Was China 6-8 K warmer in winter 6000 years ago?

Monckton, Soon, Legates & Briggs have published a paper in the Science Bulletin (formerly Chinese Science Bulletin), which, despite having an impact factor of 1.365, is “one of the world’s top six learned journals of science”. Allegedly.

In their paper, “Why models run hot: results from an irreducibly simple climate model” Monckton et al argue that the climate will only warm slightly over the next century because a process engineer would not design a climate with large net positive feedbacks that give a high climate sensitivity.

A couple of parts of the paper concern palaeoclimate data. I want to look at one of these.

Studies of paleo-vegetation and pollens [sic] in China during the mid-Holocene climate optimum 6,000 years ago find January (i.e., winter minimum) temperatures to have been 6–8 K warmer than present. Yet, Jiang et al. [31] showed that all 36 models in the Paleoclimate Modeling Intercomparison Project backcast winter temperatures for the mid-Holocene cooler than the present. Also, all but one model incorrectly simulated annual-mean mid-Holocene temperatures in China as cooler than the present [31]. Suggestions that current models accurately simulate the mid-Holocene climate optimum rely on comparisons between projected and observed summer warming only, overlooking models’ failure to represent winter temperatures correctly, perhaps through undue sensitivity to CO2-driven warming.

So was China 6-8 K warmer in winter in the mid-Holocene, and what would it mean if it was?

The Earth’s orbit was slightly different in the mid-Holocene, leading to, at China’s latitude, more insolation in summer, but less through the rest of the year than at present. The annual mean insolation was slightly reduced. Concentrations of the greenhouse gases CO2 and CH4 were also reduced relative to the pre-industrial. With winter radiative forcing being negative, warmer than pre-industrial winter climate is unexpected.

It is possible that the extra isolation in summer warms the oceans and this heat is available in winter, but to get 6-8 K of warming would require that either climate sensitivity to the summer radiative forcing is very high, or that the changes in radiative forcing caused large changes in climate dynamics. Both are exactly the sort of climate surprise we need to be aware of. Thank you Monckton et al for raising awareness of this and disproving their own argument.

Time to look at Jiang et al. (2012).

Jiang et al is a data-model comparison paper. All of the models show cooler temperatures in winter and warmer summer temperatures in the mid-Holocene, and all but one show a cooler annual mean temperature. The models are PMIP1 and PMIP2 models as the higher resolution PMIP3 models were not available at the time. If the analysis was repeated with the newer models, there would be some differences to the results, but I doubt the story would fundamentally change.

The data Jiang et al use are a compilation of January, July and mean annual temperatures reconstructed from palaeoclimate archives. Pollen data are interpreted as indicating warmer winter temperatures in the mid-Holocene. Winter temperature is certainly important for some species, for example frost intolerant species, but my work on Chinese pollen makes me suspect that it is difficult to make good winter temperature reconstructions as precipitation is such an important driver of vegetation in China.

There are only a few quantitative reconstructions of winter temperature in Jiang et al., and several of these are show considerable warmth.

1. Shi et al 1993 report that the fern Ceratopteris grew near Beijing in the mid-Holocene and that this indicates a rise of 6 K in winter temperature. The paper cited for the reconstruction is Kong Zhaochen et al. (1991) but this has been omitted from the reference list. The Kong Zhaochen et al. (1991) I have found does not concern Ceratopteris, but has the same winter and mean annual temperature anomalies. I think this citation is incorrect, and they should instead have cited Xu et al 1988 (also cited by Jiang et al), who report Ceratopteris from the same site. Xu et al report that they think the presence of this aquatic fern indicates that temperatures were 2-4K higher, but do not specify a winter temperature change. Since the fern can live as an annual, overwintering as spores, it probably isn’t very sensitive to winter temperature.

2. Kong et al (1991) report that Helicia nuts have been found in mid-Holocene deposits near Nanjing, which together with other plant fossils found indicates that a subtropical evergreen broadleaved forest grew at that site. From this, Kong et al infer that winter temperatures were 6.3 K higher using discriminant analysis. Discriminant analysis is a useful tool in quantitative palaeoecology and can be used, for example, to assign fossil pollen assemblages to vegetation types and hence qualitative estimates of climate change. It is not obvious how it can be used to make a quantitative reconstruction. From a very bad Google translation of the paper, Kong et al seem to have used a discriminant analysis to find the climate variables that distinguish sub-tropical forest from evergreen broadleaved forest and used the difference between the centroids of the two forest types as the estimate of climate change. I do not think this is an appropriate analysis – we don’t know if the centroid of the subtropical forest moved as far north as Nanjing, only that the northernmost limit of Helicia moved this far north. According to GBIF, Helicia has been recorded growing at 29°N, only 3° south of Nanjing and about 3 K warmer in winter. This is a better estimate than that given by Kong et al 1990, but some species distribution modelling would be needed to get a reliable estimate.

3. Kong et al (1990) infer from the presence of stable forest in the Qinghai Lake area during the mid-Holocene that Januaries were 8 K warmer. I cannot find a copy of the paper, so I don’t know how they inferred this large winter warming from their pollen data (the basic pollen diagram is included in Du et al (1989)). Other pollen records from Qinghai, for example Shen et al (2005) show that in the mid-Holocene Pinus, Picea and Abies forest surrounded the lake and was later replaced by forest-steppe producing lots of Artemisia pollen. Shen et al link this decline in forest to the cooling and drying of the late-Holocene climate. None of the taxa in the pollen record are obviously particularly sensitive to winter temperature, so I don’t think the estimate from Kong et al (1990) is reliable.

4. Jiang et al cite four other papers that claim that warming was greater in winter than summer in the mid-Holocene using pollen evidence. Two of these papers reconstruct mid-Holocene winter temperatures 2-3 K above pre-industrial, and one does not reconstruct winter temperatures! This leaves just one paper (Guiot et al 2008) reconstructing a reasonably large mid-Holocene winter warming, albeit using a method with an uncertainty (RMSEP) of 6.2 K (Jiang et al 2006).

I remain agnostic as to whether winter temperatures in China were warmer in the mid-Holocene despite the reduction in radiative forcing. It is possible that the proxies used are more sensitive to summer temperatures and the insolation-driven summer warming generates an apparent winter warming, as has potentially occurred in other studies. I have a funding for a small project to investigate this problem this year.

None of the evidence for 6-8 K warmer winters in China in the mid-Holocene is robust. Monckton et al do not provide evidence of a huge model failure, but a failure in their own scepticism. They accepted without checking a result they thought favourable to their argument.