A Thread for FCT's Explanations

[flacaltenn]

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
A thanks to USMB POSTERS ----

The USMB community is the best!!!
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

 

[Abraham3]

A Demonstration

One thousand windspeed readings. These are data generated in Open Office Calc. Note the standard deviation (SD:28.8): the average distance between individual readings and the mean of the entire set (Mean: 44.7).

Following here are the same data split into ten, consecutive segments of 100 data points each. Note that each has a significantly smaller standard deviation (SD) than the value we've just seen in the entire set. Note also that there is no correlation between max, min or mean windspeed and standard deviation. With a varying quantity, like windspeed, standard deviation DOES correlate with sample size.

I am working on a second set which will demonstrate how FCT's statements regarding increased accuracy with increased sample size DO apply to measurements of STATIC values

 

[Abraham3]

Here are ten sets of 1024 points centered about the origin with random radii less than or equal to 100. Think of this as the output of a GPS dancing around its actual location or guns shots scattered around a bullseye:

Here, for each set, are the distances from the origin to the average locations of the first 2, 4, 8, 16, 32, 64, 128, 256, 512 and all 1024 points:

Increased sample numbers do improve accuracy for static values ((0,0) in this case). In the case of varying values, like the speed of the wind, they do not. If the wind had a "static mean", sustained speed and gust speed would be the same value. The WMO is not shooting for objective accuracy. They are looking for a method to come up with simple numbers to pass to the public.

For a varying quantity, such as windspeed, increased sample time does NOT improve accuracy.

 

[flacaltenn]

A Demonstration

One thousand windspeed readings. These are data generated in Open Office Calc. Note the standard deviation (SD:28.8): the average distance between individual readings and the mean of the entire set (Mean: 44.7).

Following here are the same data split into ten, consecutive segments of 100 data points each. Note that each has a significantly smaller standard deviation (SD) than the value we've just seen in the entire set. Note also that there is no correlation between max, min or mean windspeed and standard deviation. With a varying quantity, like windspeed, standard deviation DOES correlate with sample size.

I am working on a second set which will demonstrate how FCT's statements regarding increased accuracy with increased sample size DO apply to measurements of STATIC values

Really, don't bother.. No one in the WORLD is interested in what the average wind speed of a hurricane was OVER A 2 DAY PERIOD.. Which is what you're comparing to..

The parameters chosen by the meterologists like max gust and AVERAGE wind speed are descriptive of the winds bounded by some RATIONAL criteria --- like over a 3 mile distance of travel or what would appear at a particular reporting location at any time.

The MEAN over the DURATION of the storm is meaningless and silly, because CERTAINLY that is NOT a stationary process. HOWEVER, as the EXPERTS have determined and I showed in ACTUAL wind field measurements --- It IS stationary over at least a 10 minute period --- which would be on the scale of your SIMULATED data --- only 3.5 sample points.

If you're data was a storm -- you'd be looking at a 48 hour time scale for the X axis. Using a 1000 points gives you 173 seconds per point. So it would violate the SAMPLING rules for capturing and digitizing data with the resolution REQUIRED to preserve almost ANY of the TRUE deviation of the storm energy.. Sampling RAW at about 3 minutes -- wouldn't give you ANY representation of the process deviation. REAL WIND DATA is not sampled at 173 seconds per point.. More like 0.25 seconds per point -- so your scaling of the problem is off by a COUPLE ORDERS OF MAGNITUDE...

That's why the 10 minute in the SCIENTIFIC STUDY that I gave you DOES show accurate tracking of the AVG 10 minute wind speed THRUOUT the ENTIRE DURATION of the storm..

Just goes to show that making the TOOLS available, doesn't make up for the knowledge of how those tools need to be applied..

I'm also disturbed by the fact that the "noise" you've created in the main TOTAL plot is noticeably almost TWICE as large as what appears in your segments. There may be other pilot errors present here. But the main point is that you are deriving statistics for events that would have NO USEFUL MEANING for analysis of hurricane wind speeds.

