冰速是海冰的重要参数，但是受资料长度限制，对卫星遥感冰速产品进行系统比较和评估的研究还较少。利用2009年—2017年国际北极浮标计划（IABP）浮标冰速数据，较系统地评估了不同时间间隔的遥感反演冰速产品在北冰洋内区和弗拉姆海峡附近海区的表现。结果显示，对北冰洋内区而言，1 d间隔的美国国家雪冰数据中心（NSIDC）遥感冰速冬季误差整体大于夏季，冬季高估了波弗特海南部海冰向西的运动，低估了从喀拉海流向格陵兰岛北部的穿极流。对于5种2 d间隔冰速产品而言，产品精度不完全取决于源数据空间分辨率的高低，反演算法的改进和数据融合均能提高冰速的精度，均方根误差大小的顺序是OSISAF-Merged <OSISAF-AMSR <Ifremer-AMSR <OSISAF-SSM < OSISAF-ASCAT。对于4种使用相同反演算法的3 d间隔的冰速产品，产品的均方根误差取决于源数据分辨率，空间分辨率最高的Ifremer-AMSR冰速均方根误差最小。比较不同种类、不同时间间隔冰速的月平均均方根误差后可知，采用3 d间隔反演的冰速由于能够忽略更多短时间尺度海冰运动，其冰速均方根误差低于2 d间隔的冰速。在弗拉姆海峡，除Ifremer-AMSR外，其余冰速产品均具有较大的经向偏差和均方根误差。在冰速较快弗拉姆海峡，冰速产品均方根误差取决于源数据的分辨率。

Sea ice motion derived from remote sensing is a key parameter for sea ice. Systematic comparison and assessment of satellite remote sensing ice motion products are still deficient because of limited data length.This study examines the performance of remote sensing ice velocity products with different time intervals in the central Arctic.The region around Fram Straitare systematically evaluated by employing the International Arctic Buoy Programme(IABP) buoy data from 2009 to 2017.The results show that for the central Arctic, the NSIDC ice velocity with one-day interval hasgreater error in winter than that in summer.In winter, the western icemotion in the southern part of the Beaufort Seais overestimated,whereasthe transpolar drift stream from the Kara Sea to the northern armof Greenland is underestimated. For five ice motion products with two-day interval, the accuracy does not depend completely on resolutions of source data.The improvement of the ice motion retrieval algorithm and the merging method can also improve the accuracy of the ice velocity.The order of the root mean square errors (RMSE) is OSISAF-Merged <OSISAF-AMSR<Ifremer-AMSR< OSISAF-SSM<OSISAF-ASCAT.The RMSEs of ice motion products with three-day interval, which employ the same retrieval algorithm, aredependent on the spatial resolution of the source data. Ifremer-AMSR has the lowest RMSEbecause of its source data has high spatial resolution. Ice velocity products with three-day interval canneglect sea ice movement in a short-time scale.The RMSE of ice velocity acquired is lower than that of thetwo-day interval. In Fram Strait, the other ice velocity products except for Ifremer-AMSR have large meridionalbiases and RMSEs.Moreover, theice velocity is fast in the strait, andthe RMSE of the ice velocity product depends on the resolution of the source data.