Gridded Logit 50% Values

[marqh]

A metadata encoding for the result of a Gridded Logit 50% Values process

• Processing steps:
  1. Begin with a collection of METAR reports and identify an interesting characteristic that is boolean (e.g., “ceiling height” less than or equal to 1000 feet or SN/SG/SNSH/SGSH occurs in present weather).
  2. Define appropriate points in time and/or spans of time.
  3. Define start and end dates for a “cool season” or “warm sea
  4. Select a “predictor” weather element. (E.g. 850-hPa temperature is generally considered to be a good predictor of the snow boolean.)
  5. Gather a set of predictors and associated booleans for a METAR site over a sample and fit a Logit distribution to them.
  6. At each METAR site, find the predictor value where the Logit distribution = 0.50.
  7. Analyze these predictor values to a grid.
  8. For a collection of stations, interpolate the gridded values to yield a value at each station.

[marqh]

http://www.uncertml.org/statistics/discrete-probability
looks appropriate for viii