A recent study, Richey et al 2007, showed a warm MWP in the Pigmy Basin, Gulf of Mexico. Julie Richey made an effective presentation at the AGU Session showing an elevated MWP from her foram studies and, together with Alicia Newton who made a similar presentation, had Mann gnashing his teeth. In a chat afterwards, Richey said that they had faced unusual opposition from one reviewer who threw many roadblocks towards publishing their results – something that was unprecedented in their experience in presenting empirical results from a qualified laboratory. Richey’s results were recently mentioned by poster at Tamino, causing Tamino (an interesting writer when he’s calm) to have one of his all-too-frequent temper tantrums, saying that he wasn’t interested in regional results, although he seems all too content to rely on composites of small numbers of much more flawed data (e.g. Moberg’s dependence on the much more questionable Arabian Sea G Bulloides series.)
I thought that it would be interesting to compare her results (also promptly archived both at the journal and at WDCP) to David Black’s Cariaco results. In a way, you’d like to see this sort of comparison in the original articles. Black discusses comparisons between the Cariaco results and 4 or 5 other series, but doesn’t illustrate them. I understand the space limitations of journals and don’t fault Black for not illustrating the comparisons, but nonetheless I think that doing some of these comparisons not done in the articles is a useful form of perspective. Note that it depends on prompt archiving of data. It’s really nice to read one of these articles, look for relevant comparanda and be able to get a perspective on them in an hour or two without months of quasi-litigation as in Hockey Team World (Mann-Bradley-Hughes-Briffa-Jones-Esper).
I’m going to show both dO18 and Mg-CA, showing the dO18 comparison first (also showing in this case Keigwin’s Sargasso Sea results.) The most obvious difference between these series is the difference in texture. The varved Cariaco series with annual to near-annual resolution has much higher resolution than the decadally resolved Pigmy Basin Basin series, which in turn is much more highly resolved than the relatively early (!996) Keigwin Sargasso Sea series, that is often used by people wanting to show an MWP (we used it in our presentation to NAS showing that, by “apple-picking”, you could get an MWP warmer than modern proxies that, in our opinion, was neither more or less convincing that cherry-picking results selecting series to show modern proxies higher than MWP.)
Can one then locate a “signal”? Here you see why paleoclimatologists work so hard to extract patterns from tree rings where, at least, you can have confidence in your dates. If you allow for dilation of the age scale in the Pigmy Basin series (which assumes a constant sedimentation rate from radiocarbon – fair enough as an assumption but which will dilate the scale if sedimentation rates vary as they undoubtedly have), I can persuade myself right now that the 1400-1800 pattern in the Pigmy Basin more or less matches the 1450-1800 pattern in the red Cariaco series, so that the signal is enhanced if the dates of Pigmy Basin are tuned to Cariaco. But maybe the dates are right and there are regional differences in signal. Without tuning, the Pigmy Basin series is completely uncorrelated to the smoothed G Bulloides dO18 series (r=0.00), but the smoothed G Bulloides dO18 series has a low correlation to the smoothed G Ruber series (0.06) as well. The Richey series has a correlation of 0.20 to the smoother Cariaco ruber series. Here’s the graphic – some more comments after the graphic.
The modern comparisons are also rather interesting. Black et al 2004, which discussed these results, presented two very different possible explanations for their dO18 results:
First, the increase in d18O may indicate that tropical summer-fall SSTs have cooled by as much as 2 deg C over the last 2000 years, possibly as a result of a long-term increase in upwelling intensity. Alternately, comparisons to other studies of ITCZ and regional evaporation/precipitation variability suggest that much of the d18O record is influenced by decadal- to centennial-scale variations in the mean annual position of the ITCZ and associated rainfall patterns.
The Pigmy Basin results, while not as resolved, like Cariaco, show “cold” results through the 19th century, with a 20th century increase. One feels that there is some information here, but it’s hard to say what it means.
Both Black and Richey also present Mg-Ca results – a relative new technique that is said to be less confounded by precipitation. I showed a graphic of Black’s temperature reconstruction the other day; here’s it is compared to the Richey reconstruction. Again one can persuade myself that there is some similarity between the series – one’s eye wants to de-align the Richey series a little, tuning it to the Cariaco series: is such tuning justified given the known dating imprecision of the Pigmy Basin series? And what are the statistical implications of such tuning? Not easy questions. If you tune the Richey series to Cariaco, it seems to me that this will bring the warm “MWP” portion of her series a little closer to the present, probably moving it to cover the 11th century known to be warm in northern Europe. (Without tuning, the correlation between the Richey series and the smoothed Cariaco series is 0.15).
One obvious difference between the two series is their core top dating. Richey’s series ends at 0 BP (thus 1950 AD) while Black’s comes to 1990 AD. Richey’s data is more or less decadally resolved and the wiggle match to Cariaco is definitely improved if the core top on her series is dated much closer to the end of Black’s series. Can this be justified? It doesn’t seem inconsistent with the evidence.
On the other hand, if you don’t tune, you quickly lose any sense of regional pattern. One series goes down while another is going up. In IPCC AR4, they show a bunch of series like this and say – aha, there’s no MWP. While this is possible, you could also get a dog’s breakfast if you have variable dating imprecisions – a point that Craig Loehle has argued for a number of years, but not mentioned by IPCC. The cases shown here are relatively clean examples of the problem – do you tune or not. And if there is no common signal at this resolution between such similar proxies, what hope is there for extracting a common signal from small populations (10-12 series) of proxies in a typical Team study.
I received some information yesterday from David Lund and William Curry on their sediment studies in the Dry Tortugas and Bahamas, which I will try to compare to this next week. When I inquired to William Curry, he responded promptly, expressing some surprise that the data had not been posted up at WDCP and immediately arranged for Lund to send me the data which he did a few hours later, together with a very cordial invitation to telephone him with any questions – a far cry from the typical quasi-litigation with the Hockey Team for scraps of data. The oceanographers seem to be a very cooperative group – I’ve had very prompt and friendly responses from Lowell Stott, Julie Richey and Alicia Newton as well.