Toward Dynamical Annual to Decadal Climate Prediction Using the IAP‐CAS Model

Authors:

Abstract
Annual to decadal (A2D) climate prediction provides key insights for public policy and individual decision‐making over the next 1–10 years, but most current dynamical models exhibit limited skill at the A2D scale. To address this challenge, the IAP‐CAS A2D dynamical ensemble climate prediction system has been developed by expanding the existing operational sub‐seasonal to seasonal (S2S) prediction system approved by the WMO/WWRP S2S panel. Using a full‐field atmosphere‐ocean initialization experiment which covers the period from 1981 to 2015, several key findings are revealed: First, the model demonstrates significant positive skill for regional surface temperature predictions globally, except for the North Atlantic, likely due to the initial shock. Despite this, the model effectively captures the global mean surface temperature warming trend. Second, the model exhibits relatively high predictability for the Pacific Decadal Oscillation (PDO), with correlation skill up to 3 years, comparable to the sixth Coupled Model Intercomparison Project Decadal Climate Prediction Project multi‐model ensemble mean. The spread‐error ratios close to 1 in the PDO predictions indicate high reliability. Additionally, the model shows significant skill in predicting the El Niño‐Southern Oscillation (ENSO) for up to 1 year, comparable to leading seasonal dynamical prediction models. Further analysis reveals an established teleconnection between ENSO and the North Pacific atmosphere in the IAP‐CAS model, likely underpinning the PDO predictive skill at forecast year 1. This study also assesses the effect of initialization by comparing initialized hindcast data with uninitialized historical simulations.
Funding Information
  • National Natural Science Foundation of China (42475155, 42175161)

This publication has 40 references indexed in Scilit: