[Stoch] Paper on representations of model uncertainty

[Antje Weisheimer]

Dear all,

I would like to draw your attention to a new paper on model uncertainty 
in monthly and seasonal forecasts.

Antje

----------------------------------------------------------------------

Weisheimer, A., T. N. Palmer, and F. J. Doblas-Reyes (2011), Assessment 
of representations of model uncertainty in monthly and seasonal forecast 
ensembles, Geophys. Res. Lett., 38, L16703, doi:10.1029/2011GL048123.

<http://www.agu.org/pubs/crossref/2011/2011GL048123.shtml>

The probabilistic skill of ensemble forecasts for the first month and 
the first season of the forecasts is assessed, where model uncertainty 
is represented by the a) multi-model, b) perturbed parameters, and c) 
stochastic parameterisation ensembles. The main foci of the assessment 
are the Brier Skill Score for near-surface temperature and precipitation 
over land areas and the spread-skill relationship of sea surface 
temperature in the tropical equatorial Pacific. On the monthly 
timescale, the ensemble forecast system with stochastic parameterisation 
provides overall the most skilful probabilistic forecasts. On the 
seasonal timescale the results depend on the variable under study: for 
near surface temperature the multi-model ensemble is most skilful for 
most land regions and for global land areas. For precipitation, the 
ensemble with stochastic parameterisation most often produces the 
highest scores on global and regional scales. Our results indicate that 
stochastic parameterisations should now be developed for multi-decadal 
climate predictions using earth-system models.