El Nino-Southern Oscillation (ENSO) is a quasi-periodic phenomenon with significant global influence, primarily characterized by variations in sea surface temperature across the tropical Pacific and associated oscillations in surface air pressure. Over the past decades, ENSO has been a focal topic of research owing to its profound influence on global climate patterns, impacting phenomena such as the Indian monsoon, Peruvian coastal fisheries, and North American climate. In recent years, with the emergence and application of new tools such as advanced climate models and machine learning in ENSO research, a comprehensive review of past studies could serve to outline the accomplishments made and shed light on the outstanding questions that persist. By integrating these powerful tools, researchers can pursue unresolved questions and further advance our understanding of ENSO dynamics. In this paper, we reviewed past studies on ENSO from both scientific and practical perspectives. As science and practice are often interconnected and inseparable, we structured our paper in a logical sequence, covering classic ENSO theories, forecasting methods, asymmetry and decadal variations, changes in ancient climates, simulated ENSO in climate models, and the impacts of ENSO in other oceans. In the realm of science, we began by discussing some classic theories. We summarized the positive feedback mechanism proposed by Bjerknes for El Nino and highlighted its shortcoming of lacking negative feedback. In addition, we covered various negative feedback mechanisms proposed by others, such as the delayed oscillator, recharge-discharge oscillator, advective-reflective oscillator, and the West Pacific oscillator theories. Furthermore, given that asymmetry and decadal variations are important features of ENSO, we reviewed several relevant studies that proposed different mechanisms, including long Rossby wave dynamics, nonlinear dynamic heating and nonlinear temperature advection. Additionally, we reviewed the studies on ENSO in scenarios of ancient and future climates, revealing how ENSO's properties, such as strength and phase-locking, differ from those of the present. Finally, we summarized the ENSO's influence beyond the Pacific, encompassing the Indian Ocean Dipole and Indian Ocean Basin Mode, as well as impacts on the North Tropical Atlantic Mode in the Atlantic Ocean. In practical applications, we first introduced different forecasting methods, including statistical and dynamic models, each with its own advantages and disadvantages. We briefly overviewed a study using a dynamic model and then emphasized the statistical models. We outlined commonly used factors for predicting El Nino, including warm water volume, the West Pacific wind field, the space-time integration of Indo-Pacific winds, and equatorial flow fields, along with discussions on the current limitations in forecasting. Recent advancements in ENSO forecasting employing deep learning are also discussed for their potential to improve prediction lead time and accuracy. In addition, some researchers have explored ENSO by analyzing outputs from climate models. In this context, we summarized some studies utilizing models from various phases of the Coupled Model Intercomparison Project (CMIP). While some studies identified inaccuracies in simulated seawater temperature and salinity during ENSO events in climate models, others have indicated a gradual improvement in the simulation of ENSO within newer phases of CMIP. Despite significant progress in the aforementioned areas of ENSO, various questions remain unanswered. In the last section, we outlined shortcomings and challenges that require further exploration in both the scientific and practical realms of ENSO research.
