Updating the global temperature graphs

The fuss recently about a fairly minor fix to the fairly ordinary estimate of global temperature change provided by satellite remote sensing has prompted some long-overdue action from me. I’ve updated and I hope improved all the global temperature graphs (click for links; you may need a browser refresh to see the latest versions):

 
The main tweaks are as follows:

1. Revised reference interval

I’ve changed the unified reference interval used for plotting the temperature anomalies from the ‘old standard’ 30 year period 1961-1990 to the 40 years from 1881-1920, a choice James Hansen has been promoting recently. This earlier interval has several advantages:

  • Temperatures in this period have been shown to be within a couple of tenths of a °C of ‘pre-industrial’, and are probably as close as it is practical to get to that metric using just the instrumental record. Earlier instrumental temperature estimates do exist in some series, but they become progressively less reliable the further back you go. A near pre-industrial base period is critical to clear understanding, because common parlance like ‘two degrees of warming’ or ‘a plus 2°C world’ refer specifically to pre-industrial conditions.
     
  • A reference interval longer than the ‘standard’ 30 years has been shown to be desirable, though that probably makes no real difference for my simple rebasing for comparative plotting purposes.
     
  • The interval 1881-1920 centres on 1900, so can be conveniently and fairly accurately referred to as ‘since 1900’ as well as ‘compared to pre-industrial’

 
The new interval does have disadvantages, but in my opinion they are tolerable:

  • Earlier instrumental global average temperatures are necessarily less well known. Particularly prior to about 1900, the uncertainty in monthly global temperature estimates begins to rise sharply. Using an uncertain monthly average for reference of course propagates through the whole anomaly record, making it potentially less reliable in terms of immediate month-to-month relativity. Again, this effect is probably slight for my simple monthly rebasing for comparative plotting purposes.
     
  • No part of the reference interval overlaps with quality reanalysis temperature estimates or with satellite remote sensing estimates, so those can only be positioned in the overall anomaly record by average matching with the instrumental estimates. That was essentially already the case with 1961-1990 for the satellite series, but it wasn’t previously for the NCEP/NCAR reanalysis.

 

2. Updated satellite series

I’ve of course updated the Remote Sensing Systems satellite remote sensing estimates of lower troposphere temperature (‘TLT’) to the new version 4. That has higher recent temperatures, mostly due to a more thorough treatment of the effects of satellite orbital decay.

I’ve also updated Dr Spencer’s UAH satellite TLT estimates to version 6, which he no longer seems to refer to as ‘beta’. Spencer, as you may know, is something of a climate sceptic, and has erred in the past in failing to adjust his remote sensing estimates for orbital decay (at all!). His new version actually has lower recent temperatures. These several things may not be related.

Frankly I find the machinations regarding satellite temperature estimates bordering on ridiculous. These are relatively poor measures of something we are not especially interested in – the average atmospheric temperature at around 5000 m elevation, or nearly 20,000 feet above sea level (the TLT temperatures are about 25°C lower than where we live).

 

Outcome

Take a step back and all the bluster and argument about data quality and this method-is-better-than-that becomes absurd. At the scale of where we’re headed, it simply does not matter.

Simple global temperature extrapolations

 
And by the way, we’ll be at +2°C well before 2050 on our current trajectory. That’s in about 30 years time.
 

1 comment to Updating the global temperature graphs

  • angech

    Taming has an interesting blog on choice if baseline and the effect it has on the anomalies and the size of their displacement. Well worth a look at