中国海表面风统计降尺度模型研究  

Studies of statistical downscaling model of sea surface wind over China seas

作　　者：李正泉[1] 肖晶晶[1] 马浩[1] 冯涛[1] 

机构地区：[1]浙江省气候中心,浙江杭州310017

出　　处：《海洋预报》2016年第4期71-79,共9页Marine Forecasts

基　　金：中国气象局气候变化专项(CCSF201427);浙江省气象科技计划重点项目(2014ZD05);公益性行业(气象)科研专项(GYHY201306050)

摘　　要：基于CCMP卫星风场数据和NECP-DOE再分析数据两者典型空间模态近20 a的长期统计关系,构建了中国海域海表面风的统计降尺度模型。降尺度模型交叉验证结果表明:模拟与观测两者风速在空间分布和变化趋势上均具有很好的吻合性,风速空间分布的相关性(R)可达到0.98以上,风速变化趋势的相关性(R)可达0.89以上。模型预测风速的均方根误差(RMSE)和绝对误差(AE),在绝大多数海区均低于0.25 m/s和0.30 m/s,相对误差(RE)<-5%或>5%的海区面积仅约占全海域面积的5%左右。从空间上而言,降尺度模型的模拟误差大值区多发生在陆域附近的近海区,主要原因是近海区影响风场的中小尺度天气因子众多。Based on long-term statistical relationship of typical spatial modes derived from CCMP wind data and NCEP-DOE reanalysis data, a statistical downscaling model of sea surface wind over China seas is constructed. The results of cross-validation indicate that the simulated wind speed agrees well with the wind speed from CCMP observation, both in spatial pattern and in annual variation. The correlation coefficient （R） of model simulating and CCMP observation is above 0.98 for wind speed spatial pattern, and higher than 0.89 for wind speed annual variation. Moreover, simulating errors of the statistical downscaling model are very low. In most of sea areas, RMSE of wind speed is lower than 0.25 m/s and AE is lower than 0.30 m/s. Only in about 5% sea areas, its RE is higher than 5% or lower than -5%. The errors spatial patterns show that large errors often occur in some offshore areas, because the offshore winds are affected by many factors that come from active local weather systems in meso-scale and small-scale.

关 键 词：海表面风 统计降尺度 经验正交函数 CCMP风场 NCEP-DOE 

分 类 号：P732[天文地球—海洋科学]