Thread by @ClimateAudit

new Climate Audit post on Antarctic proxies used in PAGES2K (2017) with a particular teaser on their addition of two borehole inversion series which were given special exemption from PAGES2K ex ante quality control standards
climateaudit.org/2019/02/01/pag…

2/ the PAGES2K (2013) Antarctic temperature reconstruction did NOT have a Hockey Stick shape. It went down gradually for past two millennia and showed no increase in 20th century.

3/ Antarctic isotope proxies of particular interest because of their use to estimate temperature deep into the last Ice Age e.g. Al Gore's famous scene in Inconvenient Truth.

4/ yet they show decline over past two millennia. Here's a 2017 compilation from Stenni et al, using a much larger collection than PAGES2K (2013).

5/ Quality control standards for PAGES2K (2013) included strict criteria on resolution and dating. These criteria were ignored by PAGES2K 2017 in order to include two ultra low resolution and poorly dated borehole inversion series. Surprise, surprise. It had a very HS blade.

6/ several years ago, I managed to get underlying data for this series introduced by PAGES 2017 and there are some interesting puzzles. Downhole temperature curve is smooth and CONVEX. But reconstruction isn't convex: it has multiple changes in curvature direction. How????

7/ I even have some source code for the borehole inversion. They deleted the top 15 meters of data (or 17 m) ostensibly "because of the influence of the weather on surface measurements”. What do you suppose that deleted data looked like?

8/ deleted 2009 data went down ~0.3 deg moving upwards to 8 m. In 2008 data (USAP), went up about 15 deg C near surface. Measurements taken in January, austral summer. Problem isn't "weather", but annual variation. Most likely, "weather" poor choice of word, not misunderstanding.

8B/ Van Ommen et al (1999) measured near surface downhole temperature variations at Law Dome. Annual variation was ~15 deg C in top meter, with amplitude damping further down. Annual variation impacted temperature down to 15 m or so.

8C/ ice core layers can be dated by counting from surface using visual and d18O annual cycles. Layers at 15-18 m dated to 1960s. In other words, Orsi et al's conclusions about late 20th century temperature don't use layers from 1960s, but estimated diffusion to earlier layers.