An investigation of low-frequency planetary wave forcing and predictability in a simple global atmospheric circulation model

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)


Meteorology and Physical Oceanography

First Committee Member

Naomi Surgi, Committee Chair


Dynamically consistent data generated by a simple 2-level global spectral model are used in conjunction with a diagnostic method to study the forcing of large-scale low-frequency (LF) atmospheric fluctuations. Run in perpetual January mode at moderate spatial resolution (i.e., R15 truncation), the model realistically simulates pertinent features of the observed time-mean climate and variability. The diagnostic method consists of calculating, from the R15 model output, the terms of the governing equations of a hypothetical low-order model that simulates only the slowly-evolving planetary-scale flow. The right hand sides of these low-pass-filtered equations are expressed in terms of resolved forcing and unresolved forcing, with the former representing autonomous processes involving only the large-scale LF components of the flow, and the latter containing the effects of high-frequency (HF) transient eddies.The roles of resolved and unresolved forcing during the various stages of composite positive and negative persistent anomalies are assessed. During anomaly maintenance, HF transient eddies anchor anomalies against downstream propagation and resist local decay. In contrast, the thermal forcing of the HF eddies acts in a destructive sense. Applying the diagnostic method to the onset stage suggests that autonomous processes typically determine the local growth of both positive and negative anomalies, consistent with interpreting their origin as a large-scale instability. The development of anomalies in HF transient eddy forcing, in general, stem from the evolution of the large-scale flow, rather than acting independently to initiate the large-scale anomaly development.A low-order model is constructed to simulate the dynamics of slowly-evolving planetary-scale flow. Relationships between the resolved and unresolved forcing gleaned from the diagnostic phase of the study are exploited in the development of a quasi-stochastic parameterization of the latter in terms of the former. The skill of the low-order model is tested against the low-pass-filtered output from the R15 model by performing a series of Monte Carlo experiments. The low-order model produces skillful forecasts out to ten days. RMS errors in the upper layer streamfunction predictions grow most rapidly in the storm track regions, and are correlated most strongly with the HF transient kinetic energy patterns.


Physics, Atmospheric Science

Link to Full Text