Global cooling again; via a comment on this rather dodgy page I find Stockton, C. W. and W. R. Boggess, Geohydrological implications of climate change on water resource development, Contract Report DACW 72-78-C-0031, for U. S. Army Coastal Engineering Res. Center. That isn’t P-R, but contains on p 159 a “Projected Climatic Trends” which in turn references A statistical study of a composite isotopic paleotemperature series from the last 700,000 years, Erik Lundtang Petersen, Søren E. Larsen. Tellus Volume 30, Issue 3, pages 193–200, June 1978; which is of course P-R. And which sayeth:
An attempt is made to identify a stochastic model that makes the best fit to a generalized temperature curve covering the last 700,000 years. A search is made for the model within the family of auto-regressive, integrated, moving average models. All the models presented forecast a decline in temperature during the next 5000 years.
That forecast, now, isn’t of any value to science; only to history of science. Of the model: “only models based on the apparent memory in the series will be discussed” and “By excluding models containing a forcing term, e.g. Milankovitch’s insolation theory (Monin, 1974), we avoid the difliculties ssociated with assigning the form of the forcing, as discussed by Kukla (1975) and Hays et al. (1976), and more-over we avoid linking our results to any specific physical model”. That seems a rather dodgy basis to me, but nevermind, its what they did. I think it tells you why its wrong: (a) it omits the greenhouse gas forcing, so its completely wrong as a forecast, but (b) it omits the differences in Milankovitch type forcing now as opposed to the past interglacials, and so is wrong there too (actually I suspect its even wronger than that, for reasons analogous to those mt has discussed elsewhere re the inevitable tendency of Fourier-type-fitting to “predict” cooling after warming, but never mind).
I think that, even for a bunch of palaeogeographers, they are curiously unaware of the world around them. The conclusion does note that its only valid if the mechanisms controlling the future are still the same as the past; the only way to rescue this is to realise that they are wrapped up in their world, which was not anywhere near as politicised as it is now. Palaeogeographers were free to speculate on what would happen restricting themselves only to their patch.
[Update: thinking about this, I should really have looked at the timescale and size of change more carefully. Their prediction has a timestep of 5 kyr, so arguably we could ignore it if we're interested in the "near future" - say the next few decades or century. But if you do try the next say 50 years, and assume linear interpolation to 5 kyr, then their prediction of 1 oC cooling over 5 kyr translates into 0.01 oC over 50 years. Which is so small it would get overwhelmed by natural variability; its equivalent to "no change / neutral".]