GISTEMP: Use Only GHCN Data

March 10, 2009
Previous articles gave an overview of GISTEMP and described how the baseline was established and validated.  That baseline is used for comparison purposes during the testing of changes to the GISTEMP software and/or data files.  This article examines the effects of only using data from the Global Historical Climatology Network (GHCN) on the surface temperature analysis.

Normally, STEP0 of GISTEMP combines temperature information from Antarctic stations, Hohenpeissenberg, and selected United States Historical Climatology Network (USHCN) records with the unadjusted GHCN data that forms the core of the temperature record.  That combined data is used as input to the rest of the processing done by GISTEMP. 

For this test we effectively bypass the processing in STEP0 and use only GHCN data.  In order to accomplish this, we replaced the lines that execute do_comb_step0.sh and copy v2.mean_comb in the run_gistemp script with the following:

		#./do_comb_step0.sh v2.mean
		#cp to_next_step/v2.mean_comb ../STEP1/to_next_step
		dump_old.exe input_files/v2.mean xxx 1880
		mv xxx ../STEP1/to_next_step/v2.mean_comb

The first two lines comment out the call to do_comb_step0.sh and copying the v2.mean_comb file to STEP1, the next line gets rid of any temperature records before 1880, and the last line moves the resulting file to the to_next_step directory in STEP1.  After making the changes, run_gistemp was executed and the results were compared to the baseline using run_compare.  The detailed output from run_compare was archived and the summary is presented below:

		STEP3 Global anomalies - Found 1834 total differences, 983 >.01C
		Higher than baseline 1415 times, lower 419 times
		Differences >.01C higher 780 times, lower 203 times

		STEP3 NH anomalies - Found 1815 total differences, 1026 >.01C
		Higher than baseline 1236 times, lower 579 times
		Differences >.01C higher 700 times, lower 326 times

		STEP3 SH anomalies - Found 1901 total differences, 993 >.01C
		Higher than baseline 1469 times, lower 432 times
		Differences >.01C higher 751 times, lower 242 times

		STEP3 Zonal anomalies - Found 1184 total differences, 722 >.01C
		Higher than baseline 908 times, lower 276 times
		Differences >.01C higher 554 times, lower 168 times

		STEP4_5 Global anomalies - Found 1209 total differences, 185 >.01C
		Higher than baseline 859 times, lower 350 times
		Differences >.01C higher 53 times, lower 132 times

		STEP4_5 NH anomalies - Found 1419 total differences, 257 >.01C
		Higher than baseline 905 times, lower 514 times
		Differences >.01C higher 88 times, lower 169 times

		STEP4_5 SH anomalies - Found 1177 total differences, 331 >.01C
		Higher than baseline 805 times, lower 372 times
		Differences >.01C higher 173 times, lower 158 times

		STEP4_5 Zonal anomalies - Found 766 total differences, 265 >.01C
		Higher than baseline 535 times, lower 231 times
		Differences >.01C higher 146 times, lower 119 times

As can be seen from the summary, 11,305 of the 17,504 anomaly values in the files had differences and 4,762 of those were more than .01C.  Next, run_extract was used to create tab delimited files that were loaded into an Excel spreadsheet in order to create charts and determine average changes.

The average change in the land only files was a warming of .008C as compared to the baseline and the average change in the land and sea files was a warming of .001C .  That average understates the true changes, which varied for the Global land only anomalies from an average warming of .018C during the period from 1880 - 1950 to an average cooling of .01C during the period from 1980 - 2008 as shown in the chart below. 

The average global land and sea change was a slight warming of .001C. The differences indicate a small warming trend for the older years and a small cooling trend for the more recent years as shown in the chart below.

The largest variance occurred in the Northern Hemisphere land only files where the warming during 1880 - 1950 averaged .022C and the cooling during 1980 - 2008 averaged .019C.

Conclusion:
All of the charts using only the GHCN data show a similar pattern where the differences essentially pivot about an axis from approximately 1955 to 1980 with the early years getting a little warmer and the most recent years getting a little cooler as compared to the baseline, which includes the Antarctic, Hohenpeissenberg, and USHCN data.  In a future article, we will determine how much of the differences are attributable to each of the other data sources.