Paper Reviewed
Harrison, S.P., Bartlein, P.J., Brewer, S., Prentice, I.C., Boyd, M., Hessler, I., Holmgren, K., Izumi, K. and Willis, K. 2014. Climate model benchmarking with glacial and mid-Holocene climates. Climate Dynamics 43: 671-688.

In their meticulous evaluation of PMIP2 and CMIP5 climate models, Harrison et al. (2014) use paleo-climatic reconstructions of the Last Glacial Maximum (LGM, ca 21,000 years ago) and the mid-Holocene (MH, ca 6,000 years ago) to evaluate the abilities of today's state-of-the-art climate models to adequately replicate these two significant global climates of Earth's distant past, which if they could do so successfully would suggest that they could probably equally well describe Earth's climate in a CO2-enriched future. So how did they do in this endeavor? ... and what did they thereby learn?

With respect to the glacial oceans, the nine researchers say that (1) "most models overestimate the ocean cooling," that (2) "they do not capture the heterogeneity seen in the reconstructions," that (3) "spatial patterns in reconstructed LGM annual SST [sea surface temperature] anomalies ... are not well predicted by the models," and that (4) "the seasonal SST anomalies show no correlation with the reconstructions."

As for the glacial continents, Harrison et al. say that (5) "all but two models underestimate the reconstructed annual cooling, with the largest median bias nearly 3.5°C and eight models having a bias larger than 1°C," that (6) "all the models underestimate the mean temperature of the coldest month reduction," with the smallest median bias being +2.4°C and the largest +7.3°C," that (7) "all models underestimate the LGM reduction in mean annual precipitation over land," and that (8) "most models also underestimate the increase in aridity."

Moving on to the mid-Holocene ocean, the nine climate scientists indicate that (9) "the models underestimate the reconstructed warming in northern mid-latitudes," and they say that they (10) "consistently underestimate the heterogeneity in sea surface temperatures." At the same time, they report that on the mid-Holocene continents, (11) "the models overestimate summer warming by 0.56-2.27°C," that (12) "the models consistently underestimate the reconstructed change in mean annual precipitation," while they also report that (13) "the models overestimate the ratio of actual to equilibrium evapotranspiration," and that (14) "simulated MH land climates show consistently less spatial variability than the reconstructed climates."

More generally, Harrison et al. report that (15, 16) "ocean temperatures are globally low and land temperatures globally high, compared with the reconstructions," that (17) "the present generation of models tends to overestimate SST cooling in the tropics," that (18) "the models underestimate precipitation changes in the regions with the largest summer warming," reflecting the fact that (19) there are "problems in the simulation of land-atmosphere heat fluxes." And, nearing the end of their list of problems, they say that (20) "all models overestimate the reconstructed summer cooling of the tropics at the LGM," just as (21) "all models underestimate the MH increase of Afro-Asian monsoon precipitation."

At the end of the day, therefore, Harrison et al. declare that (22) "there are still shortcomings in the amplitude of simulated changes," and that (23) there is still a "need for continued efforts by modelling groups to achieve accurate simulations of fundamental climate processes."