THE WEATHER MODELS , A BASIC EDUCATION ON USE AND THEIR BIAS
Good morning everyone!!
Wow , we have some great weather for most of the northern mid-atlantic / northeast this weekend.
However , quite honestly the synoptic pattern really does remind me of fall / winter.
It is quite unusual in the summer months to see a more organized / stratiform type of low pressure system.
Most of the precip in the summer months come from a convective nature (storms) type event.
However , we did see just this setup on the models this past week with an organized area of low pressure.
For every northeast weather enthusiast eyes are glued to weather models in the late fall / winter looking for storms along the east coast.
I will be doing a major blog in september for winter weather forecasting( sought of a winter forecasting handbook) , which originally would discuss models and there bias.
However , i will do a separate piece here ( and refer back to this w/ the handbook) and discuss the models.
MODELS ARE NOT CRYSTAL BALLS
Models DO NOT control the weather. they do not control or change the upper , mid or low atmospheric pattern.
The weather model is simply a tool. A very important tool that is the direct result of hundreds of extremely complicated mathematical equations.
Models have errors in certain atmospheric setups like thunderstorms , latent heat energy , and also the density of an air mass.
Any computer model or model output is only as accurate as the input. Also,some are more reliable than others at certain time frames , and should be used accordingly.
The smart meteorologist will not change his / her forecast based on changes in the GFS , or ECMWF UNLESS there is a reason in the atmospheric setup.
Perfect example of this was the Blizzard of december in 2010. If you remember 4 days out every model had an all go on a major storm to belt the northern mid atlantic.( which btw was day after xmas)
All of a sudden ,however , what happened?…2 days out models showed a complete total miss…ALL THE MODELS.
However , very important observation here…NOTHING CHANGED in the atmospheric setup!!!!
As it ended up turning out ( because you all know the outcome ;) the models were missing a very important piece of input.
Now lets go to 36 – 24 hr out , when new models came out wouldn’t you know the storm appeared again and showed 1-2 feet for nyc
IMPORTANT LESSON HERE , MODELS ARE ONLY AS GOOD AS THE INPUT.
LONG-MEDIUM RANGE MODELS
The gfs model is the model of the NCEP. important note here , the gfs usually has a cold bias that produces a storm track to the SE of eventual track in the 3 – 7 day range.
It is notorious for having a strong cold bias in the winter , which would lead to a suppressed southeast storm track. this bias becomes clear if the air mass approaching the area is colder on guidance vs actual obsverations.
Important clue here is if the model is colder than its ensemble (GEFS). The GEFS is the third best model i use for medium guidance
Ahhhh the european , some call it the king , this is really because of a greater resolution and data input.
It is IMHO the 2nd most reliable model in the medium range. However , one major bias here is the handling of disturbances coming out of the SW USA. (usually depict them slower than actual speed)
My model of choice for the medium / long range. ( tends to minor out the errors typically seen on ECMWF)
The canadian guidance tends to favor to want to over amplify low pressure , and has a warm bias due to poor equations it has for ocean / air mass interaction.
Use this guidance to look over the overall pattern , certainly not any specifics of storm track or intensity.
The UKMET is actually a pretty reliable model when it is in agreement with the ECMWF. I like to use the UKMET as a cross check to the ECMWF as both models use very similar basic equations.
The NAM is a mesoscale model and short range model. There are two types WRF & NMM , that use slightly different equations for convection.
Word of advice here is to use this to only 60 hrs out as the resolution sharply falls off after. For all you severe wx geeks the NAM WRF/NMM is a good model
for picking up on mesoscale banding setups and atmospheric instability.
Another great model for mesoscale banding / lifting up to 48 hr
See above for MM5 , also use to detect possible heavy precipation. use up to 12 hr out
BOTTOM LINE HERE , take the model guidance and compare it to observations!!! , how else would you know what the volatility may be in the guidance?
Some examples here , how could guidance show low cutting into lakes w/ high pressure parked in quebec?
Bulls eye all of a sudden just magically appears on precip output?
Models are not crystal balls but can be very useful when they are used the right way.
Stay tuned for a very extensive winter weather handbook coming in sept , questions/comments you can get me at SteveGwx ( twitter)
NOW STOP READING BLOGS AND GET OUT AND ENJOY THIS BEAUTIFUL AUGUST WEATHER IN THE NORTHEAST!!!!!!!
Rain is looking more imminent to start next week off.
Meteorologist Steve Garry