Attach the raw data and I'll see if I can help you interpret the MEANING of what you're attempting to do...

And I thought we were done.. Can I not trust your decisions? I already know I can't trust your memory or integrity for saying things offensive to people..

 

[Abraham3]

When I said I was done, I had made the mistake of thinking you would finally realize you were wrong on the debate topic, wrong to try to twist the debate topic around to allow a claim of victory and that you would have the cajones to admit it. I was wrong on all three.

I most certainly do not need your help interpreting the data I created. I was tickled to hear what time spans my graphs represented. Surely you realized you had absolutely no justification for any of those statements. They are simply a series of 1,000 random numbers.

Increasing the time span over which you collect data on a time-variant phenomenon with the aim of reducing it to a single representative mean, does NOT increase accuracy. Claiming that any portion of a time variant phenomenon (like windspeed) is CONSTANT is purely subjective bullshit. You know it. I know it. Anyone reading this knows it. Windspeed is not a constant. You lose.

 

[flacaltenn]

If I wanted to see the outcome of giving a Lemur a power drill and watching him learn how to use it without any guidance from textbooks or theory or science --- this would be it.

Your issue is no longer with me. It's with EVERY STATISTICS book ever written and the BULK of the Worlds Meterological Agencies. GO READ A FUCKING MANUAL on the tools you're attempting to use.

My Lemur experiment didn't turn out to be real pretty...

 

[flacaltenn]

In fact -- I ran plumb out of Lemurs to foster (ItFitzMe) Abraham3..
And the damage was extensive... True story...

 

[Abraham3]

If I wanted to see the outcome of giving a Lemur a power drill and watching him learn how to use it without any guidance from textbooks or theory or science --- this would be it.

Your issue is no longer with me. It's with EVERY STATISTICS book ever written and the BULK of the Worlds Meterological Agencies. GO READ A FUCKING MANUAL on the tools you're attempting to use.

My Lemur experiment didn't turn out to be real pretty...

That looks to me like you surrender.

 

[flacaltenn]

Nope ... Waiting for you TO REFUTE anything I told you in post #24.. PREFERABLY with REAL references to traditional statistical methods..

You don't know how to even SET UP a statistical question for testing.. Lemme Repeat...
NO ONE in the entire fucking world CARES what the average windspeed for a hurricane is over a 2 day interval..

And if you're gonna model it correctly --- you cannot dismiss the observations that I made about your choice of sampling. But you DID --- because I SUSPECT --- you don't know how to do it properly.

 

[Abraham3]

There's a clear indication of the problems you're having with this. The point you chose to debate said nothing about hurricanes or what people care about. Everything you bring up in this discussion is completely irrelevant to the question.

It all boils down to whether or not windspeed is constant or time-variant. You've already said you believe it's constant. Well, sorry, but that's a really dumb position. I don't need either of my statistics text books to tell me that windspeed is not a constant. You lose.

 

[Abraham3]

There's a clear indication of the problems you're having with this. The point you chose to debate said nothing about hurricanes or what people care about. Everything you bring up in this discussion is completely irrelevant to the question.

It all boils down to whether or not windspeed is constant or time-variant. You've already said you believe it's constant. Well, sorry, but that's a really dumb position. I don't need either of my statistics text books to tell me that windspeed is not a constant. You lose.

A point that bears discussion is the reason this debate ever took place. Why did FCT and several posters before him bring up this topic. A review of the discussion regarding Typhoon Haiyam will, I think, clearly reveal that a number of folks who have argued against the idea increased extreme weather, did not like hearing that Haiyam was the most powerful typhoon in recorded history. They really glommed on to the WMO conclusion that maximum sustained winds never got out of Category 4 rather than the full Cat 5 that the US NWS and several other national and international weather organizations found. That's the reason this argument took place.