ATL: Ex-Tropical Storm GASTON - Discussion

[cycloneye]

8 PM TWO:

SHOWERS AND THUNDERSTORMS REMAIN DISORGANIZED IN ASSOCIATION WITH
THE REMNANTS OF GASTON LOCATED ABOUT 60 MILES SOUTH OF PONCE PUERTO
RICO. DEVELOPMENT...IF ANY...OF THIS SYSTEM IS EXPECTED TO BE SLOW
TO OCCUR AS IT MOVES WESTWARD AT 15 TO 20 MPH. THERE IS A LOW
CHANCE...20 PERCENT...OF THIS SYSTEM BECOMING A TROPICAL CYCLONE
AGAIN DURING THE NEXT 48 HOURS.

[Bailey1777]

michael if you had to plot a center and direction of movement what would it be?

[chrisnnavarre]

Based on the burst that maybe is over the center (who knows really with no visible now..) I'm thinking the center is around 17N 66W moving 275 at around 18mph. That would be my best estimate...also seems to be pulling a little north of west movement now....

[supercane]

It's the first year the NHC TWO works with a percentage of development within 48 hours.

It would be great to have a quality test on these percentages.

If they forecast a wave to become a TC within 48 hours and give it 70% and it doesn't become a TC within 48 hours they were 70% wrong and 30 % right.

If they forecast a wave to become a TC within 48 hours and give it 30% and it becomes a TC within this margin they were 70% wrong and 30% right.

Would be nice to have those figures. The NHC gave their TWO'' something quantitative. And quantity can be reality tested.

What's the percentage they were wrong and what's the percentage they were right?

How much they were wrong on X-Gaston?

So much wrong with this post that I had to respond. From NHC's 2009 verification report:

In 2007, the NHC began producing in-house (non-public) experimental probabilistic tropical cyclone genesis forecasts. Forecasters subjectively assigned a probability of genesis (0 to 100%, in 10% increments) to each area of disturbed weather described in the TWO, where the assigned probabilities represented the forecaster’s subjective determination of the chance of TC formation during the 48 h period following the nominal TWO issuance time. Verification was based on NHC besttrack data, with the time of genesis defined to be the first tropical cyclone point appearing in the best track. ...the experimental forecasts mostly exhibited a low bias(genesis occurred more often than forecast), ...Even so, the forecasters were clearly able to distinguish gradations in genesis likelihood (evidenced by the nearly monotonic increase of the verifying percentage with forecast percentage).

Figure 15a. Reliability diagram for experimental Atlantic probabilistic tropical
cyclogenesis forecasts for the period 2007-9. The solid blue line indicates the
relationship between the forecast and verifying genesis percentages, with perfect
reliability indicated by the thin diagonal black line. The dashed blue line indicates how
the forecasts were distributed among the possible forecast values.

Also the understanding of statistics in the above post is suspect. In an individual case, the storm either develops (100%) or not (0%); there is no 70% wrong/30% right. The NHC states that given a similar setup that a storm will be a TD or stronger X out of 10 cases.

BTW, RL3AO was compiling these stats for this season in this thread, although I haven't seen updates in a while.

[expat2carib]

It's the first year the NHC TWO works with a percentage of development within 48 hours.

It would be great to have a quality test on these percentages.

If they forecast a wave to become a TC within 48 hours and give it 70% and it doesn't become a TC within 48 hours they were 70% wrong and 30 % right.

If they forecast a wave to become a TC within 48 hours and give it 30% and it becomes a TC within this margin they were 70% wrong and 30% right.

Would be nice to have those figures. The NHC gave their TWO'' something quantitative. And quantity can be reality tested.

What's the percentage they were wrong and what's the percentage they were right?

How much they were wrong on X-Gaston?

So much wrong with this post that I had to respond. From NHC's 2009 verification report:

In 2007, the NHC began producing in-house (non-public) experimental probabilistic tropical cyclone genesis forecasts. Forecasters subjectively assigned a probability of genesis (0 to 100%, in 10% increments) to each area of disturbed weather described in the TWO, where the assigned probabilities represented the forecaster’s subjective determination of the chance of TC formation during the 48 h period following the nominal TWO issuance time. Verification was based on NHC besttrack data, with the time of genesis defined to be the first tropical cyclone point appearing in the best track. ...the experimental forecasts mostly exhibited a low bias(genesis occurred more often than forecast), ...Even so, the forecasters were clearly able to distinguish gradations in genesis likelihood (evidenced by the nearly monotonic increase of the verifying percentage with forecast percentage).
http://a.imageshack.us/img841/6184/atlb ... s20079.png
Figure 15a. Reliability diagram for experimental Atlantic probabilistic tropical
cyclogenesis forecasts for the period 2007-9. The solid blue line indicates the
relationship between the forecast and verifying genesis percentages, with perfect
reliability indicated by the thin diagonal black line. The dashed blue line indicates how
the forecasts were distributed among the possible forecast values.

