New methods for the assessment and calibration of ensemble
temperature forecasts
Stephen Jewson
Risk Management Solutions Inc., London
Abstract:
We describe a new set of methods for the assessment and calibration of
ensemble temperature forecasts that involve optimal linear transformations
of the ensemble mean and ensemble standard deviation. Optimality
is defined as maximisation of the likelihood of the observations given the
calibrated forecast. These methods give various new insights
into predictability. For instance, we are able to show that the ECMWF
seasonal forecasts for Nino3 contain almost no useful information
in the ensemble spread. We also find that the predictable component of flow
dependent uncertainty in the ECMWF medium range
forecast is very small, and that forecast quality is not compromised by
ignoring the ensemble spread and using only the ensemble mean
and past forecast error statistics to build probabilistic forecasts.
Jewson, S., A. Brix and C. Ziehmann 2003:
A New Framework for the Assessment and Calibration of
Medium Range Ensemble Temperature Forecasts.
Jewson, S., F. Doblas-Reyes and R. Hagedorn 2003:
Assesment and Calibration of Ensemble Seasonal Forecasts
of Equatorial Pacific Ocean temperature and the Predictability
of Uncertainty.