New isotope based temperature reconstruction using McIntyre's 'Starbucks hypothesis' tree core samples

A new paper in GRL published Sep 6th Secular temperature trends for the southern Rocky Mountains over the last five centuries makes use of some tree core sample data gathered by Steve McIntyre and Mr. Pete. Readers of WUWT and Climate Audit may recall that in the summer of 2007, Steve left CA in my attendance for a couple of weeks while he went to Colorado to visit his sister, and to prove or disprove his ‘Starbucks hypothesis’ which asks:

…could a climate scientist have a Starbucks in the morning, collect tree rings through the day and still be home for dinner?

This came about because apparently RealClimateScientists™ don’t have the funds or time to get out of the office and gather new tree core samples, such as cores that would fill in the last 25 years that seems to be part of that “tricky” divergence problem. In A Little Secret (Oct 2007) Steve wrote:

Don’t you think that someone on the Team might have been a little curious as to what bristlecone ring widths have done during the past 25 years? For this, we have the classic excuse of Michael Mann and the Team for not updating bristlecone and proxy records is that it’s not practical within the limited climate budgets:

While paleoclimatologists are attempting to update many important proxy records to the present, this is a costly, and labor-intensive activity, often requiring expensive field campaigns that involve traveling with heavy equipment to difficult-to-reach locations (such as high-elevation or remote polar sites). For historical reasons, many of the important records were obtained in the 1970s and 1980s and have yet to be updated.

This new paper proves that you can do field science on vacation, while visiting Starbucks, and in a single day. It also has a thing or two to tell us about the accuracy of tree ring width and wood density records and their value as a temperature proxy.

First, the calibration of the d18Oc data with instrumental temperature data from the nearby Cheesman USHCN station.

Abstract:

Pre-instrumental surface temperature variability in the Southwestern United States has traditionally been reconstructed using variations in the annual ring widths of high altitude trees that live near a growth-limiting isotherm. A number of studies have suggested that the response of some trees to temperature variations is non-stationary, warranting the development of alternative approaches towards reconstructing past regional temperature variability. Here we present a five-century temperature reconstruction for a high altitude site in the Rocky Mountains derived from the oxygen isotopic composition of cellulose (d18Oc) from Bristlecone Pine trees. The record is independent of the co-located growth-based reconstruction while providing the same temporal resolution and absolute age constraints. The empirical correlation between d18Oc and instrumental temperatures is used to produce a temperature transfer function. A forward-model for cellulose isotope variations, driven by meteorological data and output from an isotope-enabled General  Circulation Model, is used to evaluate the processes that propagate the temperature signal to the proxy. The cellulose record documents persistent multidecadal variations in d18Oc that are attributable to temperature shifts on the order of 1C but no sustained monotonic rise in temperature or a step-like increase since the late 19th century. The isotope-based temperature history is consistent with both regional wood density-based temperature estimates and some sparse early instrumental records.

Berkelhammer, M.,and L. D. Stott (2012), Secular temperature trends for the southern Rocky Mountains over the last five centuries, Geophys. Res. Lett., 39, L17701, doi:10.1029/2012GL052447.

During the 20th century, the summer surface temperatures in this region are characterized by a broadly parabolic trend, with minima during the 1930s and early 1980s

and a period of relative warmth during the late 1940s to early 1960s (Figure 1). The temperature reconstruction based on d18Oc suggests that in terms of mean temperature and multidecadal variance, the 20th century is largely comparable to

the preceding 4 centuries (Figure 3).

Despite the seemingly good correspondence between tree ring width (Figures 3 and S5) and d18Oc proxies during the instrumental period, over much of the previous 400 years

the two records exhibit very different climate histories (Figure 3). Prior to the mid 19th century the width-based reconstruction indicates temperatures at this site were

approximately 0.7C cooler than during the instrumental era while the d18Oc reconstruction suggests that temperatures have remained stable. The residual between these two reconstructions (Figure 3) is sufficiently large and sustained to suggest the existence of a significant bias in one or both of these two proxies that cannot likely be explained as arising simply from random errors in the linear transfer function.

Daux et al. [2011] also note an apparent divergence between width and isotope based temperature reconstructions in Larix decidua from France. They attribute the divergence

possibly to changes in the soil hydrology (i.e., plant utilization of soil water enriched by evaporation) or moisture stress. At this site, soils are thin and the trees are characteristically shallow-rooted and it is thus unlikely that deeper, low-residence time water would be available. Further no indication of anomalous 20th water stress or abundance is seen in either d13C from pinyon pine trees across the region [Leavitt et al., 2007] or widths from lower elevation drought-stressed trees [Cook et al., 1999].

To help resolve this enigma, an additional temperature proxy that is based on a regional composite of wood density measurements is considered [Briffa et al., 1992] (Figures 3 and S5). This temperature proxy is independent of both tree growth rate and the isotopic composition of cellulose and is shown to have high skill as a growing season temperature proxy in this region (Figure S5). To test the consistency between density and isotope-derived temperatures we look at the cross-wavelet [Grinsted et al., 2004] between the records (Figure S6). In the multi-decadal window, the density and isotope reconstructions are consistently in-phase with one another through the last 400 years, implying that the two proxies are likely being influenced by a common climate parameter, which we assume to be growing season surface temperature variations. With respect to the cross-wavelet between widths and isotopes, the two appear to only be commonly forced during the 20th century (Figure S6).

Further confirmation of this is garnered by looking at early instrumental data from the region (not shown), which indicate that surface temperatures between 1850–1870 were, on average, as warm as those of the 1930s–1960s [Wahl and Lawson, 1970], which is consistent with the relative thermal stability implied by the isotopic and

density reconstructions. Taken together, the d18Oc and wood density records provide a fairly consistent perspective on multidecadal temperature variations, which suggest a cool bias in the width-based temperatures prior to the mid 19th century.

Conclusions

The isotope temperature record from this site indicates relatively stable summer season temperatures amidst decadal to multidecadal temperature fluctuations. Although

the isotope reconstruction is associated with several significant sources of uncertainty that arise from the transfer function and tree-to-tree heterogeneity, the results highlight the need for, 1) additional efforts to extend a processbased network of temperature reconstructions across the region and 2) develop pre-instrumental forward model simulations (for both widths and isotopes) that could be used to test the assumptions of linearity that underlie the proxy reconstructions.

The main points of the paper are:

• Temp. trends in the SW US can be reconstructed using isotopes in tree rings
• A process model of the proxy can be used to characterize uncertainty in proxy
• Temperature trends in SW US have been relatively stable over last 5 centuries

Acknowledgments. The authors thank M. Zhu, G. Kleber and

M. Rincon for invaluable assistance in sample preparation and analysis;

Z. Gedalof and J. Franks for cross dating the samples used in this analysis;

V. Bommarito, L. Holzmann, P. Holzmann, R. Lee, N. McIntyre S.

McIntyre, L. Thomas for sample collection; A. Ballantyne for feedback

on an earlier version of the paper; K. Yoshimura for providing outputs

from the IsoGSM simulations; 2 anonymous reviewers for suggestions

on improving the manuscript and the ITRDB for tree ring data. Funding

was provided by NOAA Award NA10OAR4310129 to LDS.

This effort, peer reviewed paper, and results just goes to show that citizen science can do what RealClimateScientists™ can’t or won’t, and do it just as effectively. For those who worry about such things, it should be noted that Steve applied for, and got permission for the core sampling of the Bristlecone pines in Colorado.

h/t to Dr. Leif Svalgaard who has the full paper on his website: http://www.leif.org//EOS/2012GL052447.pdf