Dias, D. F., A. Subramanian, L. Zanna and A. J. Miller, 2017:
Remote and Local Influences in Forecasting Pacific SST: A Linear Inverse
Model and a Multimodel Ensemble Study
Climate Dynamics, sub judice.
A suite of statistical linear inverse models (LIMs) are used to understand
the remote and local SST variability that influences SST predictions over the
North Pacific region. Observed monthly SST anomalies in the Pacific are
used to construct different regional LIMs for seasonal to decadal prediction.
The seasonal forecast skills of the LIMs are compared to that from three
operational forecast systems in the North American Multi-Model Ensemble
(NMME) revealing that the LIM has better skill in the Northeastern Pacific
than NMME models. The LIM is also found to have comparable forecast
skill for SST in the Tropical Pacific with NMME models. This skill, however,
is highly dependent on the initialization month, with forecasts initialized
during the summer having better skill than those initialized during the
winter. The data are also bandpass filtered into seasonal, interannual and
decadal time scales to identify the relationships between time scales using
the structure of the propagator matrix. Moreover, we investigate the influence
of the tropics and extra-tropics in the predictability of the SST over the region.
The extratropical North Pacific seems to be a source of predictability for the
Tropics on seasonal to interannual time scales, while the Tropics enhance the
predictability for the decadal component. These results indicate the importance
of temporal scale interactions in improving predictability on decadal
timescales. Hence, we show that LIMs are not only useful as benchmarks for
estimates of statistical skill, but also to isolate contributions to the forecast
skills from different timescales, spatial scales or even model components.