CSALT Ju-Jutsu

Cowtan and Way's hybrid correction to the HadCRUT global temperature series [1]  has provoked expected interest by auditor Steve McIntyre.  This is always welcome, because as with the majority of of these nosy irritants, the more that they try to find something wrong with a well-reasoned comprehensive analysis, the more that they lay out a cookie trail for us to follow.   So guys like McIntyre make our job easier because what they try to expose backfires on them and it just gives climate scientists further substantiation of their own models -- not exactly what McIntyre had in mind.

This case is no different, starting with McIntyre's figure below:

Fig 1: Delta between CW Hybrid (basis 1961-1990) and HadCRUT4. From McIntyre http://climateaudit.org/2013/11/18/cotwan-and-way-2013/.  Note the divergence in recent years.

Let's backtrack for a moment to understand the implication of Figure 1. In previous posts on the CSALT model, what I found puzzling about the fit against the HadCRUT4 temperature series (and others)  over the past 130+ years is a recent divergence between model and data, see Figure 2 below.

Fig 2: The ln(CO2) sensitivity matches well over the entire oil age (1880-1913) temperature record apart for the recent couple of years, which is showing a potential divergence.

A log(CO2) sensitivity should display as a straight-lined slope and so if this slope starts leveling off, the transient climate response (TCR) is also falling.

Over the last 30 years, the comparison between model and data is shown in Figure 3 below, where the divergence is clear.

Fig 3:  The agreement between the CSALT model profile has started to diverge in the last few years.

The divergence is also clear in the CSALT residual plot in Figure 4 below.  One can clearly see the actual temperature dropping in comparison to the model after the year 2010.

Fig 4 :  This shows the residual between the HadCRUt4 data and the CSALT model, with a larger negative residual in the last few years.

Now the bombshell courtesy of McIntyre.  The adjustment that C&W provide by their HadCRUT4 Hybrid model happens to accurately compensate the divergence from the CSALT model.  So combining McIntyre's Figure 1 at the top of this post and the CSALT residual in Figure 4 we get Figure 5.   Note the positive divergence that McIntyre finds in the C&W Hybrid correction almost exactly compensates for the negative divergence in the CSALT model.  With a slope change that McIntyre arbitrarily sets at 2005, the positive correction balances the negative divergence so precisely that the displayed yellow line estimates close to a zero residual over the span of the last several years, and moreover, over the last 130+ years.

The puzzling behavior is solved -- as Figure 5 shows how the jigsaw pieces fit together.

Fig 5: Compensating the effect of recent movement of thermal energy away from measured latitudes (light blue data), the C&W Hybrid model corrects for this shift (black line) . The compensation results in the near-zero yellow residual line.

The current rationale for this behavior is a significant transfer of thermal energy from middle latitudes to northern and Arctic latitudes in the last several years. This is commensurate with massive Arctic sea ice melting. Yet the global  energy balance must still hold, which the CSALT model shows.

The C&W Hybrid correction performed ju-jutsu on McIntyre and he does not even realize it.  As usual, he is conveniently skilled at helping to solve jigsaw puzzles -- an unwitting ally in our understanding of AGW.

£ Donate to C&W to make their paper [1] open source.

∫ The C&W Hybrid corrections will eventually find its way into the released CSALT model served at http:/entroplet.com/context_salt_model/navigate.

Related Posts

  1. CSALT and SST corrections
  2. Detailed Analysis of CSALT Model
  3. CSALT model
  4. Climate Variability and Inferring Global Warming

References

[1] K. Cowtan and R. G. Way, “Coverage bias in the HadCRUT4 temperature series and its impact on recent temperature trends,” Q.J.R. Meteorol. Soc., p. n/a–n/a, Nov. 2013.

 

 

6 thoughts on “CSALT Ju-Jutsu

  1. Elsewhere, Marler commented

    “Your coefficient for lnCO2 has changed to 2.818. Why is that? It’s a negligible difference, so I am guessing something liike new data added to the data used for the least-squares estimation. Is it something else?”

    The inconsequential differences around 2.82+/0.01 in the figures are due to the number of data points I use in the regression and whether I put in a yearly filter. No big deal. Unless you are confused between the TCR and the coefficient of ln(CO2) which has a factor of ln(2) multiplicative difference.

    Further, the C&W Hybrid correction to HadCRUT4 will increase the value of TCR to make it closer to that of the GISS GISTEMP value.

  2. "So guys like McIntyre make our job easier because what they try to expose backfires on them"

    Like his deconstruction of Mann's Hockey Stick, yes?

    • The Marcott hockey stick?
      The PAGES 2K hockey stick?

      Science is self-correcting. McIntyre tried to expose the hockey stick temperature trend as false yet it is just as strong now as ever. Perhaps that specific finding didn't back-fire on him but he certainly didn't invalidate the concept, if that was his objective.

      • You what? He exposed junk science for what it is. Even the IPCC, world leaders in sponsoring junk science, have quietly junked Mann's hockey stick. And the Marcott paper itself states that the hockey stick in their graph is "not robust".

        We must be living on different planets.

        • The reality is that average temperatures have not changed much over the course of centuries. That is the flat handle of the hockey stick. In the last 130 years, land temperatures have risen by 1.2C degrees. That is the blade of the hockey stick.

  3. Pingback: CSALT with CW Hybrid | context/Earth

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Optionally add an image (JPEG only)