Read article on doing a similar project

[alyoshenka]

• [x] Examining the operational use of avalanche problems with decision trees and model-generated weather and snowpack variables
• [x] Jason's article

[alyoshenka]

Article 1

Notes

• "hourly weather data were aggregated into summary statistics covering the 24 and 72 h periods prior to the hazard assessment. Previous studies have consistently found weather conditions over these time periods to be strong predictors of avalanche hazard"
• "To mimic existing forecaster practices, three interfaces were identified in the upper snowpack to produce variables relevant to surface avalanche problem types: a 24 h interface, a 72 h interface, and a storm interface (Fig. 2). The 24 and 72 h interfaces were identified by searching for all snow deposited in the past 24 and 72 h. The storm interface was identified by selecting all snow deposited since the last full day without snow, similarly to how accumulated snowfall is measured in the field on storm boards that are cleared in between stormy periods"
• "sensitivity studies have shown physical snowpack models to be most sensitive to precipitation input"
• Try with decision tree: "Decision trees are a class of machine learning methods that provide simple and interpretable visualizations of complex non-linear relationships. Unlike other machine learning methods that focus on predictive performance (e.g. neutral networks, random forests), decision trees present relationships in ways that more closely resemble human decision-making processes and are thus a helpful tool to understand the avalanche problem identification and assessment process."
  • See section 2.4 of first article for analysis of DS data structures
• look into variable selection, section 2.4.3: "The weather and snowpack variables included in our dataset exhibit natural correlations, which can negatively affect statistical analyses and make the interpretation of decision trees more challenging"
• "When only considering weather and snowpack variables, the most significant variable for wind slab avalanche problems was the maximum air temperature over the past 24 h"
• "The variables that influenced adding a new dry loose problem included skier penetration depth, the maximum air temperature over the past 72 h, and the size any of surface hoar on the surface."

  Metrics to Add

• max air temp past 24 hrs
• max air temp past 72 hrs
• previous day's danger rating
• total snowfall past 24 hrs
• total snowfall past 72 hrs
• maximum hourly wind speed past 24 hrs
• threshold amount: 7cm over past 72 hrs
• date????
  • categorical question

    Article 2

    Notes

• "Utilizing daily winter data from 1995 to 2011, results show that using Bayesian tree analysis outperforms traditional statistical methods in terms of realized misclassification costs that take into consideration asymmetric losses arising from two types of error."
• "one of the most dangerous highways in the world"
• make sure we're finding the highest elevation snotel/weather sites possible
• focuses more on deep slab avalanches as they are the ones that primarily affect road closure decisions
• Are we pigeon-holing ourselves by not including range bands in our Y variables?
• does not add snow layers metrics
• "Thus the variable RELDEN measures the ratio of the density of the snowfall on the most recent snow day to the density of the snowfall on the second-most recent snow day"
• lots of numbers I don't understand

  Metrics to Add

  In rough order of most used by forecasters for this article:

  1. Water content of new snow measured in mm
  2. Sum of maximum temperature on last three ski days, an indicator of a warm spell
  3. Difference in minimum temperature from previous day
• Interval stake: depth of snowfall in last 24 hours
• Weighted sum of snow fall in last 4 days: weights = (1.0, 0.75, 0.50, 0.25)
• Density of new snow, ratio of water content of new snow to new snow depth
• Relative density of new snow, ratio of density of new snow to density of previous storm
  • How to determine a storm with our data?
• Change in total snow depth relative to depth of snowfall in the last 24 hours
• Minimum temperature in last 24 hours
• Maximum temperature in last 24 hours
• Wind speed, mph at peak location
• Storm stake: depth of new snow in previous storm