The mid-troposphere has been warming faster than you thought

A new version of the RSS dataset has been announced in a paper in the AMS Journal of Climate. The paper, by Carl Mears and Frank J. Wentz, was published before the new version has appeared. This is in contrast to UAH, where the paper hasn't yet appeared but version 6.0 beta came out in April last year, and is now at beta 5.

The paper is about middle troposphere measurements, not lower troposphere which is what is usually discussed here. However, it's about the same instruments that are used to estimate lower troposphere temperatures: Microwave Sounding Unit (MSU) channel 2, and the Advanced Microwave Sounding Unit (AMSU) channel 5.

In the abstract, the authors say (my paras):

Previous versions of the RSS dataset have used a diurnal climatology derived from general circulation model output to remove the effects of drifting local measurement time. In this paper, we present evidence that this previous method is not sufficiently accurate, and present several alternative methods to optimize these adjustments using information from the satellite measurements themselves. These are used to construct a number of candidate climate data records using measurements from 15 MSU and AMSU satellites.

The new methods result in improved agreement between measurements made by different satellites at the same time. We choose a method based on an optimized second harmonic adjustment to produce a new version of the RSS dataset, Version 4.0.

The new dataset shows substantially increased global-scale warming relative to the previous version of the dataset, particularly after 1998. The new dataset shows more warming than most other middle tropospheric data records constructed from the same set of satellites. We also show that the new dataset is consistent with long-term changes in total column water vapor over the tropical oceans, lending support to its long-term accuracy.

The authors mention the Po-Chedley paper on diurnal cycle contamination, and talked about how they went about improving the data, correcting for this. They also discussed how they tested their methods, which seems to have been quite thorough. The authors explain:

The derivation of long-51 term trends in tropospheric temperature from satellite observations requires that the diurnally varying component for the observation be removed. This is because the local observation time for most of the satellites drifts over time (Christy et al. 2000; Mears and Wentz 2005), causing diurnal variations to be aliased into the long-term record. Ideally, we would like to use a highly accurate, independent source of atmospheric and surface temperature specify the diurnal cycle. Unfortunately, no such data are available.

They discussed three different approaches for removing the effects of the diurnal cycle, and explained why they adopted the one they did.

[Amended] Below are some charts of the mid-troposphere temperature record, using the two versions of RSS, and both versions of UAH - v5.6 and v6.0 beta 5. The chart is all years to 2014 2015. It stops at 2014 because that's as far as the comparison goes in the paper (which was probably written before the end of 2015):

Data sources: RSS v3.3 and v4; UAH v5.6 and 6.0 beta 5

In the above chart, the trend from 1979 to 2014 is a bit different to what the paper shows, so I might have the data wrong or there may have been further adjustments since the paper was written, or there's some other explanation. I adjusted the RSS baseline to match that of UAH (1981-2010). I got the v4 data from here. 

The data for RSS v4 has been changed/corrected with a new source here and here. Change the "channel" drop down box here to see the TMT v4 version, and click "download data" for the data. The previous source is no longer available (thanks Anonymous in the comments). The new source is now consistent with the paper, so I've updated the chart. The data also now goes through to the present, so in the above chart the trends are now to 2015 and are:

• UAH v5.6 0.06 C/decade
• UAH v6.0 beta 5 0.072 C/decade
• RSS v3.3 0.08 C/decade
• RSS v4 0.13 C/decade

Updated by Sou 5:00 pm AEDT 4 March 2016

From the paper, the trend goes from 0.078 C/decade with v3.3 to 0.125 C/decade in v4.0 after all adjustments. Below is the relevant chart from the paper (click to enlarge as always):

This will make UAH odd one out again, as far as mid-troposphere temperatures go. It looks as if the changes are around the time that many of us noticed in the lower troposphere (early 2000s), though the paper doesn't discuss the impact there. Since the same instruments are used, I expect there will be a difference. From the RSS website, where TLT is the lower troposphere:

TLT is a more complex data set constructed by calculating a weighted difference between measurements made at different Earth incidence angles to extrapolate MSU channel 2 and AMSU channel 5 measurements lower in the atmosphere.

I don't know when the new version will be out, however they do provide a preview website.

References and further reading

Carl A. Mears and Frank J. Wentz, 2016: "Sensitivity of satellite-derived tropospheric temperature trends to the diurnal cycle adjustment." J. Climate doi: http://dx.doi.org/10.1175/JCLI-D-15-0744.1 (subs req'd - now open access)

Stephen Po-Chedley, Tyler J. Thorsen, and Qiang Fu, 2015: "Removing Diurnal Cycle Contamination in Satellite-Derived Tropospheric Temperatures: Understanding Tropical Tropospheric Trend Discrepancies." J. Climate, 28, 2274–2290. doi:http://dx.doi.org/10.1175/JCLI-D-13-00767.1 (pdf here h/t E Swanson)

New Dataset from RSS: End of the Satellite “Pause”? - article by Tamino