2个耦合模式中MJO的预报技巧

本文使用加拿大气候模拟与分析中心(Canadian Center for Climate Modeling and Analysis,CCCma)的耦合模式预报产品,应用以信息论为基础的可预报性理论框架,诊断、分析了耦合模式中Madden-Julian Oscillation(MJO)的预报率,包括实际预报技巧和潜在预报率,以及热带季节内尺度变率(Intraseasonal Variability,ISV)最可预报模态的空间分布。并在此基础上讨论了不同时间尺度平均对MJO预报技巧的影响。结果表明:本文使用的2个耦合模式中,MJO的预报技巧与目前全球主要使用的预报模式相近,约为10 d。潜在可预报技巧可以达到30 d以上。随着时间尺度从日平均增加到10 d平均,MJO的实际预报技巧与潜在可预报技巧都相应提高,尤其是潜在可预报技巧的提高更加显著。进一步分析发现,影响实际预报技巧的一个重要因素是初始条件MJO信号的强弱,当MJO信号很强时,预报技巧较高,反之则较低。本文最后分析了模式中ISV最可预报模态的空间分布,并讨论了如何利用这种最可预报空间分布提高ISV的实际预报技巧。

Forecast skill of MJO in two coupled models

Using the production of two coupled model from CCCma, and based on the information theory,the predictability of the Madden-Julian Oscillation including the actual forecast skill and potential predictability and the most predictable compnents of ISV were evaluated.The influence of different average time scale was also discussed. It is found that MJO has skillful forecast up to the lead time around ten days in these two coupled models, similar to the current forecast models.MJO can be forecasted potentially at more than one month of lead time. The actual forecast skill and potential predictability of MJO can be further developed as the averaging time scale increased from daily to ten days, especially the potential predictability increased more significantly. Further analysis found that the MJO signal strength of initial condition is the main factor to influence the forecast skill, if the MJO signal is strong, it has the best skill, vice versa. At last,the most predictable distributions of ISV were analyzed and the methods using such distributions to improve the actual forecast skill of ISV were discussed.