Jeff asked a question in his post about the value of a forecast past 48 hours. There are always changes so is it really relevant to use a 7 day forecast. Some meteorologists would say that extended forecasts are garbage and we should not forecast past 5 days because days 5 through 7 are always changing. My response is several fold but my first reaction is that the 7 day is useful past 48 hours since at least it gives people an idea of what the weather is going to be like in the future. Large weather patterns, i.e. big storms can be seen and that is useful.
So let's get to some numbers. According to National Weather Service statistics the 'skill score' for the seven day forecast without human input, (That's just numbers from the raw computer model) is 60%. So that in itself is useful.
Here's the problem for the meteorologist. Once we get past 48 hours the data is no longer layered and gridded and really turns into a wave chart. Many times we are talking about a weather system which hasn't even made land-fall in the U.S. So my other argument is that the medium range forecasting depends on the skill of the forecaster. I will tell you that my friends at the National Weather Service often times do not spend a great deal of time on medium range forecasts and sometimes do not use all model data available to them. Part of this is political and the other part an issue of time. I use a European solution which is run in England. I also run a Canadian model called the GEM. Many times during the winter the European model sees cold air days before the American model. In fact, this is going to be the case next week. The American model is not bringing Arctic air into the picture and the cold air it brings in is way too warm in my opinion. By Monday afternoon the European has the -15 Celsius line across northern Indiana. That tells me that we could see single digit lows Monday night and highs in the teens or 20s on Tuesday.
So just to summarize. My opinion is that the medium range models are useful and a big factor in the forecasts you see on TV but is all depends on the skill of the forecaster.
Subscribe to:
Post Comments (Atom)
2 comments:
Thanks for this interesting post Greg! So often you hear forecasters talk about "the models," with very little explanation. When I visit the NWS site, I'll read their forecast discussion, and although sometimes it's hard to decipher (being in weather-person-speak), it will often talk about one model over another. And it's interesting to note which model is right!
I'm glad I could enlighten you a bit. Truth is that models and model development is a continuing learning process for me. Sometimes you need a scorecard to keep up with all the mathematical changes in various models.
I've doing this for about 22 years now and the changes have monumental to say the least. Stuff that we thought was science
Post a Comment