Ensembles, Forecasts and Predictability
143
the ratio between the inter ensemble variance and the total variance, averaged over
the duration of the simulation. Values larger than 1.0 indicate that the inter ensembIe separation (Spread) is larger than the total variance that includes the normal climate variability, whereas values smaller than 1.0 indicate that the realizations are
closer to each above the usual clutter of climate. It is possible to see how the reproducibility is space-dependent (and also time dependent, as Winter predictability is
different from Summer) with major reproducible areas in the subtropics. Interestingly, small areas of high reproducibility appear in selected portion of the middle
latitudes. These regions are those influenced by teleconnection pattems emanating
from the tropical SSTs. A clear analysis of the concept of reproducibility can be
found in Shukla (1999).
-~!L::---::,-:I :-::::::=0.2
0.6
Fig. 8.4 Reproducibility index according Stern and Miyakoda (1995). Values of the index
less than 1.0 indicate areas of potential predictability with respect boundary forcing by Sea
Surface Temperature. In these regions the signal forced by the external forcing is greater
than the internal variability of the atmosphere and therefore the ensemble members tend to
reproduce the same seasonal anomalies.
143
the ratio between the inter ensemble variance and the total variance, averaged over
the duration of the simulation. Values larger than 1.0 indicate that the inter ensembIe separation (Spread) is larger than the total variance that includes the normal climate variability, whereas values smaller than 1.0 indicate that the realizations are
closer to each above the usual clutter of climate. It is possible to see how the reproducibility is space-dependent (and also time dependent, as Winter predictability is
different from Summer) with major reproducible areas in the subtropics. Interestingly, small areas of high reproducibility appear in selected portion of the middle
latitudes. These regions are those influenced by teleconnection pattems emanating
from the tropical SSTs. A clear analysis of the concept of reproducibility can be
found in Shukla (1999).
-~!L::---::,-:I :-::::::=0.2
0.6
Fig. 8.4 Reproducibility index according Stern and Miyakoda (1995). Values of the index
less than 1.0 indicate areas of potential predictability with respect boundary forcing by Sea
Surface Temperature. In these regions the signal forced by the external forcing is greater
than the internal variability of the atmosphere and therefore the ensemble members tend to
reproduce the same seasonal anomalies.