Old temperatures are adjusted when new temperatures are added

The temperature anomaly (later simply called the temperature) of the last month is added to the older temperatures when the temperature series are updated each month. But in addition to that, for different reasons, older temperatures are often adjusted. In this blog post I will examine the adjustments done on the older temperatures. I will not discuss why the adjustments are done. The examination is done by calculating the differences between the temperatures in the new series and the temperatures in the same series that were released five years ago.

I downloaded the new temperature series in November 2019 ¹. They have temperatures up to and including October 2019. I downloaded the old temperature series in the end of 2014 and in the beginning of 2015. The old UAH and RSS satellite temperature series for the lower troposphere (TLT) have temperatures up to and including September 2014. The old surface temperature series GISTEMP, NOAA Global, HadCRUT4 and BEST have temperatures up to and including December 2014.

Figure 1: The temperature differences between new and old releases of the same temperature series. The lines show the one year moving averages of the monthly differences.

Figure 1 shows the one year moving averages of the differences between the new and the old monthly temperatures of the same temperature series. All temperatures series were made relative to the 30-years baseline 1979 to 2008 before the differences were calculated. For NOAA Global the differences are almost identical to those of the GISTEMP, and they are therefore not included in the figure.

The figure is inspired by the last figure in Nick Stokes' blog post Satellite temperatures are adjusted much more than surface [temperatures are]. The satellite temperatures are based on measurements of  microwave emissions from the lower part of the troposphere. Nick Stokes wrote that 'the task of extracting temperatures from the microwave emissions is just very hard', which explains why the old UAH and RSS temperatures are adjusted much more than the surface temperatures. Nick Stokes also discussed the reactions among the skeptics to the different way in which the UAH and the RSS temperatures are adjusted. The RSS temperatures post-2000 are adjusted to be warmer and thereby increase the global warming trend, and the UAH temperatures are adjusted in the opposite direction and thereby decrease the warming trend. The climate skeptics thrust, not surprisingly, the UAH temperatures.

The skeptics often claim that the adjustments are tampering done in order to increase the warming trend. The table below shows the trends in the 30-years span from the beginning of 1985 till the end of 2014 for both the new and the old temperature series. The new surface temperature series all slightly increase the trend. The new RSS TLT greatly increases the trend, and the new UAH TLT greatly decreases the trend. The situation is more balanced than the skeptics claim.

  Series      Trend new series  Trend old series
  
  GISTEMP     0.178 °C/decade   0.166 °C/decade
  NOAA Global 0.161   --”--     0.154   --”--
  HadCRUT4    0.168   --”--     0.166   --”--
  BEST        0.185   --”--     0.179   --”--
  RSS TLT     0.213   --”--     0.137   --”--
  UAH TLT     0.113   --”--     0.167   --”--

The differences between the monthly temperatures are the basis for the moving averages. Figure 2 and 3 show the monthly differences for the satellite temperatures.

Figure 2: The monthly adjustments done on the old RSS TLT temperatures.

Figure 3: The monthly adjustments done on the old UAH TLT temperatures.

The root mean square (rms) of the monthly differences is a quantitative measure of how much the temperatures are adjusted. It is much greater for the satellite temperatures than for the surface temperatures:

  Temperature   rms of differences 

  GISTEMP       0.032 °C
  HadCRUT4      0.007  "
  BEST          0.019  "
  UAH TLT       0.075  "
  RSS TLT       0.082  "

The satellite temperature series start in January 1979, and the figures shown so far therefore start in 1979. Also the rms values above are calculated for the time span starting in January 1979.

The surface temperature series start in 1880 and earlier. The next figures show the adjustments done on the temperatures from 1880 until the end of 2014 in these series.

Figure 4: The monthly adjustments done on the old GISTEMP temperatures.

The adjustments on the NOAA Global temperatures are similar to the adjustments done on the GISTEMP temperatures.

Figure 5: The monthly adjustments done on the old HadCRUT4 temperatures.

Figure 6: The monthly adjustments done on the old BEST temperatures.

Be aware that Figure 4, 5 and 6 have different scale on the vertical axis. The GISTEMP (and the NOAA Global) temperatures are adjusted more than the HadCRUT4 and the BEST temperatures, especially in the years around World War II.

Met Office changed their temperature series from HadCRUT3 to HadCRUT4 in 2014. The last update of HadCRUT3 has temperatures up to and including May 2014.

Figure 7: The monthly adjustments on the old HadCRUT3 temperatures done in the new HadCRUT4 temperature series.

Figure 7 shows the differences between the new HadCRUT4 and the HadCRUT3 temperatures for the months up to and including May 2014. The differences are similar to the adjustments on the GISTEMP temperatures in Figure 4. This shows that Met Office did their major adjustments in the transition from HadCRUT3 to 4. The HadCRUT3 temperatures have a warming trend 0.153 °C/decade in the 30-years span from June 1985 till May 2014. The HadCRUT4 temperatures have a warming trend 0.174 °C/decade in the same time span; that is slightly more than the HadCRUT3 temperatures have.

The first BEST temperature series that I downloaded has temperatures up to and including December 2013. The BEST series that was released one year later, had only minor adjustments on these temperatures.

Footnote

¹
In the end of November 2019 I downloaded the temperature series from these URLs:

The NASA GISTEMP temperature anomalies were downloaded from
https://data.giss.nasa.gov/gistemp/tabledata_v4/GLB.Ts+dSST.txt

NOAA global temperature anomalies were downloaded from
https://www.ncdc.noaa.gov/noaa-merged-land-ocean-global-surface-temperature-analysis-noaaglobaltemp-v5
I chose Ascii time series and then the file aravg.mon.land_ocean.90S.90N.v5.0.0.201910.asc

The Met Office HadCRUT4 temperature anomalies were downloaded from
http://www.metoffice.gov.uk/hadobs/hadcrut4/data/current/download.html
I chose Global (NH+SH)/2 monthly.

The Berkeley Earth BEST temperature anomalies were downloaded from http://berkeleyearth.org/data
I chose Land+Ocean Monthly Global Average Temperature and I used the first part of the file with 'temperature anomalies in the presence of sea ice are extrapolated from land-surface air temperature anomalies'

The RSS TLT satellite temperature anomalies were downloaded from http://data.remss.com/msu/monthly_time_series/RSS_Monthly_MSU_AMSU_Channel_TLT_Anomalies_Land_and_Ocean_v04_0.txt

The UAH TLT satellite temperature anomalies were downloaded from https://www.nsstc.uah.edu/data/msu/v6.0/tlt/uahncdc_lt_6.0.txt