Computer models and the randomness of probability

The IPCC draft review for the 2007 climate change review is now online IPCC 2007 update http://www.climatescience.gov/Library/ipcc/wg14ar-review.htm with the various possibilites and probabilities.

Firstly the suggested confidence(how confident they are correct in the modelling of the inputs.
The 2007 draft statement from the IPCC even further identifies failure band in probability(confidence)for the various exogenous variables.

Confidence of understanding.
Co2 methane 80% high
Ozone,atmospheric gases,solar forcing 50% medium
Aerosols,stratosphericwater vapour,high cirrus. 20% low
.Land surface and secondary cloud albedo. 10% very low.

This is the knowledge confidence bands for the input varibles for the various datasets that are used to produce selective outcomes ie predictions,

When constructing the models we have a correlating coefficient for the verification of the prediciton.These are merely random numbers usually with say temperature for the range of the deviation say -1 to +1.By running various montecarlo simulations we get a run time STDV.

Now this is interesting as the IPCC is suggesting time dependent (historical antecedents) whilst using the Markov chain(time discrete probabilites of outcomes)IE If you are at state S at time n, then the probability that you will move on to state x at time n + 1 does not depend on n, and only depends on the current state s that you are in.So whilst the snapshot of forecast may be correct at say 6pm gmt on the 5th may the next runtime simulation may provide a different result dependent on corrected observations.

The markov chain probabilites can provide a wide range of possible outcomes by redistributing the datasets over a wide range of forecasts.The markov montecarlo simulations can even rewrite the bible in real time so here we have selected outcomes much the same as the IPCC.Try it.