面向航运服务的海冰密集度遥感产品对比研究

Comparison of remote sensing sea ice concentration products for Arctic shipping services

北极快速升温为北极通航窗口期延长提供了重大机遇,为北极航运和极区科考船只的空间信息提供服务,海冰边缘区浮冰遥感信息产品的可靠性非常关键。对比和分析了9种海冰密集度数据产品,并采用高分雷达数据开展面向海冰边缘区精度评估。结果表明,北极地区的9种海冰密集度产品相对其平均值的最大日偏差出现在夏季6—8月航道窗口期,其中采用被动微波高频算法的ASI产品和综合算法OSI SAF产品的整体平均偏差分别为0.59%和–0.65%。经对比发现各产品空间分布的不一致性主要存在于海冰边缘区。2018年,9种产品海冰范围的逐日最大和最小值之差为2.51×10^(6)~6.26×10^(6)km^(2)。在边缘区采用高分辨率雷达数据开展对比,结果表明在密集压实型海冰边缘,除基于ASI和NT2算法的三种产品外,其余六种产品均呈低密集度区高估、高密集度区低估的特点;基于AMSR2传感器的ASI-AMSR2和OSI SAF-AMSR2产品与雷达观测具有较高相关性,R^(2)分别为0.85和0.82,且均方根误差(RMSE)最小,分别为12.19%和9.70%。在分散型冰边缘,各产品对海冰密集度全面低估,OSI SAF-AMSR2相关性最高,R^(2)为0.78,RMSE最小,为6.19%。分析认为基于AMSR2的OSI SAF-AMSR2产品在现有9种产品中更能表征航道区海冰边缘与低密集区。

Recent rapid warming of the Arctic has promoted extension of the Arctic navigation period.However,the reliability of spatial information regarding the edge of the sea ice area,provided by remote sensing products,is critical for safe operation of Arctic shipping and polar research vessels.We compared and analyzed nine sea ice concentration(SIC)products,and used high-resolution synthetic aperture radar(SAR)data to evaluate their accuracy in determining the sea ice edge area.Results showed that the largest daily deviations from the mean values of the nine products occurred during the summer navigable window(June-August).Average deviations of ASI products with a passive microwave high-frequency algorithm and OSI SAF products with an integrated algorithm were 0.59%and−0.65%,respectively.The greatest inconsistency in spatial distribution among the nine products was found in the sea ice marginal area.In 2018,the difference between the daily maximum and minimum sea ice extent of the nine products ranged from 2.51×10^(6)to 6.26×10^(6)km^(2).The SAR data were used for comparison with the nine products in the ice margin area.In the closed pack sea ice marginal area,apart from the three products based on the ASI and NT2 algorithms,the other products overestimated(underestimated)the low(high)SIC area.The ASI and OSI SAF products based on AMSR2 sensors produced the highest correlation with the SAR data,with R^(2) values of 0.85 and 0.82,respectively,and root mean square errors(RMSEs)of 12.19%and 9.70%,respectively.However,in the drift sea ice edge area,all products underestimated SIC;the OSI SAF product had the highest R^(2) value(0.78)and lowest RMSE(6.19%).Of the nine products tested,the analysis indicated that OSI SAF products based on AMSR2 sensors could better represent the sea ice margin and low SIC areas in Arctic navigation channels.