View larger image

Missed Forecast

23 Sep 2015 03:56 AM
The showers and thunderstorms that pushed into western Iowa yesterday and made it to Des Moines was a bit unexpected. One of the main high resolution and short term forecast models run by NCEP/NWS is called HRRR. This model is run every hour producing a 15 hour forecast. The featured map presents the percentage of these runs (15 in total) that had simulated reflectivity above 30 dBz (stronger showers) for 6 PM yesterday evening. The inset map shows the actual RADAR data at that time. Only the last few model runs that afternoon picked up on the showers in west central Iowa and the coverage was under predicted.

Voting:
Good: 25
Bad: 8
Abstain: 5

Tags:   hrrr   forecast  

View larger image

Bad Forecast?

14 Nov 2013 05:37 AM
A prominent meteorological forecast model, North American Mesoscale (NAM), run by the National Centers for Environmental Prediction produced what would appear to be a very bad forecast for near surface air temperature on Wednesday. The featured map shows the forecasted temperature bias for noon with many sites in the upper midwest around 10 degrees too cold. On the surface, this would appear to be a bad forecast, but there was a reason why the temperatures were too cold. The shaded areas on the map show analyzed snow depth that the model was provided to start the forecast. Of course, these snow depths are way off with much of Iowa shown in the 1-3 inch range, whereas reality was just a trace, if that. So the worst temperature predictions are shown over areas with incorrectly analyzed snow depth. This begs the academic question of if this was a bad forecast or an example of garbage in - garbage out?

Voting:
Good: 32
Bad: 2
Abstain: 4

Tags:   model   forecast  

View larger image

100% Chance

19 Feb 2013 05:41 AM
Those of you that fervently monitor the NWS forecast may have noticed something interesting this past Sunday. For some locations in the state, the NWS had a near 100% chance of precipitation forecasted for Thursday. Is it common to see a 100% chance forecast at four days out? The featured chart shows the frequency of this confident forecast based on the afternoon zone forecast for Polk County (Des Moines). Based on data back to 1 Jan 2009, this would appear to be the first time it has happened for the day 4 forecast for Polk County. The chart nicely shows that forecasters get more confident of predicting a 100% chance as the event gets closer. Today's forecast for Thursday remains at 100%!

Voting:
Good: 35
Bad: 2

Tags:   zfp   forecast   nws  

View larger image

Model Consistency

18 Feb 2013 05:43 AM
There is currently a lot of excitement over the upcoming winter storm forecasted for Thursday into Friday. The featured chart looks at forecasted precipitation totals for Des Moines from the GFS model. The chart shows the current storm along with six recent heavy snowfall events. Each bar represents one of the four model runs made per day prior to the event. The general lack of consistency creates a creditably issue for forecasters. The top series shows one model run dumping a huge amount of precipitation and then backing off on the total up until the event. The fifth and sixth chart show going from having no precip to a storm. The caveat not illustrated here is the storm track, by which small displacements can make a huge difference in forecasted precipitation for a fixed location.

Voting:
Good: 37
Bad: 3

Tags:   model   forecast