http://www.bishop-hill.net/blog/2012/11/26/lonely-old-mann.html

Comments certainly illustrate some of the variety of interests and agendas people bring to these discussions.

http://climateaudit.org/2010/02/26/the-trick-timeline/

The problem of divergence is perhaps a little more complicated than that. Depending on how you calibrate the MXD (linear relationship), the difference may seem to appear at any particular date. As I said earlier, for the full Schweingruber collection (http://noconsensus.files.wordpress.com/2009/11/before-and-after1.jpg?w=553&h=368) this divergence seems to appear as early as 1920 and not 1960.

“But still, while it’s true that even instrumental temperatures have potential problems, “bodging” the instrumental series to fit MXD is a novel way of hiding a divergence!”

That is the question, hide a MXD decline or correct an instruments bias ?

Take the issue from another angle : the melting anomalies are they a good proxy for temperature ?

Huss uses in his paper the station of Davos for reference, he compares the melting anomalies with homogenized temperatures. It should be noted that homogenization adjustments allow to obtain a good signal for high frequency to the detriment of long-term trends. However, this is what gives the comparison of melting anomalies with Davos raw temperatures : http://img837.imageshack.us/img837/5687/fontea.jpg

The curve “homogénéisation partielle” take correctly into account two adjustments (but not all).

Melting anomalies are in fact a good proxy for temperatures. And the decline which is hidden is the temperatures one.

I see now that I misread the graph in my previous post — the orange Schweingruber MXD series continues to about 1994, and so is not truncated.

It’s odd that this narrowly Swiss MXD series does not tank after 1960 the way Briffa’s broader series does. Instead, its “divergence problem” is entirely pre-1940.

But still, while it’s true that even instrumental temperatures have potential problems, “bodging” the instrumental series to fit MXD is a novel way of hiding a divergence!

Thank you for the clarification in english. I am very bad in that language but I’ll try to clarify few points (with some help of google).

Correcting the instrumental series rather than MXD and melt anomaly is quite normal for the following reasons:

1. Of the 3 series, two are consistent. It is therefore more logical to correct one set than two.

2. We have no specific reasons to correct the melt anomalies and MXD while we have a good quantity which encourages us to do so for instrumental data. Thinking a singular event caused the fall of MXD since 1960 is very strange and unsupported. The explanation of Huss et al. for the divergence of melting by shortwave is also vague and unproven. In addition, the physical phenomena involved in both cases are very different and it would be surprising that bias of completely different nature affect parallely trees and glaciers.

3. Official data are adjusted upward by 0.5 ° C per century to remove discontinuities. These adjustments should be broadly neutral and we have no reasonable explanation for this bias. These data are not reliable.

4. We know that disruption of the stations (UHI and local effects of urbanization) are not taken into account but have a significant effect.

5. UAH (and RSS) TLT for land in the northern hemisphere trend is so colder than 1 °C per century over CRUTEM while the hot spot effect should cause an opposite trend.

I used the example of Switzerland for the wonderful proxy of Huss and al but the issue is the same at least for all land in the northern hemisphere (comparison CRUTEM HN and Schweingruber entire collection). The only difference is that the divergence is 1 °C per century since 1920.

Note : the curves are weighted averages over 10 years.

phi
Posted Apr 30, 2011 at 4:14 PM
Here is another interesting graph : http://img4.imageshack.us/img4/7853/anomch.jpg

Anomalies de fonte (3 glaciers) : Huss et al. 2009. Densité des cernes : 70 swiss MXD series from Schweingruber.

Very interesting — The graph is titled april-sept temperature anomalies for Switzerland. The red line is the official instrumental series, the orange line ending around 1970 is “ring density,” presumably the 70 Schweingruber Swiss MXD series mentioned by phi, the green line is “melt anomaly”, presumably from the Huss article on glacial melting, and the blue line is the “instrumental series with a correction of -1.5dC per century since 1890″.

Is this your own graph, phi, or is it from Huss or some other souce?

Anyway, the red instrumental temperatures don’t correlate well with the orange MXD series by themselves, so the blue line “corrects” the instrumental series with a constant time trend.

First, it’s odd to be correcting the thermometer record to conform to the MXD (and melt) record rather than the other way around. The melt record only boes back to about 1920, but presumably MXD goes back centuries if not millenia. If the early portion of MXD is fairly flat, this “correction” implies that mercury thermometers that hadn’t been calibrated to treerings would have been reading about 15dC lower in 1000 AD than during the 20th century.

Alternatively, if the instrumental series is taken as the more valid, a flat MXD series would imply that actual temperatures in 1000 AD were about 15dC warmer than during the 20th century. A Medieval Sauna Period!

The glaring divergence between MXD and the actual temperature series could be at least partially explained without an ad hoc time trend by introducing CO2 into the relationship. But then there would be a lot more uncertainty to the correlation with temperature.

The stronger correlation of the green melt series to the orange MXD series than to the red instrumental series is perplexing. Perhaps melt responds to long-wave GHG radiation differently than to solar radiation or to air temperature itself? Just a guess.

Back to the topic of this thread, does the orange MXD series really end in 1970, or was it truncated there per Keith’s Science Trick to Hide an inconvenient Decline?