Guest post by Renee Hannon
This post examines regional temperature reconstructions during the past several thousand years relative to different baselines and the responses of end member deviants, the Arctic and Antarctic polar extremes. And it’s a quite interesting tug of war.
Why use Temperature Baselines?
Regional and global reconstructions are frequently shown as temperature anomalies relative to a reference timeframe. According to NOAA, temperature anomaly means a departure from a reference value or long-term average. A positive anomaly indicates observed temperatures were warmer while a negative anomaly indicates observed temperatures were cooler than the reference value. Long-term for instrumental records typically means 30 years.
The World Meteorological Organization (WMO) suggests using the latest decade for the 30-year average and recent graphs and maps show temperature anomalies relative to a 1981–2010 base period. The baseline of 1961-1990 is also frequently used because it serves as a good tie point between older proxy datasets that overlap with recent instrumental data.
The New Oxford American Dictionary states that a baseline is a minimum or starting point used for comparisons. Baselines are commonly used in science and elsewhere. For instance, a shale baseline in log analysis of lithologies is a line drawn through the minimum deflection of impermeable shales. Baseline adjustments typically do not change the shape of the individual time series or affect the trends within it. However, baselines can have a significant visual effect on the spatial distribution of data and regional temperature reconstructions.
Recent Baselines Emphasize the Hockey Stick
The Pages 2K Consortium recently produced new global mean temperature reconstructions, presumably in preparation for the next climate symposium and IPCC report (Pages 2K, 2019). This update was basically a statistical exercise of calculating global mean surface temperatures over the past 2000 years. The results are shown in Figure 1, top graph. The global mean temperatures are shown relative to a 1961-1990 baseline and tie in nicely with the instrumental global temperature anomaly. The study used seven statistical methods ranging from Bayesian models, regression-based techniques to newer linear methods to which they assign catchy acronyms.
These global means highlight the significant cold period of the Little Ice Age (LIA) around 1600 AD which is followed by the rapid “industrial” warming trend. The Medieval Warm Period (MWP) also known as the Medieval Climate Anomaly (MCA) is nowhere to be found.
Instead of using global means, this post focuses on spatial temperature reconstructions over the past thousands of years. Four regional reconstructions consisting of the Arctic by McKay, Northern Hemisphere (NH)/Europe by Luterbacher, Southern Hemisphere (SH) by Neukom, and Antarctic by Stenni are plotted in Figure 1, bottom graph. Links to the datasets and a brief description of the temperature reconstructions are listed at the bottom of this post.
Using the 1961-1990 baseline highlights the LIA which shows large temperature divergences between the four regional reconstructions. This is consistent with the deviations of Pages 2K various global mean temperatures around the time of the LIA. The present warming is pronounced, and all the regional reconstructions appear to merge visually creating the so-called hockey stick effect. It’s very difficult to see the MWP or past warm temperature events when using this baseline.
What seems unusual about this baseline normalization is the position of the Antarctic temperature anomaly. Antarctic temperatures plot about 1 deg C warmer than Northern regions. Positioning of regional temperature reconstructions seem completely reversed with Arctic and NH anomalies plotting colder than Antarctic anomalies. Whether by design or chance, using the 1961-1990 baseline visually amplifies present day warming and suppresses known past natural events like the MWP.
LIA Baseline Reveals Arctic Amplification
The same regional temperature reconstructions in Figure 1 are shown in Figure 2. The only difference is the baseline. A baseline was chosen using the LIA cold period or average minimal temperatures during the past 2000 years. The timeframe of 1600-1700 years AD appears to be a semi-stable period where temperatures are not significantly increasing or decreasing. This is different from the 1961-1990 and 1981-2010 baselines which occur during a non-stable rapidly increasing temperature trend.
Using the LIA baseline shows Arctic/NH temperature anomalies generally warmer than Antarctic/SH as they logically should be. Now regional temperature reconstructions show separation during warm periods like MWP and Present. Spatial heterogeneity shown in Figure 2 is evidence of Arctic amplification. Arctic amplification events occur about 400 years AD, during the MWP, during smaller unnamed bursts between 1400-1600 years AD and present day. The present period appears to be part of a warming trend that began with Arctic and NH temperatures rising over 150 years ago after recovery from the cold LIA.
During the present, Arctic and NH/Europe temperatures begin warming at approximately 1830 AD and before industrial times. SH temperatures begin to increase around 1925 AD and Antarctic temperatures recently began to increase around 1940. That’s over 100 years later than Arctic and NH temperature increases suggesting Arctic amplification and hemispheric temperature asymmetry.
