What does "Intitialze" mean?

[Stephanie]

I've seen it in plenty of posts before and just read it in a discussion about the GFS not being able to "initialize" a hurricane but forecasts it 2 weeks out.

Thanks!

[Stormchaser16]

Just means that when the model is run, it doesnt pick up on certain information that is actually there, it may not initialize it in the correct spot, or it may not initialize another atmospheric system. Someone else could prolly explain it better, but in easy terms it just means that when the model was run, something was not picked up on.

[Stephanie]

Thanks Stormchaser!

[MWatkins]

Stormchaser16 more or less hits it right on. Initilization is the set of data the model uses to calculate what happens next. All of the global observations get loaded into the model to initialize/start the runs.

As mentioned in the other thread, the problem with the GFS is it does a really good job of spinning up phantom storms...but cant pick up on a monster Cat 5 hurricane when it is initialized (like what happened with Isabel last year).

MW

[GalvestonDuck]

No need for your full signature...just the first letter of your first and last name will do.

~GD

[Stephanie]

Stormchaser16 more or less hits it right on. Initilization is the set of data the model uses to calculate what happens next. All of the global observations get loaded into the model to initialize/start the runs.

As mentioned in the other thread, the problem with the GFS is it does a really good job of spinning up phantom storms...but cant pick up on a monster Cat 5 hurricane when it is initialized (like what happened with Isabel last year).

MW

I remember seeing this over the past few winters as well with potential "Superstorms" and huge Nor'easters that we barely get a flake out of.

[MWatkins]

No need for your full signature...just the first letter of your first and last name will do.

~GD

1. Whoops...I didn't even notice those were back...sorry about that. These things are new to me...

2. Hey...I just noticed I can uncheck that.

Sorry about that.

MW

[GalvestonDuck]

No need for your full signature...just the first letter of your first and last name will do.

~GD

1. Whoops...I didn't even notice those were back...sorry about that. These things are new to me...

2. Hey...I just noticed I can uncheck that.

Sorry about that.

MW

I can tell Mike's had a busy couple of weeks.

Mike...I was kidding by defining "initialize" in non-met terms. Re-read and you'll get it.

*pats Mike on the back*

[Air Force Met]

Good examples of bad initializations would be these:

1) When Bonnie had a closed low of 1008mb and 40 kt winds, the model had it as an open wave with 15 kt winds

2) It rarely had Charley as a closed low early on.

3) Say a storm has a 995mb pressure. If the GFS has it as 1008 on the model run, that is bad initialization.

4) Another Example: Say a front at 12Z is sitting over SE Kansas through NE MO - E. Michigan...but the 00hr forecast on the 12Z run has it sitting back in Nebraska, then that is bad initialization.

It can initialize bad on anything...pressure...winds...temps...locations. When we are talking about hurricanes it is important. Why? Because if the intial position of the storm is initialized in the wrong spot...say 90 miles to far north...then that effects the outcome of where the storm is forecasted to go by that model because 90 miles away means different steering flows...etc. If the intensity is wrong...then the model may not forecat it moving with the right flow. For example...if the GFS has a system as an open wave...it will move it with the low level flow and the wave may actually disappear of the model run, when in actuality it will be steered with mid-level flow and will remain or strengthen.

[PurdueWx80]

There are a couple of other reasons why the GFS does so poorly at initialization:

1. Initialization isn't just made from pure observations. It is a mix of observation plus some sort of 'average' from the previous runs of the model. That helps to throw out spurious data, but it also can end up missing some very important data.

2. The resolution of the GFS is not that high, meaning that the grid points are spaced fairly far apart. Bonnie was a TINY vortex in the middle of the GULF, and if it weren't for the recon, none of us would have seen it. Small features are not picked up by the GFS, and won't be until it can run with higher resolution.

3. Data is still pretty sparse over the oceans - in fact it comes from satellites, buoys, ships and airplanes only. Satellites are still no where near perfect at picking up on simple things that may be quite important for physical or thermodynamic processes to occur.

[Air Force Met]

There are a couple of other reasons why the GFS does so poorly at initialization:

1. Initialization isn't just made from pure observations. It is a mix of observation plus some sort of 'average' from the previous runs of the model. That helps to throw out spurious data, but it also can end up missing some very important data.

Yep...and that problem can cause it to ride a forecast down to oblivion like Slim Pickens did the A-Bomb in Dr. Stranglove. Once it gets a funk...it sometimes would rather choke on it than cough it out.

[Stephanie]

Thanks for the examples Air Force Met!

[Aslkahuna]

Initialization is where the model computes the initial conditions at Time 0 of the run based upon the input of the initial conditions observed at that time and the data valid for that time which is input into the model run. That being said, it is important for the forecaster to review the initialization to make sure that the features shown in the actual data are accurately depicted since otherwise the model is not likely to forecast them well.

Steve

[Stormsfury]

Also to ADD to poor initializations are the fact that out in open waters, there's a lack of data sources as well ..

SF

[USAwx1]

Also to ADD to poor initializations are the fact that out in open waters, there's a lack of data sources as well ..

SF

yes that is also an issue