Climatic Research Unit : Papers

Reassessing the evidence for tree-growth and inferred temperature change during the Common Era in Yamalia, northwest Siberia

Keith R. Briffa, Thomas M. Melvin, Timothy J. Osborn, Rashit M. Hantemirov, Alexander V. Kirdyanov, Valeriy Mazepa, Stepan G. Shiyatov and Jan Esper
Quaternary Science Reviews (2013) 72, 83-107. doi: 10.1016/j.quascirev.2013.04.008

This is an open-access publication: anyone clicking the above "doi" link can read our article, no need for a subscription.

2016-03-09 -- minor correction issued, see below for details

Abstract

The development of research into the history of tree growth and inferred summer temperature changes in Yamalia spanning the last 2000 years is reviewed. One focus is the evolving production of tree-ring width (TRW) and tree-ring maximum-latewood density (MXD) larch (Larix sibirica) chronologies, incorporating different applications of Regional Curve Standardisation (RCS). Another focus is on comparison of independent data representing past tree growth in adjacent Yamalia areas: Yamal and Polar Urals, and the examination of the evidence for common growth behaviour at different timescales. The sample data we use are far more numerous and cover a longer time-span at Yamal compared to the Polar Urals, but Yamal has only TRW, while there are both TRW and MXD for the Polar Urals. We use more data (sub-fossil and from living trees) than in previous dendroclimatic studies in this region. We develop a new TRW chronology for Yamal, more than 2000 years long and running up to 2005. For the Polar Urals we develop new TRW and MXD chronologies that show good agreement at short (<15 years) and medium (15 to 100 years) timescales demonstrating the validity of attempts to reconcile the evidence of longer-timescale information that they provide. We use a "conservative" application of the RCS approach (two-curve signal-free RCS), guarding against the possibility of "modern sample bias": a possible inflation of recent chronology values arising out of inadvertent selection of mostly relatively fast-growing trees in recent centuries. We also transform tree indices to have a normal distribution to remove the positive chronology skew often apparent in RCS TRW chronologies. This also reduces the apparent magnitude of 20th century tree-growth levels.

There is generally good agreement between all chronologies as regards the major features of the decadal to centennial variability. Low tree-growth periods for which the inferred summer temperatures are approximately 2.5°C below the 1961–90 reference are apparent in the 15-year smoothed reconstructions, centred around 1005, 1300, 1455, 1530, particularly the 1810s where the inferred cooling reaches –4°C or even –6°C for individual years, and the 1880s. These are superimposed on generally cool pre-20th century conditions: the long-term means of the pre-1900 reconstructed temperature anomalies range from –0.6 to –0.9°C in our alternative reconstructions. There are numerous periods of one or two decades with relatively high growth (and inferred summer temperatures close to the 1961–1990 level) but at longer timescales only the 40-year period centred at 250 CE appears comparable with 20th century warmth. Although the central temperature estimate for this period is below that for the recent period, when we take into account the uncertainties we cannot be highly confident that recent warmth has exceeded the temperature of this earlier warm period. While there are clear warm decades either side of 1000 CE, neither TRW nor MXD data support the conclusion that temperatures were exceptionally high during medieval times. One previous version of the Polar Urals TRW chronology is shown here to be in error due to an injudicious application of RCS to non-homogeneous sample data, partly derived from root-collar samples that produce spuriously high chronology values in the 11th and 15th centuries. This biased chronology has been used in a number of recent studies aimed at reconstructing wider scale temperature histories. All of the chronologies we have produced here clearly show a generally high level of growth throughout their most recent 80 years. Allowing for chronology and reconstruction uncertainty, the mean of the last 100 years of the reconstruction is likely warmer than any century in the last 2000 years in this region.

Availability of Yamalia analysis products

The principal outcomes of this analysis are a number of tree-ring width (TRW) and maximum latewood density (MXD) chronologies, and several alternative reconstructions of summer temperatures, all as described in the main text. Depending on the particular focus of interest we provide the following:

Chronologies

Calibrated temperature reconstructions

[Note that these are also available from the World Data Center for Paleoclimatology.]

