Verification of heavy snow events in South Dakota

Forums (Forecast Builder)

RSS (Opens New Window)

Phil Schumacher, modified 8 Years ago.

Verification of heavy snow events in South Dakota

Youngling Posts: 11 Join Date: 9/24/12 Recent Posts

Hi there!

Several of my forecasters have commented how the SuperBlend snowfall (and QPF) were too low for our winter storms this year and felt that WPC may have provided a better first guess. So I decided to go through all the 6" snow events that impacted South Dakota and nearby areas in adjacent states and compare WPC to SuperBlend for storm total snowfall to see if this was true. In addition to calculating MAE and Bias for all stations and all events, I stratified the verification by observed snowfall amount - < 1", 1-4", 4-6", and > 6". What I found is that SuperBlend had a significant low bias on snowfall of greater than 6" and a high bias for snowfall less than 1" compared to WPC. This bias was generally worse at 60 h before snowfall began but was evident even within 24 h of the event beginning. I have attached a PowerPoint which compares WPC, SuperBlend, and NDFD (I used ISC for NDFD).

If you have any questions, please contact me at anytime.

Phil Schumacher

SOO, FSD

Jeremy Martin, modified 8 Years ago.

RE: Verification of heavy snow events in South Dakota

Youngling Posts: 14 Join Date: 1/9/14 Recent Posts

Thank you for the study! Have had so little snow down here it is been hard to get a feel for how well Superblend was working. Will make sure the CRGMAT sees this during our meeting this week. I did have one question, did you look at the QPF verification for the same events? If so, were the biases similar? Just thinking out loud to see if the problem is with the base QPF values or the snow ratios that are being determined.

-Jeremy

Phil Schumacher, modified 8 Years ago.

RE: Verification of heavy snow events in South Dakota

Youngling Posts: 11 Join Date: 9/24/12 Recent Posts

Hi Jeremy,

I did save off the QPF from these events and could make the comparison. The reason I have not done so yet is because I have low confidence in observational (COOP) QPE. I did a study last year looking at snow-liquid ratios in the FSD CWA and there is no spatial consistency - one can have 10:1 next to a station reporting 25:1. So I have been waiting for a few widespread rain events to do a similar study for QPF. However, if the CRGMAT feels it would be worthwhile, I could go through and do a QPF comparison for the same five events I did the snowfall study.

Phil