Tropical Storm Erika

[Amped]

ECMWF still shows 1006mb over the Bahamas at 96hrs, it bombs out after that. 4 days is a pretty long wait before the storm begins to deepen, wouldn't be surprised if it dissipated and reformed.

[PSUBlizzicane2007]

Quite an eastward shift in the 18Z models.

Keep in mind a lot of those are based on GFS initialization, so it's not terribly a surprise that many shifted east with the GFS finally trying to come in line with the Euro

[jojo762]

ECMWF still shows 1006mb over the Bahamas at 96hrs, it bombs out after that. 4 days is a pretty long wait before the storm begins to deepen, wouldn't be surprised if it dissipated and reformed.

Quite possible that it could dissipate in the next 24-48 hours. Extremely dry air, and strong shear await it near HSP tomorrow and Friday... And if it still has a very messy structure like we see now (and its expected to have), it probably will dissipate.

[PSUBlizzicane2007]

AMS journals are already in ruins due to similar behavior, no reason to ruin AmericanWX too.

I was going to stay out of this, but I just can't now. As someone who worked with AMS journals extensively and has worked with even the Director of Publications himself, this is a completely baseless insult to hurl at AMS journals which, along with RMetS journals, are the premiere sources of meteorological advancements and research. I personally tend to find degreed meteorologists who hurl insults at them to be ones who were either rejected by them or who couldn't handle peer review critiques of their work to the point of a tantrum throwing two year old.

Of course the AMS doesn't need your personal approval, just like the NWS doesn't need Kevin Martin's forecast seal of approval either.

[turtlehurricane]

TH, I think you are taking operational models and acting like we are using them verbatim. The op runs are valuable in showing trends and hinting at size or intensity. An op run continuously showing a large and intense TC probably means that conditions may be favorable in the model for intensification. Opposite is also true. They model the global patterns as you stated and therefore inherently should at least try to model the steering pattern of the storm. There is value in the op runs and the stats do show this as DTK pointed out. Of course I would consult TC models and not just globals, but I think you are incorrectly dismissing their value.

I've said multiple times in this thread that global models are very useful for determining large scale steering patterns, they are an essential tool no doubt. However the actual projection of a tropical cyclone in a global model is fairly dismal. I think the false alarm rate for RI is so high that the magnitude of the low in a global model projection shouldn't really be considered. Unfortunately the GFDL and HWRF have even worse false RI rates. Likewise there are numerous times where globals show a weak cyclone or open wave and it ends up being a hurricane.

 

We can all agree intensity forecasting for TCs is terrible, and has shown almost no improvement for the 2 decades I've been watching. If you can't forecast intensity then you can't forecast the TCs effects on the upper-levels, since the strength of a TC's upper-level anti-cyclone and Rossby Wave is directly connected to intensity. This results in innacurate representation of steering currents, and this effect butterflies out. Beyond 3 days track forecasting is quite poor, I think we're at the point where we can't produce better track forecasts until we solve the intensity problem. All models have this problem, but globals are particularly bad since they cant resolve the effects on steering. At least the GFDL and HWRF have enough resolution to produce the realistic upper-level features, if they can get the intensity right. Observations is another problem that feeds back into this, and might be the key to unlocking it.

 

The only time globals are quite useful is in the case of a large, mature hurricane. With something like Erika the results are abysmal.

[CoastalWx]

DTK may be able to post some evidence, but the GFDL and HWRF have also had their share of issues, so I'm not sure how they would be much better. You pointed out an issue earlier with the GFDL. Like everything else, it's guidance.

[LocoAko]

If you want a real discussion then present your arguments, so far your only argument is globals can forecast QPF without microphysics.... which is a terrible example since globals are very inaccurate for qpf due to lack of microphysics.

What do you mean lack of, or without, microphysics?

I simply repeated your insult back to you, I was calling the GFS garbage for tropical cyclone forecasting and you took it personally. If you don't like being insulted then maybe you shouldn't insult others.

 

I presented my brief arguments: "global models can barely resolve tropical cyclones, and there's not enough data to initialize with anyways. The equations aren't particularly optmal for tropical forecasting either, they are designed for mid-latitue weather. Higher resolution models with equations that are tuned for hurricanes are a much better option, although we still got a long way to go even with those." Prove that any of those statements are wrong instead of throwing red herrings. I could go deeper into it, there are dozens of things in global models that are detrimental to tropical cyclone forecasting, which is the reason hurricane models are being produced.

