A Holocene Temperature Reconstruction Part 2: More reconstructions

By Andy May

In the last post (see here) we introduced a new Holocene temperature reconstruction for Antarctica using some of the Marcott, et al. (2013) proxies. In this post, we will present two more reconstructions, one for the Southern Hemisphere mid-latitudes (60°S to 30°S) and another for the tropics (30°S to 30°N). The next post will present the Northern Hemisphere mid-latitudes (30°N to 60°N) and the Arctic (60°N to the North Pole). As we did for the Antarctic, we will examine each proxy and reject any that have an average time step greater than 130 years or if it does not cover at least part of the Little Ice Age (LIA) and the Holocene Climatic Optimum (HCO). We are looking for coverage from 9000 BP to 500 BP or very close to these values. Only simple statistical techniques that are easy to explain will be used.

Southern Hemisphere mid-latitudes

Our reconstruction for this region is shown in figure 1. The R code and the input and output datasets for the Southern Hemisphere mid-latitudes can be downloaded here.

Figure 1

This reconstruction has a more defined HCO than we saw in the Antarctic and it is placed between 8000 BP and 5000 BP. The HCO occurs at different times in different places as discussed by Renssen, et al. (2012). Following this, the temperature generally drops to a low in the LIA. In this reconstruction, we see two LIA lows, one at 1690 AD and one at 1550 AD. The Medieval Warm Period (peak 1030 AD) and the Roman Warm Period (peak 90 BC) are very distinct in this reconstruction. The Minoan Warm Period (peak 1890 BC) can also be seen.

There are only four proxies in this reconstruction, three in New Zealand and one off the coast of Chile. The locations are shown in figure 2.

Figure 2

Two of the proxies have been combined into one record, the three proxies used are plotted in figure 3.

Figure 3

Three proxies were rejected due to large sample intervals, TN057-17 (Nielsen, et al., 2004) was rejected because it was very anomalous. See the plot in figure 4.

Figure 4

TN057-17 is a sea surface temperature proxy located in the Southern Ocean, right on the Antarctic Polar Front (APF), see Figure 5. The APF has a very abrupt sea surface temperature change. It is the southern limit of synchrony with the Northern Hemisphere climate system. This location, currently, has sea ice cover about two weeks per year (Nielsen, et al., 2004) but the time of ice cover has changed a lot during the Holocene and this has probably had a dramatic effect on the proxy. The maximum sea ice cover was 4300 BP, which is also the time of the lowest TN057-17 temperature.

Figure 5, source (Nielsen, et al., 2004)

Sea ice presence (SIP) at the TN057-17 location is shown in figure 6.

Figure 6, Sea ice presence (SIP) at TN057-17 (Source: Nielsen, et al., 2004)

There is a risk that the TN057-17 proxy has been affected by local conditions that are only vaguely connected to climate change and for this reason the proxy was rejected.

The Chilean GeoB 3313-1 proxy (Lamy et al., 2002) only went back to 7000 BP and for this reason would be rejected on its own. But, the New Zealand proxy MD97-2121 (Pahnke and Sachs, et al., 2006) has nearly the same latitude and is continuous from 12464 BP to 3316 BP. So, these two proxies were merged by adjusting them to the mean of the overlapping interval 6900 BP to 4000 BP. See figure 7.

Figure 7

The logic in combining these two proxies is it gives us one more proxy in a region that has few available and, at least part of this composite is outside the New Zealand area. The most recent portion of MD97-2121 (4000 BP to 3300 BP) was not used in the composite as it looked suspicious. Pahnke and Sacks (2006) report that the MD97-2121 core top (the most recent sediments in the core) may be problematic due to lack of recent sediments at the cored location. In any case, a 3000-year-old core top, presumably close to the sea floor, has probably been churned quite a bit and should be greeted with suspicion. Older than 4000 BP, the results are consistent with GeoB-3313-1.


The reconstruction for the tropics (30°S to 30°N) is shown in figure 8. The ODP-658C proxy is problematic, so we present a reconstruction including it in figure 8A and one without it in figure 8B. The R code and the input and output datasets for the tropics can be downloaded here.

