Some news that may help European meteorologists with their seasonal forecasts in the future. We know that last fall many predicted a hard winter for Europe, and just the opposite is now occurring. What I find interesting is how the new paper reminds us of the proper way to build a model.
A team of climate scientists led by Dr. Daniela Matei and Prof. Dr. Jochem Marotzke of the Max Planck Institute for Meteorology (MPI-M) and Prof. Dr. Johanna Baehr from Hamburg University’s Cluster of Excellence “CliSAP” have shown for the first time that the strength of the Atlantic Meridional Overturning Circulation (AMOC) at 26.5 °N can be skillfully predicted for up to four years.
This can be important because, according to their press release:
The AMOC – known as “Gulf Stream” – transports warm surface waters into the high latitudes, where they cool, sink and return southwards at depth as cold North Atlantic Deep Water. Variations in AMOC can significantly affect the northward ocean heat transport and therefore the European and North Atlantic climate. Through its influence on sea surface temperature (SST), AMOC can further impact climate phenomena such as Sahel droughts or the frequency of Atlantic hurricanes. Therefore, it is of outmost importance to be able to predict these climate variations on a time horizon from years to decades.”
“Climate variations” here is the key. The authors explain the proper approach for making mid-term climate predictions in general:
In the near term (inter-annual to decadal timescales), climate variations are influenced by both anthropogenic forcing and natural variability. […] The skill of any prediction system is assessed retrospectively, by performing the so-called “hindcast” or “retrospective forecast” and comparing them with observations.”
Oh really! This is almost what the IPCC does with its famous climate models, but with a small difference of course. In the IPCC models, there are always huge discrepancies between the hindcast and observations. No problem though. The IPCC modellers simply apply the ever so versatile aerosol fudge factor wherever needed, and presto! CO2 is always the big driver and the significant natural cycles are reduced to nothing.
I don’t know if Drs. Matei and Marotzke of the MPI-M produce models for the IPCC, but if so, all they have to do now is take the same approach they mention above and include the other (very well-known, at least to the rest of the world) natural cycles like the PDO, AMO, NAO, solar, etc. in their climate models. If they did that, they would quickly discover that they could do away with the massive aerosol fudging and actually produce a multi-decadal model that works. Of course, the IPCC may not be happy with the result.
If they don’t make climate models for the IPCC, then they should at least give their climate modelling colleagues a hint or two.
Also read here (in German).
“The skill of any prediction system is assessed retrospectively, by performing the so-called “hindcast” or “retrospective forecast” and comparing them with observations.”
There is one caveat though. When you have used ALL YOUR OBSERVATION DATA to train the model, it will necessarily reproduce them. To really validate the model, it is essential to use only a subset of the data to train it and run it on the remaining data for the validation.
I don’t trust Marotzke, he’s a sea level rise alarmist and a plain vanilla warmist.
I’ll wait 4 years to see whether he botched it. I’d bet on it.
Please more respect for someone who can model the earth into an ice-covered state without getting red:
“Recently, Marotzke and Botzet (2007) have shown that in a comprehensive climate model both the current climate and a completely ice-covered Earth are stable states under today’s total solar irradiance (TSI) and greenhouse gas concentrations. By setting TSI to near-zero they were able to cause a transition from the present-day climate to the ice-covered state within 15 years [Marotzke, J. and M. Botzet (2007)”
from . Abstract [Voigt, A.; Marotzke, J.; The Transition From the Present-Day Climate to a Snowball Earth Simulated With a Comprehensive Climate Model; at: http://adsabs.harvard.edu/abs/2007AGUFMPP23B1337V%5D
Comment in German (2009) here: http://www.2007seatraining.de/Archiv/maerz_09.html