Recommended acknowledgement

When using any of these data, please cite:

Briffa K.R., Melvin T. M., Osborn T. J., Hantemirov R. M., Kirdyanov A. V., Mazepa V., Shiyatov S. G. and Esper J. (2013) Reassessing the evidence for tree-growth and inferred temperature change during the Common Era in Yamalia, northwest Siberia. Quaternary Science Reviews 72, 83-107. doi: 10.1016/j.quascirev.2013.04.008

Correction issued (2016-03-09)

Error. The data values were misaligned in one line of the raw data file raw/polar/purla/purlasi.mxd because of two spurious characters. This error affected only the 1970s decade of maximum latewood density (MXD) measurement values for one tree core (id 148 211).

Consequences. This resulted in spurious values being read in for that tree core, with knock-on effects throughout the study but with diminishing effects as the data for this tree core were aggregrated with additional data (first from a second core from the same tree, then with data from other trees sampled at this site at the same time, then from trees sampled during other fieldwork at this site and at nearby sites). The influence of this error is most noticeable when the purlasi site is considered in isolation (SM4 Fig. PC13). It affects particularly the 1973 and 1976 values but also has a small effect throughout the chronologies because (a) the RCS curves used to standardise MXD data are slightly affected, and (b) the standard deviation of the data at this site is affected, which influences the rescaling step used to combine MXD data from different sites (rescaling is necessary to remove systematic differences between MXD measurements from different equipment/times -- see SM4).

We have compared the corrected results with those originally published and confirm that the error diminishes as more data are aggregated. The transformation of tree indices to follow a normal distribution is particularly effective at reducing the sensitivity of the final chronology to these outlier values, confirming the benefits of this procedure (introduced in section 4.4 of the main paper).

The error does not affect the Yamal, Polar Urals or combined (Yamalia) tree-ring width (TRW) chronologies, which provide the main basis for the long-term summer temperature changes reported in the main paper. The figures in the main paper that depend on the MXD data are hardly affected (see a visual comparison). We have re-evaluated our main findings reported in the paper and none need to be modified.

Correction. New versions of SM3, SM4, SM5, SM6 and SM7 are provided with the affected figures and tables replaced. A new version of the final Polar Urals MXD chronology is provided in the chronologies data file (PY_Chrons.prn) in the box above and in the data files section below. Previous versions are retained for comparison (see a visual comparison). While issuing this correction, we noticed that some figures (SM5 Fig. PY06, PY09, PY14-16) were based on the polarx.mxd compilation (excluding some root collar samples) instead of the intended polarxs.mxd compilation, so these have been corrected as well.

Acknowledgement. We thank Steve McIntyre for identifying the data error in his Climate Audit blog.

Supplementary Material

Extensive supplementary material is provided to document in detail the tree-ring data, the temperature data, and the multiple stages of data processing and analysis. This material is organised as nine documents (SM1 to SM9). These documents, together with data files containing raw measurements, cross-dating reports, some figure datasets, adjustment logs, chronologies and reconstructions, are available here.

Supplementary Material Documents

Additional description, results and figures are structured into nine Supplementary Material sections. These documents are identical to the supplementary material available at the journal's website.

Data Files

Each zip file contains multiple text files with data and reports, together with a readme.txt file that describes the contents of each text file.

The temperature reconstruction data (including estimates of uncertainty) are also available from the World Data Center for Paleoclimatology, as are the tree-ring measurement data (new data used in our study can be located by searching for "Yamalia" at the WDC Paleo search page).

Source Code

The tree-ring chronologies were constructed using the CRUST (CRU Standardisation of Tree-ring data) software. A separate article has been published that describes the processing methods that CRUST implements, and the full CRUST software (including source code) is now available.

We have created a reduced-functionality version of the CRUST software that is specific to the Yamalia paper (i.e. it does the specific processing necessary to construct the tree-ring chronologies, undertake the various analyses reported in the paper, and to create most of the figures shown in the paper).

This Yamalia version of CRUST is available in this corrected 2016-03-09 zip file (replaces this original zip file). It contains all the raw data, together with executable versions for Linux and Windows. It also includes the Fortran source code, from which it can be compiled. We are not able to offer any technical support, but it does include a "makefile" to assist with compiling the software (though we recommend using the pre-compiled, executable versions if they work on your computer system).

The full-functionality version of CRUST is available from the CRUST software page.

Amendment record