Global Mean Temperature Flattens the Past

Introduction There have been recent discussions about ‘flattening the curve’ and some curves are easier to flatten than others. The Pages 2K Consortium calculates global mean temperature in a manner that flattens the long-term trend and makes present day temperatures appear warmer relative to past temperatures. Across the globe, temperature reconstructions show cooling millennial temperature trends with one exception, the Pages 2K global mean. Millennial Temperature Trends Show Global Cooling Global mean surface temperature anomalies were recently calculated by the Pages 2K Consortium led by Nuekom, 2019. Their statistical means are a conglomeration of seven different averaging methods for 7000 proxy records over the past 2000 years. The median across all global mean methods is plotted as a dashed line in Figure 1 and compared to Pages 2K’s published regional reconstructions. All means demonstrate similar trends as the median and will be simply be referred to as the global mean(s). Regional temperature reconstructions are chosen that utilize similar proxy datasets used in the global mean calculation. The Arctic reconstruction by McKay incorporates a balance of proxy records consisting of ice cores, tree rings, lake and marine sediments north of 60 deg N. The Northern Hemisphere (NH) European reconstruction by Luterbacher is tree ring proxy based. And Stenni’s Antarctic reconstruction uses predominantly ice core isotopes. The Pages 2K global mean appears to be reasonable compared to regional reconstructions from Present through the Little Ice Age (LIA) until about 1250 AD. Although it is difficult to see how the mean compares to regional reconstructions during the Present when using a 1961-1990 baseline as all reconstructions converge creating the “hockey stick” effect. Pre-1250 AD, the global mean appears to parallel NH Europe temperatures largely ignoring the Antarctic.

Figure 1: Top graph are surface temperature reconstructions with a 50-year loess filter plotted with Pages 2K global mean of the 7000-member ensemble across all methods. Bottom graph shows linear trends over the past 2000 years.
Linear regression analysis of the temperature reconstructions in Figure 1, bottom graph, shows cooling trends over the past 2000 years. Surprisingly, both the Arctic and the Antarctic show a similar long-term cooling trend of -0.4 deg C per 1000 years. As a matter of fact, all regional reconstructions show a negative slope, or cooling trend, in temperature anomalies during the past 2000 years shown in Table 1. Interestingly, all the global means are nearly flat or show a subtle cooling trend. The global mean cooling trend is more aligned with the NH Europe temperature reconstruction. Note the 97.5% global mean cooling of -0.2 deg C per 1000 years is still flatter than both the Arctic and Antarctic mean cooling trend of -0.4 deg C per 1000 years. Also, the global mean low 2.5% range of -0.04 deg C per 1000 years is much flatter than the low range of any regional reconstruction.
Table 1: Millennial trends of spatial temperature reconstructions over the past 2000 years compared to Pages2K global means. Means and ranges are from McKay for the Arctic, Stenni for Antarctic, Luterbacker for Europe, and Neukom for global means.
