Comments on: GISS Estimation Case Study

By: Cedarville Sausage « Watts Up With That?

Cedarville Sausage « Watts Up With That? — Sat, 19 Jul 2008 03:38:26 +0000

[...] May I began a quest to better understand how GISS does its homogeneity adjustment, also known as GISS Step 2. Steve McIntyre took the ball from that scrum and ran with it, producing [...]

By: John Goetz

John Goetz — Fri, 23 May 2008 20:16:18 +0000

#17 Joe … I did a quick run on one of the Russian stations I had not yet looked at. That is, I calculated the annual average the way GISS does: first calculate monthly averages, then calculate seasonal averages from the three monthly averages, then calculate annual average from the four seasonal averages. Then I calculated the annual average a second way: simply take all days in the year and calculate the mean at once.

I had assumed the difference between the two methods was small. I was wrong. For the single station I examined it was substantial, ranging from -1.11 C to +0.87C.

I have Meteo data from four other stations. I want to go through those before I make any general conclusions.

By: John Goetz

John Goetz — Fri, 23 May 2008 15:51:13 +0000

#19 Gaelen: Presumably, the data I am working with is raw, unadjusted data.

By: Gaelan Clark

Gaelan Clark — Fri, 23 May 2008 15:34:47 +0000

The contemporaneous “adjustments” to the PAST record, that is.

By: Gaelan Clark

Gaelan Clark — Fri, 23 May 2008 15:32:44 +0000

Now, let’s see how the “trend” changes when one takes all of the “adjustments” to the temperature record out. Can this be done? Have all of the temperature “adjustments” been recorded and have they been justified?

By: Sam Urbinto

Sam Urbinto — Fri, 23 May 2008 14:29:02 +0000

The GISTEMP anomaly trend is less than 6% of the nominal 14 it’s based off of. If this trend is off 15%, what does that say.

By: Joe Black

Joe Black — Fri, 23 May 2008 13:41:11 +0000

RE 15

Sure it’s minor, but this supossed to be SCIENCE. Access to computers has been available to SCIENCE since the 60′s if not the 70′s. Computers to the people since the 90′s (arguably the 80′s).

Using a daily weighting would seem to be the “right thing to do”, and certainly not beyond the capabilities of available hardware.

Engineering shifts to Science at somewhere between 10% and 1% of point. Engineering Safety Factors have shifted from 10x (Hoover Dam) to approx 1.1x these days.

By: EW

EW — Fri, 23 May 2008 13:30:05 +0000

12 (JohnG)
The ftp Meteo directory was empty yesterday already – I’ve looked.

By: John Goetz

John Goetz — Fri, 23 May 2008 13:17:44 +0000

Joe, that’s an interesting question. I doubt it amounts to much but is easy enough for me to calculate with what I have. I will look at it later tonight.

By: Mark H.

Mark H. — Fri, 23 May 2008 13:16:52 +0000

Re: #5

After reading the various analysis on temperature reading on this site (from changes in methods in measuring sea temps to ‘adjustments’)one wonders what the record might look like without all the point shaving. While the general trendline would not change (or so I suspect) it gives one pause to think that GCM’s and proxy calibrations to instrumented data would then be off – but then what would that say about GCM robustness?