本文采用北美多模式集合(NMME) 1982年至2010年5个模式的月平均降水回报数据，ERA5再分析月平均降水数据进行验证，评估了单个模式及简单多模式集合对海南岛区域降水的的预报技巧和潜在可预报性，并研究两者之间的关系。受台风、暴雨等强对流影响，夏季是海南岛降水最集中的季节，对当地农业影响巨大，造成经济损失，因此特别对NMME模式在海南岛夏季降水的预报技巧进行了分析。发现各模式的潜在可预报性均明显高于模式预报技巧，说明对降水预测仍有较好提高空间，CFSv2预报技巧优于其他单个模式，距平相关系数为0.52，CM2.2和CanCM4的预报技巧较差，距平相关系数分别为0.25、0.20，随着预见期的增加，各模式的预报技巧均有所下降。简单集合平均的预报技巧处于最优模型和最差模型之间，距平相关系数为0.48。此外，进一步研究潜在可预报性和预报技巧之间的关系，发现存在一定的线性关系，潜在可预报性较高的模式其预报技巧也相对较高，比如CFSv2、CESM1、CCSM4，两者之间存在正相关关系。针对海南岛夏季降水的预报技巧研究，发现所有模式对海南岛北部及中部地区具有较好的预报技巧，而南部地区均较差，同时均方根误差也较大。具有较低预报技巧的CM2.2，其均方根误差也是所有模式中最大的，CFSv2对海南岛北部及中部地区具有较高的预报技巧，仍然高于其他模式，均方根误差也具有类似的空间分布，简单集合平均介于中间。
In this paper, five models, the hindcast monthly Precipitation from North American Multimodel Ensemble (NMME) during 1982 to 2010. and compared with ERA5 reanalysis monthly mean precipitation data to verify the prediction ability of each model and multimodel ensemble mean for re-gional precipitation in Hainan Island, evaluate their potential predictability. Affected by typhoons and rainstorms, precipitation in Hainan Island is most concentrated in summer, which has a great impact on local agriculture and causes economic losses. The forecast skill is lower than the potential predictability, indicates significant room for improvement Precipitation forecasting. CFSv2 forecast skill is higher to other single models. The AC is 0.52. The forecast skill of CM2.2 and CanCM4 is lower than other models, with the AC of 0.25 and 0.20 respectively. With the increase of the lead time, the forecast skill of each model is decreased. The ensemble mean forecast skill is between the best model and the worst model, the AC is 0.48. Here are statistically significant linear relationships be-tween forecast skill and predictability for NMME models, where models with higher predictability also have higher forecast skill, such as CFSv2, CESM1, CCSM4. According to the research on the forecasting skills of summer precipitation in Hainan Island, it is found that all single model have better forecasting skills for the northern and central parts of Hainan Island, while the southern part is poor, the RMSE is larger in the south and northwest of Hainan Island, but smaller in the middle of Hainan Island. The CFSv2 has a high forecasting skill for the northern and central parts of Hainan Island, which is still higher than other models. The RMSE of CFSv2 also has a similar spatial distribution, and the Ensemble mean is in the middle.
Becker, E., Dool, H.V.D. and Zhang, Q. (2014) Predictability and Forecast Skill in NMME. Journal of Climate, 27, 5891-5906. https://doi.org/10.1175/JCLI-D-13-00597.1
Ma, F., Yuan, X. and Ye, A.Z. (2015) Seasonal Drought Predictability and Forecast Skill over China. Journal of Geophysical Research Atmospheres, 120, 8264-8275. https://doi.org/10.1002/2015JD023185.
[7]
Yao, M.N., Yuan, X., et al. (2018) Superensemble Seasonal Fore-casting of Soil Moisture by NMME. International Journal of Climatology: A Journal of the Royal Meteorological Society, 38, 2565-2574.
https://doi.org/10.1002/joc.5436
[8]
Kirtman, B.P., Min, D., Infanti, J.M., et al. (2014) The North American Mul-timodel Ensemble: Phase-1 Seasonal-to-Interannual Prediction; Phase-2 toward Developing Intraseasonal Prediction. Bulletin of the American Meteorological Society, 95, 585-601. https://doi.org/10.1175/BAMS-D-12-00050.1
