HR: 08:30h
AN: A31A-01 INVITED [Abstracts]
TI: Dynamics of the most predictable patterns in week two forecasts
AU: * Whitaker, J
EM: jeffrey.s.whitaker@noaa.gov
AF: NOAA Climate Diagnostics Center, 325 Broadway R/CDC1, Boulder, CO 80305 United States
AB:
The phenomena that yield skill in the second week of a forecast are generally large scale and low frequency, and hence there
may be only a few independent samples of these events each season. In addition, the predictable signal may be small compared to the uncertainty in a single forecast, so ensembles may be needed to extract that signal. Therefore, quantifying the
nature of the predictable signal in week two therefore requires a large sample of ensemble forecasts with a fixed model,
spanning many years. Using the CDC MRF reforecast dataset (a 25-year dataset of ensemble forecasts with a fixed model), we
will attempt to shed light on a few basic questions such as "How much predictive skill is there in week 2?" and "Where does
that skill come from?". A canonical correlation analysis is performed to isolate the most predictable patterns in week two
for Northern Hemisphere winter. The three most predictable patterns are very similar to the "Tropical/Northern Hemisphere"
(TNH), "Pacific/North American" (PNA) and "North Atlantic Oscillation" (NAO) teleconnection patterns identified in previous
studies of low-frequency variability. Regression analyses are used to elucidate the relevant mechanisms responsible for the
remarkable predictability of these patterns, including the role of tropical convective forcing, large-scale eddy-mean flow
interactions, and synoptic-scale transient eddy feedbacks.
DE: 3319 General circulation
DE: 3337 Numerical modeling and data assimilation
DE: 3364 Synoptic-scale meteorology
SC: Atmospheric Sciences [A]
MN: 2005 Joint Assembly