News
By Joel Gratz, Founding Meteorologist Posted 12 years ago March 28, 2012
DIY Meteorology, Part 2
As I wrote about yesterday, there was some serious uncertainty between the models concerning the forecast for the Sunday storm.
Today, the picture is a little clearer. There will be much colder air (back to normal temperatures) moving into the northern half of Colorado on Sunday, with some snow.
Here's the model comparison for the forecasted weather at 6pm on Sunday. This forecast was made Tuesday night (a 5-day forecast), and the blue line shows the leading edge of the colder air. If the blue line is further south, the storm is stronger.
But how do we choose between models? The European is usually the best and shows more cold air and storminess than the American GFS model for this storm.
One way to extend our analysis is to see which forecast model has been most consistent in its forecast. Consistency doesn't always correlate to accuracy, but if a forecast model is consistent, it gives me more confidence in its forecast. Think of it like an NBA player. You never know when they'll have a great shooting game and when they'll miss most of their shots -- life can be random sometimes. But if you pick the player that has the most consistent performance night after night, chances are you'll have an easier time predicting how well they'll do over the next few games.
So, how consistent are the models? The graphic below shows the 5-day forecast made on Tuesday night (thick blue line) and the 6-day forecast made on Monday night, 24 hours earlier (lighter blue dashed line). The closer these lines are together, the more consistent the forecast. The further they are apart, the more "all over the place" the model is and the less I trust it.
Well, sure enough the European model is the most consistent. The worst model might be the American, as it forecasted a strong, cold storm yesterday and today's forecast now shows a much weaker storm.
The summary? I think the storm will come in like the European shows, with a decent amount of cold air for northern Colorado on Sunday and perhaps a few inches of snow.
Today, the picture is a little clearer. There will be much colder air (back to normal temperatures) moving into the northern half of Colorado on Sunday, with some snow.
Here's the model comparison for the forecasted weather at 6pm on Sunday. This forecast was made Tuesday night (a 5-day forecast), and the blue line shows the leading edge of the colder air. If the blue line is further south, the storm is stronger.
But how do we choose between models? The European is usually the best and shows more cold air and storminess than the American GFS model for this storm.
One way to extend our analysis is to see which forecast model has been most consistent in its forecast. Consistency doesn't always correlate to accuracy, but if a forecast model is consistent, it gives me more confidence in its forecast. Think of it like an NBA player. You never know when they'll have a great shooting game and when they'll miss most of their shots -- life can be random sometimes. But if you pick the player that has the most consistent performance night after night, chances are you'll have an easier time predicting how well they'll do over the next few games.
So, how consistent are the models? The graphic below shows the 5-day forecast made on Tuesday night (thick blue line) and the 6-day forecast made on Monday night, 24 hours earlier (lighter blue dashed line). The closer these lines are together, the more consistent the forecast. The further they are apart, the more "all over the place" the model is and the less I trust it.
Well, sure enough the European model is the most consistent. The worst model might be the American, as it forecasted a strong, cold storm yesterday and today's forecast now shows a much weaker storm.
The summary? I think the storm will come in like the European shows, with a decent amount of cold air for northern Colorado on Sunday and perhaps a few inches of snow.