I kind of get what you're trying to say, but dtk has posted (albeit not data) that the GFS has some of the best verification scores for TC tracks. How do you respond to that?

 

A major problem with the atmospheric science field is people that act like you massively slow down scientific progress by nitpicking ideas and derailing them, while presenting none of their own. There is almost zero progress in government laboratories and even UW-Madison since scientists from other institutions torpedo ideas, so it takes years to publish and then other people have already published it, which is quite inefficient since it wastes money to have many institutions working on the same thing.

So scientists from other institutions "torpedo" ideas, and then other people publish it instead? I don't even understand what you're trying to accuse the field of with comments like this. Why wouldn't everyone get torpedoed? Why do other scientists' opinions matter before trying to publish?

 

There is almost always a confrontational stance between meteorologists, where if someone disagrees with something they take it personally and then it becomes a pride thing. And then scientists complain when all their funding gets cut.... that's what happens when you produce nothing and spend all your time ruining other people's projects.

 

This is mostly unrelated to you, this is not a serious scientific debate obviously. What I'm saying is the last thing we need is people demanding hard empirical evidence on a casual discussion board, when the subject being discussed is not fully understood.

I find the debate about the utility of global models interesting, especially in a scientific way.

 

AMS journals are already in ruins due to similar behavior, no reason to ruin AmericanWX too.

:eyeroll:

[dtk]

If you want a real discussion then present your arguments, so far your only argument is globals can forecast QPF without microphysics.... which is a terrible example since globals are very inaccurate for qpf due to lack of microphysics.

 

I simply repeated your insult back to you, I was calling the GFS garbage for tropical cyclone forecasting and you took it personally. If you don't like being insulted then maybe you shouldn't insult others.

 

I presented my brief arguments: "global models can barely resolve tropical cyclones, and there's not enough data to initialize with anyways. The equations aren't particularly optmal for tropical forecasting either, they are designed for mid-latitue weather. Higher resolution models with equations that are tuned for hurricanes are a much better option, although we still got a long way to go even with those." Prove that any of those statements are wrong instead of throwing red herrings. I could go deeper into it, there are dozens of things in global models that are detrimental to tropical cyclone forecasting, which is the reason hurricane models are being produced.

 

A major problem with the atmospheric science field is people that act like you massively slow down scientific progress by nitpicking ideas and derailing them, while presenting none of their own. There is almost zero progress in government laboratories and even UW-Madison since scientists from other institutions torpedo ideas, so it takes years to publish and then other people have already published it, which is quite inefficient since it wastes money to have many institutions working on the same thing.

 

There is almost always a confrontational stance between meteorologists, where if someone disagrees with something they take it personally and then it becomes a pride thing. And then scientists complain when all their funding gets cut.... that's what happens when you produce nothing and spend all your time ruining other people's projects.

 

This is mostly unrelated to you, this is not a serious scientific debate obviously. What I'm saying is the last thing we need is people demanding hard empirical evidence on a casual discussion board, when the subject being discussed is not fully understood.

 

AMS journals are already in ruins due to similar behavior, no reason to ruin AmericanWX too.

I'm at a workshop right now so I can't dive into this entirely.  However, I'll concede that my example with QPF was probably a terrible one.  Secondly, I apologize for the personal insult and saying this was out of your league.

 

However, I will not back down from arguing against your original comment that "all globals are garbage for TC forecasting".  We have strong, quantitative and anecdotal evidence to the contrary.

 

The fact of the matter is that global models still produce lower track errors than the hurricane specific models, much of which is most likely related to data assimilation (a topic in which I happen to know something about).  

 

Global models are not "designed for mid-latitude weather", that's nuts.  They are designed to represent all kind of atmospheric (global) phenomena.  In terms of the "GFS is garbage thing"...I'm not the GFS and I don't take it personally.  It's just that it is hard to take anyone seriously that makes such claims.  I already apologized for my previous comment and will choose to move on. 

 

For what it's worth, I spend no amount of time "trying to ruin other's projects".  In fact, I go out of my way to try and help people succeed, sometimes to the detriment of my own work.  I think your generalizations about the science community is a bit overstated, though I think I know where you are coming from as there is competitiveness when it comes to funding opportunities and having to sell ones self.