The Arctic temperature anomaly shows an abrupt warming and gradual cooling during the MWP. NH/Europe temperature reconstructions do not show temperatures increasing as strongly during this time. This may be due to the sparseness of tree proxy data available pre-1000 years AD.
The other key difference between the Present and MWP is Antarctic temperature anomalies are colder during the Present. The Antarctic is just now approaching temperature anomalies seen during the MWP and are still lower than pre-1000 AD. Over the past two thousand years, Antarctic temperatures have been generally decreasing by 2 degrees C per 1000 years as described by Stenni. The Antarctic temperature anomaly continued to steadily decrease until about 1940 AD.
Both the MWP and Present have an abrupt warming signal characterized by an initial high rate of temperature increase coupled with a high interhemispheric difference between the Arctic and Antarctic reconstructions. The difference during the MWP is approximately 0.75 deg C and over 1 deg C for the Present. The smaller interhemispheric difference during the MWP is mostly due to higher Antarctic temperature anomalies of 0.35 deg C versus -0.1 deg C for the Present. Another element could be McKay’s Arctic temperature reconstruction where the use of multiple data proxies tends to smooth Arctic temperature peaks.
Both Present and MWP show Similar Latitudinal Trends
The MWP is the most recent analog for the Present warming. There are numerous articles debating the presence and absolute warmth of the MWP (Ljungqvist, Nuekom, Wilson). Most of the discussion is based on NH temperature reconstructions and lack of temperature response in southern latitudes.
Figure 3 is a histogram of the temperature differences between the MWP (950-1100 AD) and Present (1950-2000 AD) relative to the LIA (1600-1700 AD) for thirteen individual reconstructions. Ljungqvist, 2019 conducted a detailed study using similar temperature reconstructions of MWP using a longer timeframe for the LIA (1450-1850 AD). Surprisingly, both the MWP and Present show similar latitudinal temperature differences, larger in the northern hemisphere shown in greens and blues and smaller in the southern hemisphere shown in orange and reds.
Absolute temperatures tend to be slightly lower during the MWP than Present relative to the LIA. McKay’s Arctic temperature reconstruction shows that Present is about 0.2 deg C warmer than during the MWP. Pages 2K 2013 Arctic reconstruction shows the biggest discrepancy with Arctic temperatures of almost 0.6 deg C warmer during Present than MWP. Their Arctic reconstruction has recently been updated. Tree ring proxy data is sparse during the MWP leading to greater uncertainty for NH and Europe temperature reconstructions. Otherwise, most of the temperature differences between Present and MWP are within 0.2 to 0.3 deg C and almost within the error of the data.
Instrumental data using GISS Surface Temperature Analysis reveals diverging NH and SH trends around 1940 and again after 1980 as shown in Figure 4. The SH temperature anomaly in red is very similar to ocean surface temperatures in blue. Both are flat to decreasing until around 1930 and then begin to increase while NH temperatures are increasing.
The NH and SH interhemispheric difference shows latitude divergence at 1940, convergence by 1980 and increasing divergence since. Arctic and Antarctic polar temperature anomalies show a large difference of almost 1 deg C at 1940.
Climate Models Do Not Reproduce Antarctic and SH Temperature Reconstructions
The global spatial pattern of the Present suggests that internal variability plays a major role in driving heterogeneous warming and is underestimated in model simulations. Both Luterbacher and Ljungqvist state that NH and European temperature reconstructions have larger temperature differences between the Medieval period and the Little Ice Age than in climate simulations. Luterbacher states this discrepancy may be due to inflated variability of the reconstructions and/or an underestimation of climate model sensitivity to internal variability on centennial and longer time scales. Ljungvist suggests the disagreement is related to too-cold initial ocean conditions and too-weak internal variations.
Climate simulation results are also poor in matching Southern Hemisphere temperature reconstructions (Nuekom, 2018). In the SH, climate models show a much greater response to volcanic, aerosol and greenhouse gas forcings than seen in the temperature reconstructions. Also, climate models do not recognize the warming delay of the SH and Antarctic and therefore overestimate southern hemisphere warming. This suggests climate models have less interhemispheric variance and greater consistency between hemispheres than in actual temperature reconstructions. Nuekom speculates that temperature responses in the ocean-dominated SH are delayed and buffered by the large heat capacity of the oceans and by more monotonic oceanic processes. Pages 2K, 2015, agrees that better knowledge of internal and forced variability in the ocean is necessary to understand the influence of climate modes on temperature variability.
