GISTEMP: 2009 tied for second warmest year on record

[swansont]

Of course everyone makes mistakes, but when Hansen makes a declaration on temperatures in a region and it is later found that data in that region has an obvious disjointed 0.3 step increase almost across the board it gives the lasting impression that Hansen does no checking of the data at all.

 

I will grant you that NOAA and GHCN should do a better job of cleaning their data before Hansen gets his hands on it, but it is really all of their responsibility to check this data.. and Y2K error and Siberia are particularly easy to spot with even a casual check of the data. Three organizations tasked with collecting and evaluation temperature data missed a large error with a large portion of their data, and continued missing seeing it for 6 years.

 

The data error in the US is less about the impact on the global temperature estimate (because it was weighted away) but is more about the pecentage of stations used that were effected.

 

So I guess what I am trying to say is what good is credibility if the person remains careless? And can you even be credible if you show no signs of being more careful as a result?

 

 

I can't speak to the details of Hansen's routines for processing and analyzing. But I can compare my own experiences as an experimentalist in a different field, and say that mistakes can be made without "careless" being an appropriate label. I also suspect that if some of the critics had their professional accomplishments scrutinized as carefully as Hansen's, that we would discover that Hansen is being held to an unfair standard in this regard. That certainly seems to be true of the more popular "skeptical" blogs, many of whose claims are readily debunked and yet they still seem to have credibility amongst their followers.

[jryan]

Here is something I think is worth mentioning on this topic:

 

In August 2007 Reto Rudy and James Hansen tell Doyle Rice of USAToday in an email that they don't use their own GISSTemp records to evaluate their own models. (pdf) GISS uses NCDC and HADCrut.

[earthandbeyond]

It has been the coldest winter in a very long while in Europe.

[iNow]

It has been the coldest winter in a very long while in Europe.

 

Since roughly the 1970s, correct, but local temperature variations don't negate the overall global average which is the focus of this particular thread. Global averages have gone up steadily and consistently, and we are seeing warming temperatures around the world (the warmest averages ever recorded), despite the fact that it gets chilly and snows sometimes in some areas (which, oddly enough, is also foretasted by climate models due to the extra energy in the system).

[swansont]

Here is something I think is worth mentioning on this topic:

 

In August 2007 Reto Rudy and James Hansen tell Doyle Rice of USAToday in an email that they don't use their own GISSTemp records to evaluate their own models. (pdf) GISS uses NCDC and HADCrut.

 

Wait, what? I though GISSTEMP was a model output. You can't evaluate your model by comparing it to itself.

---

Merged post follows:

Consecutive posts merged

Global averages have gone up steadily and consistently

 

Note well, though, that this is not the same as linearly.

[jryan]

Wait, what? I though GISSTEMP was a model output. You can't evaluate your model by comparing it to itself.

---

Merged post follows:

Consecutive posts merged

 

Yeah, I worded that wrong. In the emails Reto Rudy is telling the USA Today writer to use HADCrut and NCDC, not GISTemp, for his reporting as temp records are "what they do best" and GISS does "rudimentary analysis" of global observed data.

 

So I guess my question is this: Then what exactly IS GISTemp?

 

 

Side note: Since GISS is using NCDC and HADcrut to check their GISTemp model then it's no surprise that GISTemp, HADCcrut and NCDC look alike, is it? I think the earlier attempt to validate GISTemp by it's agreement with HADcrut fall apart on that revelation.

 

I wonder what record HADcrut used to validate their temperature model?

[JohnB]

I wonder what record HADcrut used to validate their temperature model?

 

No idea. They apparently lost the data.

 

swansont, AFAIK GISSTEMP is not a "model" as such, in that it has no predictive power. It is a method of combining global station readings into a coherent whole. On that basis, I see little reason not to use it to validate the NASA models. However, it strikes me as a good idea to use someone elses temp series to prevent possible conflicts of interest.

 

Conflict of interest is a major concern in the business world but is unfortunately given little thought in the climate community.

Edited March 16, 2010 by JohnB

[bascule]

Of course everyone makes mistakes, but when Hansen makes a declaration on temperatures in a region and it is later found that data in that region has an obvious disjointed 0.3 step increase almost across the board it gives the lasting impression that Hansen does no checking of the data at all.

 

He made a mistake, therefore he does no checking of his data at all! Slippery slope much?

 

In the emails Reto Rudy is telling the USA Today writer to use HADCrut and NCDC, not GISTemp, for his reporting as temp records are "what they do best" and GISS does "rudimentary analysis" of global observed data.

 

*facepalm*

 

jryan, seriously, can you knock off the completely out of context quotes? Here's what you paraphrased Reto Rudy as saying:

 

GISS does "rudimentary analysis" of global observed data.

 

Here's the quote where he actually said the words "rudimentary analysis":

 

We are basically a modeling group and were forced into rudimentary analysis of the global observed data in the 70's and early 80's since nobody else was doing that job at the time.

 

Since GISS is using NCDC and HADcrut to check their GISTemp model then it's no surprise that GISTemp, HADCcrut and NCDC look alike, is it?

 

If they didn't look alike, then it would be an indication that one of them was doing something wrong. So yes, it's no surprise.

 

AFAIK GISSTEMP is not a "model" as such, in that it has no predictive power. It is a method of combining global station readings into a coherent whole.

 

Models don't have to have predictive power. Models provide estimates of the behavior of larger systems.

[swansont]

No idea. They apparently lost the data.

 

swansont, AFAIK GISSTEMP is not a "model" as such, in that it has no predictive power. It is a method of combining global station readings into a coherent whole. On that basis, I see little reason not to use it to validate the NASA models. However, it strikes me as a good idea to use someone elses temp series to prevent possible conflicts of interest.

 

Conflict of interest is a major concern in the business world but is unfortunately given little thought in the climate community.

 

"Model" is basically a proxy for "math." Science is models, all the way down.

 

AFAIK GISSTEMP predicts temperatures in area where there are no stations via interpolation algorithms. You can test that by interpolating even where there are stations, and checking against the data or another model's output. You can't check it against itself.

[JohnB]

Sorry swansont and bascule, terminology mismatch.

 

I was differentiating GISSTEMP (interpolating the past) from "Climate Models" (predicting the future).