Of the regional temperature reconstructions, the NH Europe mean millennial trend shows the least amount of cooling during the past 2000 years of only -0.20 deg C per 1000 years. This temperature reconstruction consists entirely of tree ring proxy data. There is a notable shift in data quantity and quality of tree ring datasets around 1000 AD. The number of tree ring records are reduced significantly from 400 records post-1600 AD to less than 30 records pre-1000 AD (Luterbacher, 2016). McKay reports an Arctic cooling trend of -0.47 deg C per 1000 years during the period 0 to 1900 AD. The cooling trend reported here is for the period 0 to 2000 AD and includes the Present. Including the Present slightly reduces the Arctic millennial cooling trend from -0.47 to -0.40 deg C per 1000 years. As expected, the Arctic reconstruction with large centennial temperature swings shows the highest spread of millennial trends ranging from -0.10 to -0.70 deg C per 1000 years. Stenni, 2017, shows cooling trends ranging from -0.30 deg c per 1000 years for the East Antarctic Plateau to -0.52 deg C per 1000 years for the Antarctic Peninsula during 0-1900 AD. She breaks out the last 100 years separately which shows the higher frequency or shorter-term centennial warming of the Present. Including the Present slightly increases the range of the Antarctic millennial cooling trend. The E. Antarctic is the last place on Earth where the Present centennial warming has occurred. This delayed warming is not captured by climate models which tend to overestimate Antarctic warming (Stenni, 2017). Global Mean Falls outside of the Arctic Antarctic Envelope As discussed in my previous post, I prefer using the LIA 1600-1700 AD as a baseline rather than the 1961-1990 baseline for extended temperature reconstructions. Using the LIA baseline maintains convergence of temperatures during the cold LIA and temperature divergences between the Arctic and Antarctic during warmer periods. It allows the Medieval Warm Period (MWP), Roman Warm Period (RWP) and LIA climate events to be prominently visible on the Arctic data shown in Figure 2. Additionally, these Polar regions are placed in a proper climate context with the Antarctic showing colder temperature anomalies than the Arctic. General observations show the MWP and RWP to have a peak Arctic temperature like the 1940 Present peak. All three peaks are approximately 1.3 deg C warmer than the LIA baseline. In contrast, Antarctic temperatures are 0.25 to 0.50 deg C warmer during the MWP and RWP than Present. The global mean is basically flattened backwards in time by not incorporating the underlying millennial cooling trends. When reconstructions are datumed on the LIA, the global mean falls outside the Arctic and Antarctic envelope pre-1250 AD. From 0 to 1250 AD, the mean shows colder global temperatures than even the Antarctic. A simple difference in average temperature between the LIA and reconstructions from 0-1000 AD is revealing. It shows the global mean with a slight increase of only 0.25 deg C warming prior to the LIA in contrast to both the Antarctic and Arctic which show increases of 0.5 and 0.8 deg C warming, respectively, prior to the LIA. The global mean appears reasonable during the Present.
Figure 2. Antarctic and Arctic temperature reconstructions plotted with Pages 2K Global Mean relative to a 1600-1700 baseline. Temperature reconstructions filtered with a 50-year Loess. Bottom graph is a zoom in which shows linear trends from the MWP to the LIA. Reconstructions are filtered with a 30-year loess.
As expected, the millennial underlying trend is transient with time. As an example, the MWP cooling descent into the LIA is faster than the cooling over the past 2000 years, bottom graph in figure 2. The Arctic shows a cooling of -1.1 deg C per 1000 years and the Antarctic is cooling at a rate of -0.6 deg C. Unbelievably, the Pages 2K global mean shows a LIA cooling rate of only -0.2 deg C per 1000 years that is even slower than the Antarctic. The global mean flattening effect reduces the temperature anomaly of warm periods prior to the LIA and does not properly preserve the LIA cooling trend. Global Mean is Biased by Tree Ring Proxies The data used by Pages 2K to calculate the global means is based on 7000 proxy records. However, the majority (59%) of the records are tree rings which are located primarily in the Northern Hemisphere (Pages 2K, 2017). Nuekon, 2019, acknowledges that tree ring records are detrended and therefore, do not capture centennial and multi-centennial trends. They also do not retain longer-term millennial trends. Furthermore, he confirms this problem compounds backwards in time and results in underestimation of low-frequency variability especially during the first millennium of the Common Era. This would be during warm period analogs