In my posts, I have urged that the focus of climate modeling research change from focusing on providing multi-decadal climate predictions to the assessment of predictability; e.g. see
I was alerted by Jos de Laat of KNMI to an important new research paper that specifically addresses this issue. This paper is
Oldenborgh, G.J. van, F.J. Doblas-Reyes, B. Wouters and W. Hazeleger, 2012: Skill in the trend and internal variability in a multi-model decadal prediction ensemble. accepted, Clim. Dyn.
The abstract [as it reads here] is [highlight added]
Decadal climate predictions have skill due to predictable components in boundary conditions (mainly greenhouse gases) and initial conditions (mainly the ocean). We investigated the skill of temperature and precipitation hindcasts from a set of four coupled ocean-atmosphere models. Regional variations in skill with and without trend due to global warming point to separate effects of the boundary forcing and the ocean initial state. In temperature most skill comes from the prescribed boundary forcing. The trend of the global mean temperature is represented well in the hindcasts, but variations around the trend show little skill. The models have non-trivial skill in hindcasts of North Atlantic SST beyond the trend. The same may hold for the decadal ENSO region, although the signal is less clear. Hence we conclude that the ocean initial state contributes significantly to skill in forecasting SST in these regions.
The conclusion contains the text
A 4-model 12-member ensemble of 10-yr hindcasts has been analysed for skill in SST, 2m temperature and precipitation. The main source of skill in temperature is the trend, which is primarily forced by greenhouse gases and aerosols. This trend contributes almost everywhere to the skill. Variation in the global mean temperature around the trend do not have any skill beyond the first year. However, regionally there appears to be skill beyond the trend in the two areas of well-known low-frequency variability: SST in parts of the North Atlantic and Pacific Oceans is predicted better than persistence. A comparison with the CMIP3 ensemble shows that the skill in the northern North Atlantic and eastern Pacific is most likely due to the initialisation, whereas the skill in the subtropical North Atlantic and western North Pacific are probably due to the forcing.
In the Atlantic, the ensemble shows clear skill in predicting an AMO index that is orthogonal to the trend in yrs 2–5, and reasonable skill in yrs 6–9. The skill in decadal ENSO is lower, not statistically significant, but in agreement with other studies. The CMIP3 ensemble shows less skill in both these indices. There is also an indication of skill in hindcasting decadal Sahel rainfall variations, which are known to be teleconnected to North Atlantic and Pacific SST. The uninitialised CMIP3 ensemble that includes volcanic aerosols reproduces these variations as well, but the models without volcanic aerosols do not. It therefore remains an open question whether initialisation improves predictions of Sahel rainfall.
The modelled trends agree well with observations in the global mean, but the agreement is not so good at the local scale.
These experiments are only a first step towards decadal forecasting using non-optimised methods from seasonal forecasting. The skill assessment does not take into account the considerable biases and drift of the models. It is based on only nine or ten data points and hence suffers from large statistical uncertainties. Larger ensembles sizes per model and more frequent and earlier starting dates will be required to characterise the skill of decadal forecasts better. The verification of decadal hindcasts can then be used to improve the climate models, their forcings and initialisation procedures to give more reliable and skilful climate forecasts.
The authors should be commended for focusing on this assessment of predictability. We need more such excellent studies!