[CoastalWx]

Perhaps it's just how you use the guidance. The NHC references the globals many times...I guess it's just how you view the models and utilize it. I've had plenty of experience using them to my advantage. I don't take their solutions 100% verbatim. 

[dtk]

Speaking of NWP...I noticed there is a parallel run of the GFS labeled as "PR4DEV" showing up on the verification charts (and I might add it's doing quite well). The website says this is a 4D-ENKF/VAR experimental version. Anyone have the low-down on it? dtk?

 

By the way, the 00Z run on 8/25 showed Erika intensifying and swinging east past Florida for what it's worth.

This is a near real-time prototype of the hybrid 4D EnVar system that I have been working on with collaborators for several years.  This is still an EMC-maintained experiment, but I believe the plan is for an implementation sometime early next year.  The package includes several other DA related changes including:  expansion to include cloudy/precip affected MW radiances, variational bias correction for aircraft temperatures, and a few things related to the EnKF design and execution.

[CoastalWx]

I'm at a workshop right now so I can't dive into this entirely.  However, I'll concede that my example with QPF was probably a terrible one.  Secondly, I apologize for the personal insult and saying this was out of your league.

 

However, I will not back down from arguing against your original comment that "all globals are garbage for TC forecasting".  We have strong, quantitative and anecdotal evidence to the contrary.

 

The fact of the matter is that global models still produce lower track errors than the hurricane specific models, much of which is most likely related to data assimilation (a topic in which I happen to know something about).  

 

Global models are not "designed for mid-latitude weather", that's nuts.  They are designed to represent all kind of atmospheric (global) phenomena.  In terms of the "GFS is garbage thing"...I'm not the GFS and I don't take it personally.  It's just that it is hard to take anyone seriously that makes such claims.  I already apologized for my previous comment and will choose to move on. 

 

For what it's worth, I spend no amount of time "trying to ruin other's projects".  In fact, I go out of my way to try and help people succeed, sometimes to the detriment of my own work.  I think your generalizations about the science community is a bit overstated, though I think I know where you are coming from as there is competitiveness when it comes to funding opportunities and having to sell ones self.

That's my line of thinking as well.  Again, maybe it's how people interpret solutions.

[CoastalWx]

Show the stats for tropical cyclones broken down by category, I already know the stats for weak and disorganized TCs is terrible. The only thing bringing up the numbers is large hurricanes like Isabel, at that point globals can be decently accurate.

 

In the case of Erika and any other disorganized system the verification scores are gonna be terrible. Case in point, GFS forecast from a mere 54 hours ago showed an open wave at this point.

 

Well just a quick example from what I am saying. The tropical models, including GFDL and HWRF were pretty darn strong this morning. Others were as well. The GFS as we all know was pretty weak and a mess. 12z run finally started to develop it a bit. the euro has been altering between a weak storm and a strong storm. Globals may indicate (esp GFS) that Erika may struggle with shear for a bit and land interaction. Sort of a yellow flag. NHC official call is on the lower side of intensity guidance and states their reason. Again, the idea is guidance is struggling a bit and the globals IMO are not inspiring a lot of confidence to go with the higher end intensity that some of those hurricane models suggest. Perhaps it does strengthen rapidly near the Bahamas, but I don't see a big reason to go nuts with intensity yet, as the next two days are pretty important in terms of track and strength. 

[heavy_wx]

This plot is from the NHC's 2014 Verification report.

 

 

The GFS and Euro had higher skill in forecasting track than the high resolution models (e.g. HWRF, GHM/GFDL).

[turtlehurricane]

I'm at a workshop right now so I can't dive into this entirely.  However, I'll concede that my example with QPF was probably a terrible one.  Secondly, I apologize for the personal insult and saying this was out of your league.

 

However, I will not back down from arguing against your original comment that "all globals are garbage for TC forecasting".  We have strong, quantitative and anecdotal evidence to the contrary.

 

The fact of the matter is that global models still produce lower track errors than the hurricane specific models, much of which is most likely related to data assimilation (a topic in which I happen to know something about).  

 

Global models are not "designed for mid-latitude weather", that's nuts.  They are designed to represent all kind of atmospheric (global) phenomena.  In terms of the "GFS is garbage thing"...I'm not the GFS and I don't take it personally.  It's just that it is hard to take anyone seriously that makes such claims.  I already apologized for my previous comment and will choose to move on. 