Figure 8A, with the ODP-658C proxy

Figure 8B, without the ODP-658C proxy

There is a very distinct LIA at 1630 AD. The two peaks around the classical MWP are 1090AD and 930 AD. The Roman Warm Period (90 BC) and the Minoan Warm Period are apparent. In this reconstruction, the HCO is from 9600 BP to 7700 BP.

There are eight usable proxies in the tropics using our criteria, they are plotted in figure 9.

Figure 9

The location of the proxies is shown in figure 10, there are quite a few in Indonesia so we did not put arrows on the figure for each one. One of the Indonesian proxies, Ros_BJ813GGC, is not a Marcott et al. (2013) proxy. It is from Rosenthal, et al. 2013.

Figure 10

The rejected proxies, except for ODP-658C, were all because of resolution or because they did not cover the period from the Holocene Climatic Optimum to the Little Ice Age. ODP-658C (deMenocal et al., 2000) is the brown line in figure 9. It is located off West Africa, the northern most arrow off West Africa in figure 10. The proxy is plotted in figure 11. This proxy was left out of the final reconstruction, but figure 8A shows a reconstruction that includes it.

Figure 11

The sharp break in the proxy at 5700 BP appears to be a data problem until we consider that this is the end of the African humid period when the Sahara turned into a desert due to the abrupt movement of the Intertropical Convergence Zone or “ITCZ” (deMenocal et al. 2000 and Javier, 2017). Figure 12 shows the worldwide change that took place around 5700 BP, it is from Javier’s essay here. This change in the location of the climatic equator (ITCZ) is often called the Mid-Holocene Transition when the world goes from the Holocene Climatic Optimum (HCO) period to the Neoglacial period. This data suggests that the Mid-Holocene Transition, in this area, occurred in less than 120 years between 5808 BP and 5683 BP using the dates given by deMenocal, et al. This is how long it took for sea surface temperatures at the ODP-658C location to increase over 2°C. The core location is shown in figure 12 with a red star. Around 5700 BP the ITCZ migrated from north of this location to south of it.

Figure 12 (Source: Javier, here)

The proxy temperature record labeled 17940 (Pelejero, et al., 1999) from the South China Sea could also be considered slightly anomalous since in the Neoglacial period it trends warmer, rather than cooler. It also shows no Holocene Climatic Optimum. The proxy is displayed below in figure 13.

Figure 13

The South China Sea is a very large marginal basin in the western Pacific. It is bounded by broad shallow shelves in the northwest and southwest that emerge during periods of low sea level. Sea level was low enough during the early Holocene for these shelves to be emergent. The position of the shelves can be seen by following the 100-meter isobath (water depth) contour in the map in figure 14. In addition, core 17940 is only 400 km from the mouth of the Pearl River, the second largest river in China. Due to the large sediment discharge from the river the location has a very high sedimentation rate, further it was larger in the past as the glaciers retreated to their present position and sea level was lower. We kept the proxy in the reconstruction, but recognize that it is sensitive to the changes in sea level experienced during the Holocene and to changes in the discharge rate from the Pearl River. The sea surface temperatures of the other cores in the South China Sea are similar.

Figure 14 (Source: Pelejero, et al. 1999)

In the next post, we will discuss the Northern Hemisphere Holocene proxies and the Arctic proxies. The supplementary materials for the tropics reconstruction can be downloaded here.


The Southern Hemisphere mid-latitude reconstruction is built from very little data. The data is mostly in the New Zealand area and cannot be considered representative of the whole region. But, we work with what we have.

The tropics reconstruction is built from widely dispersed proxies that sample all major ocean basins. It is probably representative of the region. The reconstruction without the anomalous ODP-658C proxy is probably the best to use. The Mid-Holocene Transition is a real event, but probably had no global temperature effect. It is merely a shift in the climatic equator or the Intertropical Convergence Zone or “ITCZ.” The ODP-658C location warmed dramatically 5700 BP, but presumably another location, that is not sampled, cooled as dramatically. Including the warming at the ODP-658C site, without the cooling elsewhere distorts the regional and global picture. Therefore, we excluded the proxy.

I am very grateful to Javier who has read this post and made many very helpful suggestions. Any errors are the author’s alone.