• Security incident: ISF was recently accessed by intruders. Please change your password, and change it anywhere else you used it. Read more

How To Be A Global Warming Sceptic

lomiller said:
Why would it mean anything different then usual? :confused:
Click to expand...

I'm sorry, my understanding of a plot such as this:

Average data over a certain period is the zero abcissa
Individual data points are plotted based upon their difference from the average. This difference from the average is the 'anomaly'.

Can someone please tell me if my understanding is correct?

Thanks in advance!
 
Alric said:
Not to be snarky but which word is giving you trouble? "Temperature" or "anomaly".
Click to expand...

No, I know that you are not being snarky....

I guess the "anomaly" part...............

To be honest, based upon my understanding (I'm the first to admit my understanding may be flawed), the anomaly line would be at the "grand mean" of the data set.

Then, you would plot the individual data points above and below that, based upon how far they are from that average.

The 0 deg. C line in the plots shown is clearly not at the mean of the data sets that are shown.

That's what led me to believe that my interpretation of "anomaly" may be in error.
 
Alric said:
Not to be snarky but which word is giving you trouble? "Temperature" or "anomaly".
Click to expand...

Also, I thought that they might mention it in the paper, but the only times the word "anomaly" appeared was in the graphs.
 
My apologies. I am so used to mhaze my first impulse is to be snarky.

All this data are normalized to relative differences in temperature to be able to exist in one chart. Also it gives an idea of the magnitude of the change in all cases. Rarely in science do you get so many disparate data points to agree in a similar trend.

More importantly there is no single set of data that contradicts this trend.
 
Has the temp really increased around half a degree in the last seven years? Or is that chart an old prediction from 2000? Is it supposed to stop at 2000? The end is hard to read.
 
Alric said:
My apologies. I am so used to mhaze my first impulse is to be snarky.
Click to expand...

Apology is certainly not needed! No offense taken! Although things might get snarky later..who knows? :)

All this data are normalized to relative differences in temperature to be able to exist in one chart.
Click to expand...

That's what confuses me about the appearance of this chart..if all data were normalized, why don't we see equal amounts of temperatures above 0 deg C, as we see below?
 
Last edited:
Wangler said:
No, I know that you are not being snarky....

I guess the "anomaly" part...............

To be honest, based upon my understanding (I'm the first to admit my understanding may be flawed), the anomaly line would be at the "grand mean" of the data set.

Then, you would plot the individual data points above and below that, based upon how far they are from that average.

The 0 deg. C line in the plots shown is clearly not at the mean of the data sets that are shown.

That's what led me to believe that my interpretation of "anomaly" may be in error.
Click to expand...


They usually take a base period to calculate the anomaly from. So they might use a time period like 1951 to 1980 or so when calculating anomalies for the instrumental record. It doesn't change the shape of the curve anyway.
 
Last edited:
fsol said:
They usually take a base period to calculate the anomaly from. So they might use a time period like 1951 to 1980 or so when calculating anomalies for the instrumental record. It doesn't change the shape of the curve anyway.
Click to expand...
Yes. If the data were normalised in the usual way, each graph would be hard to compare with any other as the baseline would be different every time. It is usual to plot anomalies relative to the average for a period of several decades in the recent past.

NASA GISS tend use 1951-80, Hadley CRU tend to use 1961-90; in this case Mann et al used the latter. You can get the PDF here.
 
fsol said:
They usually take a base period to calculate the anomaly from. So they might use a time period like 1951 to 1980 or so when calculating anomalies for the instrumental record. It doesn't change the shape of the curve anyway.
Click to expand...

Yes, I understand...if the data more closely bracketed the 0 deg C curve, the far left of the curve would have an even larger anomaly (it would be much higher than the 0 deg C than it is now).

They don't have a reference (that I have found) to their basis for the 0 deg C curve. Yes, they

might use a time period like 1951 to 1980 or so when calculating anomalies for the instrumental record.
Click to expand...

but how do we know?

Am I the only one who thinks that this is a poor way to construct a graph for a scientific paper?
 
Wangler said:
No, I know that you are not being snarky....

I guess the "anomaly" part...............

To be honest, based upon my understanding (I'm the first to admit my understanding may be flawed), the anomaly line would be at the "grand mean" of the data set.
Click to expand...

