中国海洋再分析全球2.0 版（CORA2）是一个同时包含气象场和引潮力驱动的全球涡分辨率冰—海耦合再分析，提供的海流产品包含大洋环流和潮流。首先基于含水帆的SVP 浮标漂移速度，统计了2019 年CORA2 的日平均15 m深度海流误差，并与再分析产品GREP 和GLORYS12v1、卫星遥感海流分析产品GLOBCURRENT进行对比，评估它们对近表层大洋环流的再现能力，结果显示，CORA2 和GREP 信风带西向流和西风带东向流均比SVP 浮标偏弱；虽然各产品流速相对SVP 浮标偏低，但高分辨率CORA2 和GLORYS12v1 与低分辨率GREP 和GLOBCURRENT 相比，流速偏低减弱，且海流标准差更接近SVP 浮标，因此再分析分辨率提高有助于提升海流流速、增加海流波动；虽然CORA2 海流的均方根误差与相关系数不占优势，但是数值试验结果显示改进海面高度异常同化方法有利于提升CORA2 海流精度。其次，基于地波雷达观测资，利用误差统计与调和分析方法，评估了2022 年CORA2 和SMOC逐3 小时表层海流，结果显示，在潮流占主导的站位CORA2 潮流与观测十分吻合，且优于SMOC；但是在潮流偏弱的站位，CORA2 的潮流过强，导致其误差增大。

Abstract：The China Ocean Reanalysis Global Version 2.0 (CORA2) is a global eddy-resolving ice-ocean coupled reanalysis forced by atmospheric fields and tidal forcing, whose currents resolve both large-scale ocean circulations and tidal currents. The two components can be separated using a 24-hour averaging approach, in which the 24-hour averaged ocean currents approximate the large-scale circulations represented in CORA2. First, a statistical error analysis of CORA2 24-hour mean currents at 15 m depth was performed using 2019 drift velocities from SVP drifters with drogue. The results were benchmarked against the reanalysis products GREP and GLORYS12v1, alongside the satellite-derived current dataset GLOBCURRENT, to assess their performance in reproducing near-surface ocean circulation. The results reveal that CORA2 and GREP feature weak zonal westward (eastward) currents within the trade wind belt, alongside strong eastward (westward) currents in the westerly wind belt; GLORYS12v1 displays the inverse pattern. While current magnitudes from all datasets are underestimated relative to SVP drifter observations, this bias is far milder in the high-resolution CORA2 and GLORYS12v1 than in the coarser GREP and GLOBCURRENT. Additionally, their zonal and meridional velocity standard deviations better match the drifter measurements. These findings suggest that finer horizontal resolution improves simulated current magnitudes and better captures strong oceanic current variability. Though CORA2 does not outperform other datasets in terms of root-mean-square error (RMSE) and correlation coefficients for zonal and meridional velocity (U/V) components, sensitivity numerical experiments confirm that optimized sea level anomaly (SLA) assimilation schemes can substantially boost the accuracy of CORA2 surface currents. Second, a comparative evaluation of 3-hourly surface currents from CORA2 and SMOC was carried out using error statistics and harmonic analysis, based on 2022 High-Frequency (HF) Radar measurements at the LSIYKO, SPUDCN and NADHL stations. The results show that at stations dominated by strong tidal currents, CORA2- simulated tides agree closely with radar observations, making CORA2 markedly more accurate than SMOC. At stations with weak tidal and residual circulations but prominent Stokes drift, CORA2 considerably overpredicts tidal currents, resulting in elevated overall errors. At locations where tidal and residual currents are comparable in magnitude and both far exceed Stokes drift, CORA2 and SMOC each exhibit unique strengths.