Also the understanding of statistics in the above post is suspect. In an individual case, the storm either develops (100%) or not (0%); there is no 70% wrong/30% right. The NHC states that given a similar setup that a storm will be a TD or stronger X out of 10 cases.

BTW, RL3AO was compiling these stats for this season in this thread, although I haven't seen updates in a while.

Excuse my understanding of statistics. Why do the forecasters use percentage? Why don't the use the term will/will not develop within 48 hours? (as in your reasoning) Why don't they use in 7 out of 10 cases/similar setup cases we know of, these kind of storms develop into a TC within 48 hours?

I understand 70% chance of development as 70% possibility in this unique setup and 30% chance on non development in this unique setup.

The mistake is that they are using quantitative, subjective data in a qualitative context.

Just my .25 cent

[mvtrucking]

I think this one could be on its last gasp, Just not looking too good.

[AJC3]

Excuse my understanding of statistics. Why do the forecasters use percentage? Why don't the use the term will/will not develop within 48 hours? (as in your reasoning) Why don't they use in 7 out of 10 cases/similar setup cases we know of, these kind of storms develop into a TC within 48 hours?

I understand 70% chance of development as 70% possibility in this unique setup and 30% chance on non development in this unique setup.

The mistake is that they are using quantitative, subjective data in a qualitative context.

Just my .25 cent

Short answer - making a deterministic (single value - in this case "will develop within H48" or "will not develop within H48") rather than a probablistic forecast would fail to express the inherent uncertainity in TC-genesis forecasts. To put it bluntly, in most cases, forecasters just aren't good enough to make an accurate, straight-up "yes/no" forecast when it comes to whether or not a TC will develop from a potential area of genesis (of which there are many types!), especially early on.

Percentage type probablistic forecasts are generally easier for the general public to comprehend, compared to, say, the coverage (e.g. isolated, scattered, numerous, widespread) or uncertainty (e.g. slight chance, chance, likely) terms that you see in precipitation forecasts.

[expat2carib]

Excuse my understanding of statistics. Why do the forecasters use percentage? Why don't the use the term will/will not develop within 48 hours? (as in your reasoning) Why don't they use in 7 out of 10 cases/similar setup cases we know of, these kind of storms develop into a TC within 48 hours?

I understand 70% chance of development as 70% possibility in this unique setup and 30% chance on non development in this unique setup.

The mistake is that they are using quantitative, subjective data in a qualitative context.

Just my .25 cent

Short answer - making a deterministic (single value - in this case "will develop within H48" or "will not develop within H48") rather than a probablistic forecast would fail to express the inherent uncertainity in TC-genesis forecasts. To put it bluntly, in most cases, forecasters just aren't good enough to make a straight-up "yes/no" forecast when it comes to whether or not a TC will develop from a potential area of genesis (of which there are many types!), especially early on.

Percentage type probablistic forecasts are generally easier for the general public to comprehend, compared to, say, the coverage (e.g. isolated, scattered, numerous, widespread) or uncertainty (e.g. slight chance, chance, likely) terms that you see in precipitation forecasts.

OK! Thanks for your reply! So it's all about communication to the public in general.

Back on topic. Gaston is looking pathetic. Was a non event in the Lesser Antilles. On Dominica we had some torrential rain in an hour and some thunderstorms.

[supercane]

Excuse my understanding of statistics. Why do the forecasters use percentage? Why don't the use the term will/will not develop within 48 hours? (as in your reasoning) Why don't they use in 7 out of 10 cases/similar setup cases we know of, these kind of storms develop into a TC within 48 hours?

I understand 70% chance of development as 70% possibility in this unique setup and 30% chance on non development in this unique setup.

The mistake is that they are using quantitative, subjective data in a qualitative context.

Just my .25 cent

NHC could've decided to use a yes/no system as you describe, but there are problems. The big intrinsic one is that meteorology is not deterministic; that is, you cannot guarantee the same result given identical initial conditions. Given that reality, making a dichotomous yes/no determination is difficult, would give you less information, and hides the uncertainty surrounding cyclogenesis.

Your local weather forecast is replete with examples. Take, for instance, pop-up thunderstorms in the summer. I would be hard pressed to tell you any given day that it will definitely rain or definitely be dry. I could say "yes" or "no" depending on how sure I am, but that would be based on some probability threshold anyway (i.e., I say it will rain if I am 60% or more sure). Thus, while some lament seeing the 30% chance of thunderstorms every day during the summer, it is a true reflection of the uncertainty. The same thought process goes into the cyclogenesis percentages. Another example would be percentages in the medical literature. For someone diagnosed with cancer, it would be true that statistics could not predict whether or not he would ultimately survive, but it is hard to argue that there isn't a difference between a 5% and 90% 5-yr survival.

BTW, what the NHC expresses in their development probabilities is more remarkable than just yes/no. The graph above shows that if NHC states that a system has a 10% chance of development, that it would actually develop close to 10% of the time, and so on for each additional 10%. Outstanding!

PS, nothing prevents you from doing statistics on categorical vs pure quantitative data. Otherwise, Gregor Mendel couldn't have done his pea plant experiments (short v tall, curled v wrinkled, yellow v green, etc). (He may have fudged the results, but that's for another thread).

Edit, AJC3 beat me to the punch.

[cycloneye]

To reinforce what expat2carib said about the effects in Dominica, here in Puerto Rico,yes there were some flood advisories issued, but for the most part, nothing out of the ordinary occured.Even the winds were not over 30 mph.Is over here.

[expat2carib]

Excuse my understanding of statistics. Why do the forecasters use percentage? Why don't the use the term will/will not develop within 48 hours? (as in your reasoning) Why don't they use in 7 out of 10 cases/similar setup cases we know of, these kind of storms develop into a TC within 48 hours?

I understand 70% chance of development as 70% possibility in this unique setup and 30% chance on non development in this unique setup.

The mistake is that they are using quantitative, subjective data in a qualitative context.

Just my .25 cent

NHC could've decided to use a yes/no system as you describe, but there are problems. The big intrinsic one is that meteorology is not deterministic; that is, you cannot guarantee the same result given identical initial conditions. Given that reality, making a dichotomous yes/no determination is difficult, would give you less information, and hides the uncertainty surrounding cyclogenesis.

Your local weather forecast is replete with examples. Take, for instance, pop-up thunderstorms in the summer. I would be hard pressed to tell you any given day that it will definitely rain or definitely be dry. I could say "yes" or "no" depending on how sure I am, but that would be based on some probability threshold anyway (i.e., I say it will rain if I am 60% or more sure). Thus, while some lament seeing the 30% chance of thunderstorms every day during the summer, it is a true reflection of the uncertainty. The same thought process goes into the cyclogenesis percentages. Another example would be percentages in the medical literature. For someone diagnosed with cancer, it would be true that statistics could not predict whether or not he would ultimately survive, but it is hard to argue that there isn't a difference between a 5% and 90% 5-yr survival.

BTW, what the NHC expresses in their development probabilities is more remarkable than just yes/no. The graph above shows that if NHC states that a system has a 10% chance of development, that it would actually develop close to 10% of the time, and so on for each additional 10%. Outstanding!

PS, nothing prevents you from doing statistics on categorical vs pure quantitative data. Otherwise, Gregor Mendel couldn't have done his pea plant experiments (short v tall, curled v wrinkled, yellow v green, etc). (He may have fudged the results, but that's for another thread).

Edit, AJC3 beat me to the punch.

Thanks! I get your point. There should be a separate topic for these kind of discussions.

BTW Mendel was using qualitative data in a quantitative context.

Sorry mods I'm back on topic now!

[gatorcane]

Cloud tops are really warming, just thunderstorms from a few hours ago that are pushing out outflow boundaries, may be not much even visible on satellite anymore by tomorrow. Gaston looks quite dead indeed. I suspect most members will be turning their attention to 91L over the coming days as chances for development appear quite good with good model support, while the Gaston threads languishes and fades off.

[fasterdisaster]

Wow, Gaston just hit Hispaniola and it looks like he's dying in a hurry. There was one point where he looked this bad before, but he wasn't over Hispaniola and he may have just gotten lucky. I have to agree that Gaston is on the embalmnig table right now. It's possible he'll wake from the dead, but at this point I doubt it.

[ROCK]

whatever left of the LLC did not go over Hispa, its south of it....

http://www.ssd.noaa.gov/goes/flt/t3/flash-ir2.html

[fasterdisaster]

whatever left of the LLC did not go over Hispa, its south of it....

http://www.ssd.noaa.gov/goes/flt/t3/flash-ir2.html

Ah, yes, it's difficult to see on water vapor. Still, it certainly looks as though it's interacting with it somewhat. It looks truly pathetic right now.

[ROCK]

whatever left of the LLC did not go over Hispa, its south of it....

http://www.ssd.noaa.gov/goes/flt/t3/flash-ir2.html

Ah, yes, it's difficult to see on water vapor. Still, it certainly looks as though it's interacting with it somewhat. It looks truly pathetic right now.

oh yeah no doubt about that....

[ColinDelia]

I think the "pink tracks" summarize how most of us feel about gaston after his 2010 showing

[ColinDelia]

thank you colin i love when yyou get infomative answers to questions and let me say i have read many of your post and you bring alot to the storm2k family..

Thanks. I feel like a real novice compared to a lot of people around here. This is a great learning environment for us all to learn from each other for sure.

[chrisnnavarre]

I was thinking this might dissipate over Hispaniola, but it looks to have passed south. If there's any type of circulation left it will probably really take off between Haiti and Jamaica. Guess he's still headed due west after all.

[ColinDelia]

8 am NHC sez ... "near 0%"