The anomaly isn’t going to be based on the mean of the data set. You ant your reference to be a period where you more confidence in the numbers, so most reconstructions use a reference someplace in the 20th century. Even the various temperature sets you see for the 20th century don’t all use the same reference point, which can get confusing for some of the more mathematically challenged denialists…

To expand a little, the reason anomalies are used is that they are a powerful tool for removing static error. In a noisy data set generated with an instrument that may or may not be well calibrated to the actual temperature you can still identify relative values with good accuracy.

Take a clock that is out by 96 min, unless you know exactly how much it’s out you can’t tell exactly what time it is. You can still use it to find out exactly how much time has passed because the 96 min error cancels out. I.E. from 3:15 to 3:36 on your clock is still 21 min no matter how far out your clock is set. Now lets sat you have a clock that in addition to not being set properly has no faceplate or numbers. You can still use it if you can calibrate it against another clock and find out how fast/slow it’s moving.

The situation is similar for temperature stations and climate proxies. You can identify when it’s warmer and when it’s colder, but the data is to noisy overall to say precisely what the average temperature is in a given year. If you take the anomaly, however most of this noise drops away. You can then take the information that remains and calibrate it against an overlapping period where your data is much better to find out how warm/cool it was relative to that reference period.

For this, you want a reference period in the 20th century where you have actually measured the temperature. It’s not always the latter half of the 20th century however because some proxies are skewed by human activity from the middle of the century onwards.
 
Wangler said:
but how do we know?
Click to expand...

You read the paper.

Wangler said:
Am I the only one who thinks that this is a poor way to construct a graph for a scientific paper?
Click to expand...

How so? The reference itself doesn’t change the data but a poor choice for the reference period could be a problem. In addition, you have data sets that have been in use since before the earliest data in other sets. For example satellite data has only been available since 1979, while GISS was already in existence. It would be impossible for the two sets to use the same reference period, and compensating for the reference value isn’t very difficult.
 
lomiller said:
The anomaly isn’t going to be based on the mean of the data set. You ant your reference to be a period where you more confidence in the numbers, so most reconstructions use a reference someplace in the 20th century. Even the various temperature sets you see for the 20th century don’t all use the same reference point, which can get confusing for some of the more mathematically challenged denialists…
Click to expand...

I don't know about mathematically challenged deniers, but it certainly would be confusing even to mathematically adept AGW-proponents if the temperature sets don't clearly define how they obtained the reference point, like this paper we are talking about.

Iomiller said:
To expand a little, the reason anomalies are used is that they are a powerful tool for removing static error. In a noisy data set generated with an instrument that may or may not be well calibrated to the actual temperature you can still identify relative values with good accuracy.

Take a clock that is out by 96 min, unless you know exactly how much it’s out you can’t tell exactly what time it is. You can still use it to find out exactly how much time has passed because the 96 min error cancels out. I.E. from 3:15 to 3:36 on your clock is still 21 min no matter how far out your clock is set. Now lets sat you have a clock that in addition to not being set properly has no faceplate or numbers. You can still use it if you can calibrate it against another clock and find out how fast/slow it’s moving.

The situation is similar for temperature stations and climate proxies. You can identify when it’s warmer and when it’s colder, but the data is to noisy overall to say precisely what the average temperature is in a given year. If you take the anomaly, however most of this noise drops away. You can then take the information that remains and calibrate it against an overlapping period where your data is much better to find out how warm/cool it was relative to that reference period.

For this, you want a reference period in the 20th century where you have actually measured the temperature. It’s not always the latter half of the 20th century however because some proxies are skewed by human activity from the middle of the century onwards.
Click to expand...

Iomiller,

Thanks for the explanation. It makes sense to me that it doesn't truly matter where the anomaly is placed, as long as the deviations from that point are accurately represented, you will have a good visualization of the data.

However, I am going to read that short paper again, because it seems only reasonable in a scientific paper to state what your basis for the 0 deg C point is, otherwise it gets confusing for all, regardless of their level of proficiency at maths..................
 
Last edited:
lomiller said:
The reference itself doesn’t change the data but a poor choice for the reference period could be a problem.
Click to expand...

Iomiller, I understand that the choice of reference will not change the data, but could you explain how the reference period choice could be a problem?

I guess it seems to me that the reference is a reference; if you say "based upon this reference, this is how the data looks".

Do you mean this:

  1. We could chose yesterdays' mean temperature for St. Paul, Minnesota, as our reference.
  2. We then plot all the data against that reference.
  3. We have then a nice graph, that shows the data correctly, but it isn't very useful.
 
You must log in or register to reply here.

Share this page

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom