Chairpersons of the supervisory committee:
Professors David S. Battisti and Edward S. Sarachik
Department of Atmospheric Sciences, University of Washington
Aspects of the El Nino-Southern Oscillation (ENSO) are investigated with
numerical models of the tropical Pacific region. The topics addressed
include: the relevance of the 'delayed oscillator theory' to the real
world ENSO cycle; the cause for the aperiodic variability in the
Zebiak-Cane (ZC) model's simulated ENSO cycle; the differences between
the ZC and Battisti (B88) coupled ocean-atmosphere models; and the
forecast skill of the B88 coupled ocean-atmosphere model.
The relevance of the delayed oscillator theory for ENSO is
investigated by analyzing the output from a wind-driven hindcast of
variability in a reduced gravity ocean model for the tropical Pacific
basin. The interannual variability in the output from the hindcast
simulation reproduces that found in observed tide-gauge data and is
shown to be consistent with the delayed oscillator scenario for the
evolution of individual warm and cold phases of the ENSO cycle. Output
from coupled ocean-atmosphere models that contain a simulated ENSO
cycle, along with that from a series of idealized cases, demonstrates
that simple lag/lead correlations predicted by pure delayed oscillator
theory are sensitive to the regularity of the ENSO cycle in the system
being studied.
The aperiodic interannual variability in the standard ZC model is
found to result from interactions between the mobile mode--a sub-annual,
westward propagating unstable mode of air-sea interaction--and the
interannual ENSO mode. The mobile mode is most active during the cold
phases of the model ENSO due to asymmetries in the prescribed
basic-state fields of sea surface temperature (SST) and zonal surface
current, and the non-linearity in the subsurface temperature
parameterization. The physics of the mobile mode is similar to that of a
coupled, gravest-oceanic-Rossby mode. The cyclic nature of the ENSO
mode, by itself well-described by the delayed oscillator theory, is
interrupted by ocean disturbances produced by the mobile mode. The B88
coupled model behavior is dominated by an interannual ENSO mode and does
not exhibit the mobile mode instability. The differing stability
characteristics of the ZC and B88 coupled models result from
dissimilarities in the parameter regimes in which the two models
operate.
Forecasts of an index for ENSO variability are carried out with
three different versions of the B88 coupled model. The best forecast
skill is obtained from a model that operates in a parameter regime that
is intermediate between that of the ZC and B88 models. Forecast skill is
found to be sensitive to the strength of the thermocline/SST coupling.
email:
mantua@atmos.washington.eduurl http://www.atmos.washington.edu/~mantua/abst.DISSERTATION.html