Baselines are important in the visual presentation of global climate heterogeneity. The 1961-1990 baseline commonly used emphasizes the LIA and the present rapid warming. Past warm anomalies are obscured. Conversely, converging spatial temperature anomalies during the cold LIA demonstrates separation of northern latitudes from southern latitude temperatures during warm events such as the MWP and Present.
The MWP analog is frequently questioned as to whether it is a global event and not having a global impact, especially in the southern latitudes. Ironically, the Present shows similar smaller temperature increases and delayed warming in southern latitudes. The Present warming began as an abrupt warm upturn from the cold LIA around 1830 initially in the Northern Hemisphere and Arctic.
The Arctic and NH are on different climate pathways than the SH and Antarctic. Northern latitudes respond rapidly to short-term influences such as volcanic activity, GHG and solar whereas ocean-dominated southern latitudes lack many of these signals. Until science and climate models can sort out natural underlying polar causes and effects, only then can science unravel the potential influence of anthropogenic intervention on climate change.
Acknowledgements: Special thanks to Donald Ince and Andy May for reviewing and editing this article. This article is dedicated to my Delaware friends.
Cowtan, K. & Way, R. G. Coverage in the HadCRUT4 temperature series and its impact on recent temperature trends. Q. J. R. Meteorol. Soc. 140, 1935–1944 (2014). https://rmets.onlinelibrary.wiley.com/doi/full/10.1002/qj.2297Documents
Ljungqvist, F.C.; Zhang, Q.; Brattström, G.; Krusic, P.J.; Seim, A.; Li, Q.; Zhang, Q.; Moberg, A. Centennial-Scale Temperature Change in Last Millennium Simulations and Proxy-Based Reconstructions. J. Clim. 2019, 32, 2441–2482. https://journals.ametsoc.org/doi/pdf/10.1175/JCLI-D-18-0525.1
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
Neukom, R., Schurer, A. P., Steiger, N. J. & Hegerl, G. C. Possible causes of data model discrepancy in the temperature history of the last Millennium. Sci. Rep. 8, 7572 (2018). https://www.nature.com/articles/s41598-018-25862-2
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-PMIP3 group: Continental-scale temperature variability in PMIP3 simulations and PAGES 2k regional temperature reconstructions over the past millennium, Clim. Past, 11, 1673– 1699, https://doi.org/10.5194/cp-11-1673-2015, 2015.
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.
Wilson, R. et al. Last millennium northern hemisphere summer temperatures from tree rings: Part I: The long-term context. Quat. Sci. Rev. 134, 1–18 (2016). https://www.st-andrews.ac.uk/~rjsw/N-TREND/Wilsonetal2016.pdf
Temperature Reconstruction Datasets
Surface temperature reconstructions over the past thousands of years are numerous (Luterbacher, McKay, Nuekom, Stenni, and Wilson) to name a few. Most of these authors, if not all, are part of the Pages 2K Consortium. Pages 2K compiled a temperature proxy database that includes almost 700 records from ice cores, sediment, corals, tree rings, and pollen in 2013 and updated it in 2017.
Regional reconstructions used in this post are the Arctic by McKay, Northern Hemisphere (NH)/Europe by Luterbacher, Southern Hemisphere (SH) by Neukom, and Antarctic by Stenni. The Arctic reconstruction by McKay utilizes proxy records consisting of ice cores, tree rings, lake and marine sediments north of 60 deg N. The Europe reconstruction used is tree based and like Wilson’s NH reconstruction. Wilson’s NH data ends at 900 AD and includes a couple of peak temperature differences during the MWP (not used). Neukom’s SH reconstruction is based on ice cores, tree rings, and corals and ends at 1000 AD. Stenni used mostly ice core isotope data to produced Antarctic temperature reconstructions.
There is a notable shift is data quantity and quality of temperature datasets around 1000 AD especially tree rings. The number of tree ring records are reduced significantly from 400 records post-1600 AD to less than 30 records pre-1000 AD. Unfortunately, this reduction of proxy data occurs in the middle of the past analog warm period, the MWP. For an objective review of Pages 2K datasets, I recommend reading Steve McIntyre’s articles particularly in reference to tree ring data. He discusses the accuracy of tree ring data, the divergence problem and cherry picking of data.
Arctic McKay, 2016. 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
NH Wilson, 2016. https://www.ncdc.noaa.gov/paleo-search/study/19743
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.
Instrumental Data. https://data.giss.nasa.gov/gistemp.
Credit: Source link