Note: For the full collection of archived data files and NASA GISS-generated "Fig.D" graphs see:  http://www.sealevel.info/GISS_FigD/
 

NASA GISS "Fig.D" U.S. 48-State Surface Air Temperature Anomaly


Dear Steve,

I love your article showing the progression of the NCDC and/or GISS "adjustments" to the U.S. surface air temperature data. However, what you called the "1999" version of the NASA "FigD.txt" (a/k/a Fig.D.txt) data is actually a version from mid-2000. Based on the file's name and contents I'm pretty sure that John Daly downloaded and saved it in June, 2000. That's why it includes the full-year average for 1999, and also why it doesn't match the (earlier) graph found in Hansen's August, 1999 article and paper (doi:10.1029/1999JD900835).

These are all the versions of John Daly's saved copies of the GISS "FigD.txt" or "Fig.D.txt" data file which I have been able to find on his web site:

"Contiguous 48 U.S. Surface Air Temperature Anomaly (C)"

There might be more, but if so I didn't find them. I've cc'd Jerry Brennan; perhaps he can check.

Note that John died on January 29, 2004, so only the two oldest versions could have been downloaded by him. I don't know who downloaded the 2004-2007 versions of the data file and put them on John's site. My guess is that it was Jerry Brennan, who is the current webmaster of John's site, or perhaps John's widow, Amy, or his daughter, Rachel.

The data files have apparently resided on NASA's servers in six locations, over the years:

Note that the "February, 2007 (early)" version is identical to the "March, 2001" version for the years 1880 through 1961, but the "February, 2007 (late)" version differs considerably from both of them, all the way back to 1880.

Even back in 2000-2001 NOAA NCDC and/or NASA GISS were busily revising the data to shrink the 1930s temperature spike and increase post-WWII warming. John Daly created this blink comparator to illustrate what was going on:

big version: http://www.john-daly.com/giss-wig.gif

small version:
Daly blink comparator, June 2000 vs. March 2001

It compares the June, 2000 version of the data to the March, 2001 version. Because John's "before" version was the June, 2000 version, I suspect that is the earliest data file he had.

 

Recovery of missing 1999 data

Unfortunately, NASA GISS seems to have "disappeared" the 1999 version of the data file, which was depicted in Hansen's graph. (They also used robots.txt to prevent the data from being archived.)

In late 2011 I asked the CSRRT (Climate Science Rapid Response Team) to help me find that lost data, but they were stumped. Here's the conversation:
http://tinyurl.com/revstoustemp2

So I've reconstructed the lost 1999 data file by digitizing the annual temperatures from Hansen's 1999 graph.

My first attempt to reconstruct the 1999 version of the data was from this little graph, in which 1934 was shown as 0.6 °C warmer than 1998. The graph's small size, and the fact that the "5-year mean" is opaque and covers up several data points, made it difficult. I enlarged it to 500%, converted the red 5-year mean to green with Irfanview (to contrast with red digitized data points), and used WebPlotDigitizer to manually digitize the 119 data points, which are shown as red dots here:

big version: http://www.burtonsys.com/climate/fig1x_1999_US_temp_500pct_green_digitized.jpg

small version:
1999 version


However, I subsequently noticed that the 1999 Hansen JGR paper has a high-resolution B&W version of the same graph, as "Figure 6" on p.37. Here it is:

big version: http://www.burtonsys.com/climate/fig1x_1999_highres_fig6_from_paper4.png

small version:
Fig 6 from Hansen, et al (1999)

So I digitized that version. The digitized points are indicated in red here:
http://www.burtonsys.com/climate/fig1x_1999_highres_fig6_from_paper_digitized.png

The resulting reconstructed 1999 FigD.txt data file is here:
http://www.burtonsys.com/climate/usatemps_1999_digitized_from_graph_v2.txt

The corresponding Excel spreadsheet (in which the temperatures have not been rounded to the nearest 0.01 °C) is here:
http://www.burtonsys.com/climate/FigD_1999_digitized_hires_v2.xls

All of the temperature values in the reconstructed data files should be within ± 0.01 °C of those in the lost 1999 data file, except for the 1880 and 1881 "5-year mean" temperatures, which are rough estimates made by eyeballing the graph (since the 1878 & 1879 temperatures weren't reported).

The version from the low-resolution graph is not as accurate, so don't use it:
http://www.burtonsys.com/climate/usatemps_1999_digitized_from_graph_v1.txt


Warmest regards,
Dave
http://www.sealevel.info/