Revised Snow Algorithm


Mar 1, 2004
Lansing, MI
Dan Cobb from NWS GRR has revised his snow accumulation algorithm (a very old version is in Bufkit as it's snow table) and I've incorporated that into the web feed... (and sub in the desired location into output, so Lansing MI would be )

1) Ptype logic has been adjusted to better detect mixed precip.
2) Output Ptypes of SNZRPL and SNRAPL are now possible.
3) Default sleet ratio has been reduced from 3:1 to 2:1 and default FZRA ratio has been reduced from 1.09:1 to 1.05:1. This should result in more conservative sleet and ice accumulations.
4) Now provide separate accumulations of sleet (ice pellets) and freezing rain.
5) Accumulation of sleet or snow is not allowed above a temp of 32.9F.
6) Output has been reorganized into an hourly group and storm-total group. I have added 6 hour (12 hour for GFS) separators for ease of reading the table.
7) Fixed bug that resulted in crashes when using GFS bufkit files. The issue was due to a slight variation in GFS bufkit file formats generated locally versus Penn State website.

The critical temperature for refreezing rain drops is -6C. The critical temperature for ice in cloud remains at -8C with an RH threshold of 85%.
Accumulation of sleet or snow is not allowed above a temp of 32.9F

I must ask, why is accumulation of sleet and snow not allowed over a temperature of 32.9? In some wet snow scenarios, heavy snowfall accumulations can result up to right around 36 degrees on grassy and above ground locations. Would not it be better to create an algorithm that incorporates ground temperature, precipitation intensity, and precipitation type?
How many observation sites report ground / soil temperature? The Oklahoma mesonet sites do (a few with obs at many different depths), but ASOS sites don't. It'd be nice to have ground / soil temperature reading for this, though.

I do agree that snow intensity rates are important... As we know, snow accumulation can occur with >33F 2m temperatures if it falls heavily enough. In this case, though, this would be another error-prone variable that may cause more harm than good.
Ground temps are not reported in the BUFKIT file, so there's no way they can add that in. It's not too often that you get snow accumulating with mid-30's. If it's a 'special' event, you'll still see the QPF and the ratio and precip type so you can do the math yourself, but overall less snow sticks when the temp is 37 vs 27 :>

Also remember what does the public care about? If you tell them 6" of snow is coming, but they wake up and roads are wet with the only snow on the grass -- it's all BUST. They want to know what impacts them, and roads are MUCH higher on the impact scale than my backyard.
At this point in time, due to lack of obs leading to a general lack of knowledge I do not feel that any threshold should be set for accumulating snow. I've seen snow in northern Minnesota stick around 42F simply because it was the end of January and the frost line in the soil extended down 5+ ft at that point. The ground was well below freezing but the air was not, snow still accumulated. It did melt shortly after but it DID accumulate while it was snowing. I think at this point in time educated instincts and observational experience are the best tools for determining whether snow will accumulate. I'm sure someone could come up with an empirical formula if the observations were there. It would likely be a function of air temperature, ground temperature & snowfall rate (to be more precise... it would have to be evaluated for several cm below the sfc due to conduction.) This whole idea of accumulating snow involves physics we do not fully understand yet. I can guarantee that if the ground temp and air temp is below freezing then snow will accumulate. :) If someone can hash out the money for more soil temp observations (the more the better) then an empirical algorithm should be doable.
Thanks for the bump. Somehow I missed it last time. Just in time for the upcoming storm that, per tonight's model runs anyway, looks like a snow bust for STL, despite some media excitement this evening. (Doesn't take much to get that here.)