基于海冰分析图获取北极固定冰分布的方法研究

对北极沿岸地区固定冰的分析研究，将有助于更全面认识北极海冰的变化趋势，并为进一步探索北极航道的开发利用提供数据支撑。基于美国国家冰中心发布的2008-2018年逐周/逐两周海冰分析图发展了一个数据转化处理方法，准确提取了全北极逐周的固定冰分布格点数据。分析该时间序列发现，北极平均固定冰范围为（9.0±6.3）×10⁵ km²，呈现出（-1.0±3.9）×10³ km²·a^-1的减小趋势；4月份固定冰范围最大，多年平均值为1.6×10⁶ km²，处于北极冬季的1-6月份的范围均高于百万平方公里，而7-10月份为北半球夏季，几乎无固定冰存在。将MODIS图像作为基准数据对不同月份的七组结果进行对比验证，结果显示：结冰末期和融冰前期的平均偏差较小，如2012年3月12日的固定冰外缘线平均偏差为-0.8 km；而融冰末期的偏差较大，如2009年9月14日的固定冰外缘线平均偏差为-9.4 km；但所有月份的验证误差都小于格点网格的大小（25 km），因此认为该方法是可行的，得到的格点数据可以代表固定冰的真实分布情况。

Study on the method of extracting Arctic landfast ice distribution based on sea ice analysis charts

Analysis of landfast ice in Arctic coastal areas is very important to provide a comprehensive understanding of the variations of Arctic sea ice, and to generate further basis data on the development and utilization of the Arctic shipping routes. Based on the weekly/biweekly sea ice analysis chart (2008-2018) released by the National Ice Center (NIC), we present a conversion and processing method to accurately extract the gridded data of the whole Arctic landfast ice distribution. The results showed that the average extent of Arctic landfast ice was (9.0±6.3)×10⁵ km² and has gradually decreased with an average rate of (-1.0±3.9)×10³ km²·a^-1. The landfast ice extent was the largest in April, with a multi-year mean of 1.6×10⁶ km². The extent from January to June in the Arctic winter was more than 1 million square kilometers, while almost no landfast ice in July-October, which was named Arctic summer. The MODIS images were used to verify the seven others in various months. The results showed that the average deviation in the end of freeze-up and the start of breakup was smaller. For example, the average deviation of the outer edge of the landfast ice was -0.8 km on March 12, 2012, and the deviation in the end of breakup was larger, such as the average deviation of the outer edge of the landfast ice was -9.4 km on September 14, 2009. However the verification error of all months was smaller than the grid size (25 km). Therefore, the method is considered to be credible, and the extracted gridded data may represent the true distribution of landfast ice.