Progress and future prospects of decadal prediction and data assimilation: A review

Abstract

Decadal prediction, also known as “near-term climate prediction”, aims to forecast climate changes in the next 1–10 years and is a new focus in the fields of climate prediction and climate change research. It lies between seasonal-to-interannual predictions and long-term climate change projections, combining the aspects of both the initial value problem and external forcing problem. The core technique in decadal prediction lies in the accuracy and efficiency of the assimilation methods used to initialize the model, which aims to provide the model with accurate initial conditions that incorporate observed internal climate variabilities. The initialization of decadal predictions often involves assimilating oceanic observations within a coupled framework, in which the observed signals are transmitted through the coupled processes to other components such as the atmosphere and sea ice. However, recent studies have increasingly focused on exploring coupled data assimilation (CDA) in coupled ocean–atmosphere models, based on which it has been suggested that CDA has the potential to significantly enhance the skill of decadal predictions. This paper provides a comprehensive review of the research status in three aspects of this field: initialization methods, the predictability and prediction skill for decadal climate prediction, and the future development and challenges for decadal prediction.