final_program_abstracts[1]

11 IMSC Session Program
Uncertain climate forecasts: When to use them, when to
ignore them, and when and how to adjust them
Wednesday - Parallel Session 11
Steve Jewson 1 and Ed Hawkins 2
1
Risk Management Solutions
2 Reading University, UK
Suppose that we have a forecast for a change in a climate variable (perhaps derived
from an ensemble mean), and that we also have a reasonable estimate of the
uncertainty around that forecast (perhaps derived from an ensemble spread). If the
ratio of the size of the change to the size of the uncertainty is large then it makes sense
to use the forecast. If, however, the ratio of the size of the change to the size of the
uncertainty is small, then, when minimizing expected mean squared error is the goal,
the forecast would be better ignored. In between these two extremes there is a grey
area where it may make sense to make a compromise between using and ignoring the
forecast by reducing the forecast towards zero.
First, we discuss this problem and its relation to standard statistical ideas of
overfitting, model selection, bias-variance tradeoff, shrinkage and biased estimation.
We then discuss the application of standard model selection rules (including BIC and
Bayes Factors) to decide between using a forecast and ignoring it.
Secondly, we discuss the development of new minimum mean squared error
shrinkage estimators that attempt to do the best we possibly can in the grey area (see
[3]). We call the new adjusted forecasts that result “damped” forecasts.
Thirdly, we apply these ideas to AR4 rainfall projections. We inflate the spread of the
AR4 ensemble to account for positive correlations due to common errors between
different models using the method described in [1]. We then test whether it would be
best to use, ignore, or „damp‟ the predictions, as a function of lead time. We find that
at short lead times it would be best to ignore the forecasts, at intermediate lead-times
it would be best to damp the forecasts, and at long lead times it would be best to use
the forecasts as is (see [2]).
Finally we discuss the relevance of these ideas to other kinds of forecasts such as
seasonal forecasts and decadal forecasts.
[1] “CMIP3 Ensemble Spread, Model Similarity, and Climate Prediction Uncertainty”
(2009), Jewson S. and Hawkins E., http://arxiv.org/abs/0909.1890
[2] “Improving the Expected Accuracy of Forecasts of Future Climate Using a Simple
Bias-Variance Tradeoff” (2009), Jewson S. and Hawkins E.,
http://arxiv.org/abs/0911.1904
[3] “Improving Uncertain Climate Forecasts Using a New Minimum Mean Square
Error Estimator for the Mean of the Normal Distribution” (2009), Jewson S. and
Hawkins E., http://arxiv.org/abs/0912.4395
Abstracts 193