一种新的模式倾向误差估计算法及其在 ENSO 模拟中的应用

Climate models are important tools for us to understand, simulate and forecast the evolution of the climate. However, even with the current state-of-the-art coupled models, due to the inevitable systematic errors in the tendency equation of model, the model tendency error, the simulations and forecasts are still far from the true state of the atmosphere/ocean. Therefore, reducing the model tendency error is of great significance to improve the simulation and forecasting effect of the model. A novel algorithm was developed for estimating the tendency error of a model using assimilation technique with local ensemble transform Kalman filter (LETKF). The new algorithm was applied to the Zebiak-Cane (ZC) model to estimate the space-time dependent tendency error by assimilating the observed data of sea surface temperature anomaly (SSTA), and the calculated tendency error was used to correct the model, and then an integral simulation was carried out. Results reveal a high correlation between the tendency error and the simulation error of the ZC model. The corrected model improved some important characteristics of the simulation of El Niño-Southern Oscillation (ENSO). Overall, the new algorithm is very effective and simple computationally, shows good application value in ENSO simulation, and can be easily applied to various models, and thus shall be promoted.