本文利用2012年夏季中国第五次北极科学考察期间雪龙船在北极东北航道走航观测的海冰密集度数据（OBS-SIC），初步评估了7种基于被动微波遥感的海冰密集度产品（PM-SIC）。7种PM-SIC因传感器和反演方法不同，分辨率差异较大（4-25 km）。在海盆尺度的海冰范围反演上7种PM-SIC基本相同，但对小范围浮冰区的反演差异较大。与MODIS可见光图像对比发现，MASAM数据（4 km）对局部小区域海冰刻画较好，是研究近岸区域或海峡岛屿海冰覆盖范围或面积时的首选产品；7种PM-SIC纬向平均后对比分析显示，不同PM-SIC对网格内是否存在海冰的判断基本一致，但对网格内海冰所占的比例（密集度）判断差异较大。结合OBS-SIC按航线、区域、密集度大小3种不同情况对7种PM-SIC进行分类定量评估，结果表明基于AMSR2传感器的AMSR2/ASI和AMSR2/Bootstrap数据与OBS-SIC偏差较小，平均偏差约±1%，均方根偏差仅11%和12%；而SSMIS/NT数据的偏差最大，平均偏差约-15%，均方根偏差为21%，其严重低估了网格内的海冰密集度值；因此具有更高分辨率的AMSR2/ASI数据（6.25 km）是关注海冰密集度大小时的首选产品。

The rapid decrease in Arctic sea ice makes normalized commercial shipping through Arctic passages possible. The accuracy of Sea-Ice Concentration (SIC) data is a crucial basis for Arctic shipping. Filed SIC data, however, is difficult to acquire. Passive microwave (PM) satellite is an efficient tool for obtaining large-scale SIC. Unfortunately, satellite SIC in the Arctic can only be evaluated with limited field observation data. Ship-based sea ice concentration observations (OBS-SIC) have been collected in the Antarctic to evaluate PM satellite sea-ice concentration (PM-SIC) (Worby, et al., 1999). In this paper, seven PM-SIC datasets that were released by Bremen University, NSIDC, and EUMETSAT were compared and assessed using ship-based OBS-SIC during the 5^th CHINARE Arctic Northeast Passage cruise from July to September 2012. A total of 604 OBS-SIC pairs that were obtained from approximately 20 days of the cruise is evaluated. We selected another 604 SIC pairs from PM-SIC datasets based on the same OBS-SIC latitude and longitude. To avoid bias from daily sea ice changes and different spatial resolutions, the daily mean PM-SIC and OBS-SIC for comparison is calculated using a method that is based on a similar evaluation work in Antarctica (Beitsch, et al., 2015). MODIS images are also used to evaluate the sea ice distribution near the continent, narrow strait, and island.
Results show that the seven satellite datasets have a similar pattern of large sea ice distribution, but have dissimilar patterns near the continent, island, and strait. MASAM successfully detected the small ice floe area near Poluostrov Taymyr on July 25, 2012, and near Ostrov Vrangelya on September 1, 2012, whereas other methods failed to do so. Latitude mean comparisons demonstrate that the seven PM-SIC have highly similar abilities to detect that the grid was completely water or sea ice, but highly differed in the detected percentage of sea ice in the ice grid. Quantitative evaluation via OBS-SIC comparison indicatesthatAMSR2/ASI, AMSR2/Bootstrap, SSMIS/ASI, and SSMIS/Bootstrap performed well, whereas SSMIS/NT and MASAM performedbadly.AMSR2/ASI has the lowest bias of 1% and root-mean-square error (RMSE) of 11%. However, SSMIS/NT largely underestimates the SIC with a mean bias of -15% and RMSE of 21%. AMSR2/ASI has a higher spatial resolution than the well-performing group. More importantly, it is updated near real time with only a delay of one day. High resolution and timely updates are the most important factors for operational ice service, which make AMSR2/ASI the best choice as areal-time shipping guide. High-resolution MASAM (4 km) can detect small sea ice distribution near the continent and narrow strait. Therefore, it is the most suitable for sea ice area and should be used for further studies in special regions. However, near-real-time higher-resolution AMSR2/ASI (6.25 km)has a smaller bias and RMSE with OBS-SIC. Hence, it is the best dataset for SIC quantity studies and real-time shipping guide.