The dissertation analyzes the spatio-temporal structure of ocean-atmosphere interannual variability associated with the El Niño - Southern Oscillation (ENSO) using COADS observations, ECMWF and NCEP reanalyses (including diagnosed 3D heating), and ocean data-assimilation products in order to advance our understanding of ENSO evolution. In the pursuit of this objective, the linkage between Asian summer monsoon and ENSO - the two global-scale climate systems - is investigated through numerical experiments with the modified Cane-Zebiak (CZ) coupled ocean-atmosphere dynamical model.

The ENSO mature phase in atmospheric observations and state-of-the-art climate model simulation (NCAR/CCM3) is obtained using the rotated principal component analysis (RPCA) of the year-round combined interannual variability of surface winds and SST. The ENSO covariant 3D structure of residually diagnosed diabatic heating is calculated from the ECMWF and NCEP reanalyses, and compared with the ENSO heating anomalies in the NCAR/CCM3 simulation. The ENSO deep heating anomalies are in general agreement, with positive heating in the equatorial central Pacific accompanied by diabatic cooling in the western Pacific and off-equatorial regions. While broadly similar, CCM3's ENSO heating can be characterized as being more of a meridional redistribution ("Hadley-like"), than a zonal one ("Walker-like") as in the reanalysis anomalies; comparison of the CCM3 and Xie/Arkin ENSO precipitation anomalies corroborates this assessment, and moreover shows ECMWF deep-heating to be more consistent with the precipitation anomaly structure. In the vertical, the CCM3 heating anomalies are notably larger in the 600-850mb layer, leading to a "bottom heavy" profile. CCM3's ENSO surface winds, on the other hand, contain stronger easterlies in the off-equatorial latitudes and stronger equatorward flow across the Pacific.

The ENSO covariant 3D circulation and thermal structure extracted from ECMWF reanalyses is used to examine the validity of the premises upon which extant simple atmospheric models are based. The 3-term surface momentum balance works well in the interannual variability case as well, but only with a 2-3 times larger meridional Rayleigh friction coefficient; the 3-term meridional momentum approximation however breaks down at 850mb, undermining Wang and Li's strategy for combining the Gill and Lindzen-Nigam models.

The RPCA of band-pass filtered surface and subsurface observations has lead to a clear separation of the constituent biennial and low-frequency (LF) oscillation modes inherent in ENSO variability, and therefore to accurate identification of the mature and transition phases of the LF oscillation mode in SST and ocean heat-content. The mature phases of the two modes are similar in the equatorial region, but their transition phases are radically different. Interestingly, the LF mode's transition phase in SST is found to closely resemble the well-known `North Pacific' SST mode.

An important component of the doctoral research was to investigate the impact of the Asian summer monsoon - ENSO feedback in the intermediate Cane-Zebiak Pacific anomaly model. The Asian summer-monsoon midtropspheric heating anomalies are parameterized in terms of the concurrent ENSO SST anomalies, and used as additional forcing in the CZ Gill-type atmospheric model. The Asian heating parameterization is developed from the RPCA of combined interannual variability of the tropical Pacific SSTs, tropical diabatic heating at 400mb (from ECMWF's reanalyses and operational analyses), and the 1000mb tropical winds during the 1979-1997 summer months of June, July and August.

Analysis of the 95,000-year long model integrations conducted with and without the interactive Asian-sector heating anomalies reveals that their influence on the Pacific surface winds leads to increased ENSO occurrence - an extra ENSO event every 20 years or so. An examination of the ENSO event distribution w.r.t.\ the peak SST-anomaly in the eastern equatorial Pacific shows a population-shift due to the strengthening of weak El Niño events in the monsoon run; this finding attests to the positive feedback between the two phenomena. The interaction of ENSO-related Asian summer monsoon heating with the CZ model's ocean-atmosphere also results in a wider period-distribution of ENSO variability, but with the El Niño peak-phase remaining seasonally locked with the Northern winter months.