version of this project, this would be carried out within the parsing script, by taking the average of the two numbers, and rounding up to eliminate decimals. Although this would generalize the analysis, it eliminate falsely high danger ratings cases which classify as "low probability, high consequence," and vice versa while continuing to provide accurate ratings when both the probability and consequences of avalanches are high or low.

 

The theoretical parsing script outputs an attribute table containing a danger rating for wind slabs, storm slabs, and persistent slabs, as well as the slope aspects affected by each of those avalanche types. From there, an ArcGIS model, built using ModelBuilder, takes over.

 

For each avalanche type, the danger rating is assigned to the aspects for which it is valid through a table join. It is then reclassified and combined in a weighted sum with slope, where the slopes more likely to avalanche (centered around 38 degrees) are given a higher weight. That output raster is then combined again with elevation relative to treeline, where Alpine terrain is more dangerous than treeline and below-treeline terrain, to produce the final hazard raster for that avalanche type.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

To allow map users to visualize each avalanche type in isolation, each of these hazard rasters is published overlaid on the previously made base map. Although avalanche types are not independent of one another, this will help map users understand where the final combined map comes from.

 

Finally, the three avalanche types are put together in a weighted sum and combined with the base map to provide a general avalanche hazard map that takes into account wind slabs, storm slabs, and persistent slabs.

 

Each map is produced with a conventional, easy to understand color scheme, with a green-yellow-red scale to indicate avalanche hazard.