such as the MWP and RWP. The Pages 2K global mean calculations are driven by NH proxy data with an overemphasis on tree ring proxies which is the primary reason for the flattening in the past. Christainsen et. al, 2017, has an excellent analysis and discussion on the lack of preservation of low-frequency or longer-term variability in proxy records and large-scale temperature reconstructions. He states that tree ring records have absolute annual dating control and can be cross dated with other chronologies. However, he confirms tree ring data has problems related to preserving the very low frequencies and longer-term trends. Additionally, he states that averaging proxies acts as a low-pass filter resulting in the signal being “flattened out,” thus preventing the true magnitude of cold and warm periods in temperature reconstructions from being captured. Both issues apply to the global mean across the RWP and MWP and therefore, should not be directly compared to the Present centennial warming in absolute temperature terms. Additionally, pre-1000 AD tree ring records are reduced in data quantity and quality. For an objective review of Pages 2K tree ring proxies, I recommend reading Steve McIntyre’s articles. He discusses the accuracy of tree ring data, the divergence problem and cherry picking of data. Conclusions Pages 2K global mean published in 2019 does not capture the millennial cooling trend observed in Arctic and Antarctic regional temperature reconstructions. Their global mean relies on a database biased with Northern Hemisphere tree ring proxies which do not preserve long-term temperature trends of the polar regions. The overall effect of the Pages 2K dataset and mean is to flatten temperature trends backwards in time, especially during the RWP and MWP which are key present-day analogs. The cooling descent into the LIA is largely removed. Warmer Arctic and Antarctic temperatures during the RWP and MWP are minimized and not represented by the global mean temperature. Thus, the Pages 2K Consortium has flattened the global mean temperature profile in the past. Acknowledgements: Special thanks to Donald Ince and Andy May for reviewing and editing this article. References Cited: Christiansen, B. & Ljungqvist, F. C. Challenges and perspectives for large-scale temperature reconstructions of the past two millennia. Rev. Geophys. 55, 40–96 (2017). https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2016RG000521 Luterbacher J et al. European summer temperatures since Roman times. Environmental Research Letters 11, 024001, DOI: 10.1088/1748-9326/11/2/024001, 2016. McKay, N. P. and Kaufman, D. S.: An extended Arctic proxy temperature database for the past 2,000 years, Scientific Data 1:140026, doi:10.1038/sdata.2014.26, 2014 Dataset: https://www1.ncdc.noaa.gov/pub/data/paleo/pages2k/arctic2014temperature-v1.1.txt McIntyre, S. Climate Audit blog. https://climateaudit.org/?s=Pages PAGES 2k Consortium: Continental-scale temperature variability during the past two millennia, Nat. Geosci., 6, 339–346, Published online 21 April 2013, https://doi.org/10.1038/NGEO1797, 2013.1c PAYWALLED. Dataset available see above. PAGES 2k Consortium- Neukom, R., Barboza, L.A., Erb, M.P. et al. Consistent multidecadal variability in global temperature reconstructions and simulations over the Common Era. Nat. Geosci. 12, 643–649 (2019). https://doi.org/10.1038/s41561-019-0400-0. Paywalled, but shared by the author at the following link. http://pastglobalchanges.org/science/wg/2k-network/nature-geosc-2k-july-19 Stenni, B., Curran, M. A. J., Abram, N. J., Orsi, A., Goursaud, S., Masson-Delmotte, V., Neukom, R., Goosse, H., Divine, D., van Ommen, T., Steig, E. J., Dixon, D. A., Thomas, E. R., Bertler, N. A. N., Isaksson, E., Ekaykin, A., Werner, M., and Frezzotti, M.: Antarctic climate variability on regional and continental scales over the last 2000 years, Clim. Past, 13, 1609–1634, https://doi.org/10.5194/cp-13-1609-2017, 2017. Temperature Reconstruction Datasets Arctic McKay, 2014. https://www1.ncdc.noaa.gov/pub/data/paleo/pages2k/arctic2014temperature-v1.1.txt Antarctic Stenni, 2017. https://www1.ncdc.noaa.gov/pub/data/paleo/pages2k/stenni2017antarctica/CPSrecons/All_regions_recons_CPS.csv Europe Luterbacher, 2016. https://www1.ncdc.noaa.gov/pub/data/paleo/pages2k/EuroMed2k/eujja_2krecon_nested_cps.txt SH Nuekom, 2014. https://www1.ncdc.noaa.gov/pub/data/paleo/contributions_by_author/neukom2014/SH_Fig2_recons_Ens-means_wrt1000-2000.txt Pages 2K 2013 dataset. www.ncdc.noaa.gov/paleo/pages2k/pages-2k-network.html Pages 2K-Nuekom Ensemble Means 2019. https://www.ncdc.noaa.gov/paleo-search/study/26872.   This article appeared on the Watts Up With That? website at https://wattsupwiththat.com/2020/05/03/global-mean-temperature-flattens-the-past/]]>

Join our effort to build a greener future.