Quantifying the Predictability of Low-Resolution Medium-Range Weather Forecasts [report]

Craig H. Bishop
2001 unpublished
LONG-TERM GOALS To predict the probability distribution function (pdf) of medium range weather forecast errors as accurately as possible. OBJECTIVES Objective 1: To compare the Bishop et al.'s (2001) recently developed Ensemble Transform Kalman Filter (ET KF) ensemble generation technique against the breeding of growing vectors (BGV) technique (Toth and Kalnay, 1993, 1997) in a GCM. Objective 2: To quantify the limits of an ET KF ensemble that does not explicitly account for model error to
more » ... model error to predict forecast error variance in a GCM. Objective 3: To identify and remove (a) model error bias, (b) model error that correlates with variations in key parameters controlling the model's parameterizations of unresolved processes and (c) model error that correlates with deviations of the model trajectory about the climatological mean. (NWP failure to predict cold air damming due to poorly resolved topography is a fine example of a systematic model error that would correlate with the deviation of the model trajectory about the climate mean.) Objective 4: To create and test an ensemble generation scheme that accounts not only for the loss of predictability due to initial condition error but also for the loss of predictability due to model error.
doi:10.21236/ada625753 fatcat:xlz2fa5gy5b5jopmaf3yl77gqm