The Problem of Snow Density


In order to convert precip to new snow depth, the simplest solution if to use the ’ten to one’ rule; ten inches of snow per tenth inch of precip, regardless of temperature. Not very realistic.


Another solution is to use the National Weather Service New Snowfall to Estimated Meltwater Conversion table. For every tenth inch precip, the table estimates the new snowfall as follows:

34 to 28 deg. – 1.0 inches

27 to 20 deg. – 1.5 inches

19 to 15 deg. – 2.0 inches

14 to 10 deg. – 3.0 inches



This may be a better solution, but it requires a lot of work to implement in the model, and it’s still incremental. Also, since temps on the valley floor are often near freezing, estimating an inch of new snow per inch at a temp of 34 degrees will lead to significant errors.


The model currently uses a power function times a multiplier with easily adjustable constants that have global effect. The constants are tweaked based on field info. Here is an example based on the  settings from January 2007. I’ve since modified them slightly.

Based on the NWS table, the function graphed above may be underestimating snow depth at lower temperatures, but that is not supported by field data. Easy to fix in any case!


As another approach, I compared recent NWAC Stevens Pass telemetry data for snow depth and precip vs. temperature, and found little correlation. So it looks like one way or another I’ll have to use some sort of empirical  formula.