In September 2011, we presented a medium-term (79 months) quantitative prediction of monthly global mean temperatures based on an interdependent system model of the atmosphere developed by KnowledgeMiner, which was also discussed at Climate Etc. in October 2011. This model describes a non-linear dynamic system of the atmosphere consisting of 5 major climate drivers: Ozone concentration, aerosols, radiative cloud fraction, and global mean temperature as endogenous variables and sun activity (sunspot numbers) as exogenous variable of the system. This system model was obtained from monthly observation data of the past 33 years (6 variables in total: the 5 variables the system is actually composed of (see above) plus CO2, which, however, has not been identified as relevant system variable), exclusively, by unique self-organizing knowledge extraction technologies.

Now, more than a year has passed, and we can verify what has been predicted relative to the temperatures, which have really been measured (fig. 1).

Fig. 1: Ex-ante forecast (most likely (red), high, low (pink); April 2011 - November 2017) of the system model as of March 2011 vs observed values (black and white square dots; HADCRUT3) from April 2011 to December 2012. These 21 months are used for verification of the out-of-sample predictive power of the system model.

Verifying the prediction skill of the system model from April 2011 to December 2012, the accuracy of the most likely forecast (solid red line) remains at a high level of 75%, and the accuracy relative to prediction uncertainty (pink area) is an exceptional 98%. Given the noise in the data (presumably incomplete set of system variables considered, noise added during measurement and preprocessing of raw observation data, or random events, for example), this clearly confirms the validity of the system model and its forecast.

How does the IPCC AR4 A1B Scenario compares to the recent observed data and the system model forecast (fig. 2)?

Fig. 2: Ex-ante most likely forecast (red; April 2011 - November 2017) of the system model as of March 2011 vs observed values (black and white square dots; HADCRUT3) from April 2011 to December 2012 vs IPCC A1B projection (yellow; until November 2017) vs CO2 concentration (light gray; until November 2017).

The IPCC A1B scenario is derived from a number of million-dollar General Circulation Models (GCMs), which depend on atmospheric CO2 as the major driver for Global Warming. Consequently, the IPCC A1B projection follows the development of CO2 concentration, which - in contrast to observed global temperatures - has only been rising in the past and which will continue to do so for the next future. This IPCC projection currently shows a prediction accuracy of 23% (September 2007 - December 2012, 64 months) and just 7% accuracy for the same forecast horizon as applied for the system model (April 2011 - December 2012, 21 months).

In Fig. 2, two different models - IPCC model and atmospheric system model - with two very different modeling approaches - theory-driven vs data-driven modeling - are compared. The IPCC model is based essentially on AGW theory by emission of greenhouse gases, namely CO2, the presented atmospheric system model on the other hand is a CO2-free prediction model. It is described by 5 other variables. The IPCC model shows a prediction accuracy of 7% and the atmospheric system model an accuracy of 75% for the same most recent 21 months of time...

It is only fair to mention that the objective of the IPCC models is long-term (centennial) qualitative projection of global temperatures while the proposed system model is for medium-term (decadal) quantitative forecasting purposes. It is a property of every long-term prediction model, by definition, that it does not necessarily fit to short-term variations of the observed variable. Therefore, it is not surprising that the IPCC model shows lower prediction accuracy on shorter time horizons (here, 21 and 64 months) than a dedicated short- to medium-term prediction model does.

However, for more than 6 years now observed global temperatures have been constantly below the IPCC projection. And the gap between observed global temperatures and projected IPCC scenario is expected to grow every month that passes given the confirmed system model forecast (fig. 2). By the end of 2017, within 10 years then, the prediction error of the IPCC A1B projection might have been accumulated to around 0.4 °C or 100%, already.

This one-sided drift of the IPCC projection seen in Fig. 2 is not common for long-term prediction models. A similar drift situation is not observed for the time before 2007, for the data the GCMs were developed on. Here, over- and underestimation of observed values is balanced, as expected. In modeling, such a drift is seen as clear evidence of low (or decreasing) descriptive power of the model, the lack of skill to describe the underlying phenomenon sufficiently. This evidence, for the IPCC model, is not surprising since the simplistic linear cause-effect relationship „growing atmospheric CO2 concentration leads to proportionally growing global temperatures“, which the model is based on, does not adequately describe the complex and interdependent nature of the atmosphere-ocean system.

Additionally, our climate system is essentially influenced by external, cosmic climate drivers such as the Earth Orbit Oscillation in centennial time frames, the multidecadal tri-synodic Jupiter/Saturn cycle, or the well-known orbit eccentricity Milankovitch cycle, which causes glacial and interglacial ages on Earth. These cosmic climate drivers are responsible for most of the variation of solar radiation received on Earth, resulting in medium- to long-term warming and cooling trends, independently from the sun‘s own rather small changing activity and radiation.

System model data and example models are available here for download.