Improving ENSO Simulation And Prediction Using An Intermediate Coupled Model With A Four-Dimensional Variational Data Assimilation Method

El Ni?o and Southern Oscillation(ENSO), the strongest signal of interannual variability on Earth, is a phenomenon involving ocean-atmosphere interactions in the tropical Pacific. In recent decades, great progress has been made in its mechanism understanding, theory development, numerical simulation and prediction. So far, ENSO forecast has been one of the most successful fields in terms of short-term climate forecasts. However, there still exist great challenges for ENSO-related studies. For example, the onset and diversity of ENSO events are still not fully understood, and it is still difficult to accurately make real-time predictions of the onset, development and transition during ENSO cycles. This thesis is focused on mechanism analyses, simulation and prediction of ENSO event using an improved Intermediate Coupled Model(ICM), including dynamical processes that can be responsible for the onset and evolution of ENSO events. Then, a four-dimensional variational(4D-Var) data assimilation system is successfully developed based on the ICM; experiments with optimal initialization and parameter estimation are performed to demonstrate the improved ENSO simulation and prediction. The main findings and innovative results are as follows:(1) The role of the temperature of subsurface water entrained in the mixed layer(Te) is reveled in the ENSO cycle, and a new mechanism for the onset of ENSO event is put forward. Unlike the well-known delayed oscillator mechanism associated with Rossby wave reflections at the western boundaries of the tropical Pacific, here we present a new mechanism for the onset of El Ni?o event based on this ICM simulation. Interannual sea surface temperature(SST) anomalies can be produced in the central basin off the equator along the Northern Equatorial Counter Current(NECC) pathways, which can induce westerly wind anomalies over the western tropical Pacific. Thereafter, the resultant SST and wind anomalies develop systematically, amplifying in strength and extending spatially into the equatorial region, triggering the onset of the El Ni?o. An example is given for the evolution of La Ni?a conditions in model year 2 to El Ni?o conditions in year 4. Right after a La Ni?a event(e.g., in year 2), there is a clear signature of reflections at the western boundary in early year 2, with related equatorial signals propagating eastward along the equator into the eastern basin in middle year 2. However, these reflected signals on the equator do not directly lead to an onset of an El Ni?o event at that time. Instead, with approximately one-year delay, a major El Ni?o event is seen to develop in the following year(late year 3), at a time when there is no reflected signal explicitly from the western boundary, indicating that the origin of the El Ni?o event cannot be directly ascribed to the reflection processes. Instead, Kelvin waves in the ocean that actually triggers the El Ni?o event in early year 3 are generated by interior wind anomalies near the date line that are associated with the first appearance of the warm SST anomalies off the equator. Persisted Te anomalies off the equator in the western tropical Pacific initiate the warm SST anomaly near the date line along the NECC region, which induces wind anomalies and an ocean-atmosphere coupling, leading to the El Ni?o event in year 4. The observed 1991-92 El Ni?o event is one of good examples fitting the processes discussed in this modeling work.(2) The roles of atmospheric wind forcing and subsurface thermal forcing both are demonstrated to be important in the second-year cooling of the 2010-12 La Ni?a event, which provides a theoretical foundation for improving ENSO forecasts. It has been seen that many coupled models fail to forecast the second-year cooling in 2011, following the major cooling in 2010. There are many factors affecting the real-time forecast, including propagating heat content anomalies, local thermocline effect(referred as Te anomaly) and wind forcing. The factors are coherently investigated using the ICM in this study within the context of real-time prediction of the 2010-12 La Ni?a event. Results show that two competing processes were at work in determining the SST condition in the eastern equatorial Pacific in 2011: a warming effect from the western tropical Pacific positive heat content propagation and a cooling effect locally associated with the negative Te anomalies(which was further associated with easterly wind forcing). The fate of the SST condition was thus determined by the effect which was dominant. As the persisted negative Te anomaly and easterly wind anomalies over the tropical Pacific appeared to make dominant contributions to the SST condition in 2011, the SST cooling re-emerged in 2011. The results of our study further demonstrate that the intensity of interannual wind forcing is equally important to the SST evolution during 2010-11 compared with that of the thermocline effect. To correctly predict the observed second year cooling, the ICM needs to adequately represent the intensity of both the wind forcing and the thermocline effects.(3) The real-time predictions of the 2015-16 El Ni?o event are made using the ICM. One striking feature associated with this event is the slow evolution of a warm SST anomaly in the western tropical Pacific through 2014 and early 2015; subsequently, the related ocean-atmosphere anomalies were coupled and amplified in spring 2015 and developed rapidly into a warm event in late spring 2015. The ICM predicted the warming and cooling tendencies of the SST well during 2014-16. Positive heat content anomalies accumulated in the western Pacific in 2014 and tended to propagate eastward along the equator in early 2015. Their arrival in the east in spring 2015 remotely influenced the subsurface temperature in the central-eastern equatorial Pacific, and created a significant warming effect on the SST there. Warm SST conditions emerged in the eastern equatorial Pacific in early spring 2015 and were accompanied by westerly wind anomalies in the west. The related anomalies of the SST and surface winds were amplified through this coupling, leading to rapid amplifications of the warm SST anomaly in late spring 2015. The system was then ready to evolve into a warm event. By summer 2015, large SST anomalies are predicted to occur in the central and eastern equatorial Pacific, indicating an El Ni?o event. The anomalies were amplified and reached a mature stage in late 2015. As the 2015 El Ni?o event developed, some related negative feedback processes came into play as well. For example, negative heat content anomalies emerged in the western tropical Pacific and propagated eastward along the equator. Its arrival to the east in early 2016 produced a cooling effect on the SST there, producing a cold SST anomaly. A transition to normal conditions is predicted to occur with a shift to cold conditions in mid and late 2016.(4) A 4D-Var data assimilation system is successfully developed based on the ICM. The ICM exhibits a good performance for ENSO simulation and prediction; we implement the 4D-Var method into the ICM, including the constructions of the tangent linear model and the adjoint model associated with the ICM and a minimization scheme. Additionally, strict testing is performed to validate the adjoint model and the minimization algorithm. Finally, after the 4D-Var method is successfully implemented into the ICM, a series of sensitivity experiments are conducted to optimize assimilation configurations and settings for better performances, including model parameters and data sampling.(5) The optimal initialization and parameter estimation based on the 4D-Var method are examined in terms of ENSO simulation and prediction. For the optimal initialization based on the 4D-Var method, it is demonstrated that model state fields obtained by assimilating SST anomalies into the ICM are more consistent with the truth fields compared with a non-assimilation case. For the parameter estimation, the model parameter, αTe, which represents the strength of the subsurface thermal forcing on the surface, is tested in addition to the SST anomalies that are assimilated into the model. The results indicate that optimizing both the initial conditions and model parameters can more effectively improve ENSO forecast. Thus, the ICM-based 4D-Var data assimilation system provides a modeling tool and theoretical guidance for improving the real-time ENSO forecasts. It is expected that this innovation modeling platform will have wide applications to future studies on ENSO analyses and predictions.