 

For what it's worth, I spend no amount of time "trying to ruin other's projects".  In fact, I go out of my way to try and help people succeed, sometimes to the detriment of my own work.  I think your generalizations about the science community is a bit overstated, though I think I know where you are coming from as there is competitiveness when it comes to funding opportunities and having to sell ones self.

Show the stats for tropical cyclones broken down by category, I already know the stats for weak and disorganized TCs is terrible. The only thing bringing up the numbers is large hurricanes like Isabel, at that point globals can be decently accurate. Also break them down by RMW radius, storms like Charlie are handled poorly. Also when citing stats the dataset needs to be defined for the stats to have any meaning.

 

It's actually a testament to our lack of understanding that global models on average produce better track forecasts statistically. Zooming in on a hurricane's structure amplifies the number of unknowns the model has to deal with, more assumptions and estimates are made, and every single mistake versus reality can ruin the whole forecast. This is why when we zoom out to a crude representation of a hurricane in global models there are better results on average. We have been locked in a stalemate where more resolution and data degrades the forecast, because there are fundamental problems with our understanding of hurricane structure and the equations which drive it.

 

When it comes to Erika the only thing of real use of globals is the representation of large scale steering patterns, which is just about enough to make an accurate forecast when dealing with a weak TC, even if the projection in the model is inaccurate.

 

What's more useful: A projection that's more accurate for the wrong reasons, or a projection that's less accurate for the right reasons?

[AvantHiatus]

18z GFS coming in to settle the score. This one is going east of FL and strengthening viciously north of PR.

[downeastnc]

Little flare up just east of the center might have some legs.......

[AvantHiatus]

[IsentropicLift]

GFS looks GGEM esque

[IntenseBlizzard2014]

Trying to think, was Katrina really the last Hurricane to landfall in SE Florida? 10 years ago.

Does Hurricane Fay from 2008 count?

[AvantHiatus]

GFS looks GGEM esque

Exactly.

 

[AvantHiatus]

My god...It's a good thing Coastal SC/GA is one of the most isolated sections of coast in the country.

 

[andyhb]

My god...It's a good thing Coastal SC/GA is one of the most isolated sections of coast in the country.

 

It's also 6 days out.

[AvantHiatus]

It's also 6 days out.

Why wouldn't it happen tho? I could see an argument for further east tho. This region of the Atlantic ocean has some of the most potent ocean heat potential in the world

[andyhb]

Why wouldn't it happen tho? I could see an argument for further east tho. This region of the Atlantic ocean has some of the most potent ocean heat potential in the world

 

Calling for the doom of some area along the EC when the storm is essentially a naked swirl right now is alarmism at its finest.

 

Just because an area has high OHC doesn't suddenly mean run for the hills. You're ignoring a massive number of variables.

[UnionWeatherWx]

Calling for the doom of some area along the EC when the storm is essentially a naked swirl right now is alarmism at its finest.

You act like we've never seen this kind of situation before. It's a model run, we are going to analyze it.

[Stebo]

Why wouldn't it happen tho? I could see an argument for further east tho. This region of the Atlantic ocean has some of the most potent ocean heat potential in the world

Many reasons already mentioned in this thread. Might be best if you read more in the thread instead of posting any thought that comes to mind.

[AvantHiatus]

Many reasons already mentioned in this thread. Might be best if you read more in the thread instead of posting any thought that comes to mind.

Fair enough, one must acknowledge that these scenarios are possible and take the appropriate actions. Don't mess with this stuff my man. Bitcoin bets aside, i'll take the over.

 

http://www.tropicaltidbits.com/analysis/models/gfs/2015082618/gfs_pres_wind_atl_28.png

[andyhb]

You act like we've never seen this kind of situation before. It's a model run, we are going to analyze it.

 

Well yeah posts like the one made by AvantHiatus aka Weatherguy#s there make it really seem like it!

 

Impressive run though.

[Stebo]

Fair enough, one must acknowledge that these scenarios are possible and take the appropriate actions. Don't mess with this stuff my man. Bitcoin bets aside, i'll take the over.

 

http://www.tropicaltidbits.com/analysis/models/gfs/2015082618/gfs_pres_wind_atl_28.png

Possible sure but there is a multitude of other solutions including this running into the islands that could happen.

[AvantHiatus]

Possible sure but there is a multitude of other solutions including this running into the islands that could happen.

No model support thus it is more unlikely than the above scenario. Unless you subscribe to the TurtleH philosophy.