Assessment of AMSR-E ASI sea ice concentration using ship observations and Landsat-7 ETM + imagery[J]. Journal of Remote Sensing, 2013, 17(3): 514-526. DOI: 10.11834/jrs.20132081.
而相应的均方根误差RMSE(Root Mean Squared Errors)从13.7%到33.8%。依据Landsat-7ETM+海冰分类
AMSR-E数据海冰密集度最大的误差产生于新冰区。
Abstract
The purpose of this research is to assess the accuracy of sea ice concentrations derived from AMSR-E data using ship observations from the 26 th Chinese National Antarctic Research Expedition ( CHINARE) and Landsat-7 ETM + Imagery in Antarctic. The comparison of AMSR-E sea ice concentrations with ship observational data indicated a general linear relationship between the two types of observations ( R2 = 0 . 816)
but two types of data didn’t match each other quite well. AMSR-E tended to underestimate sea ice concentrations ( compared to ETM + ice concentrations)
with average ice concentration biases ranging from - 5. 33% to - 21. 5% and corresponding Root Mean Square Errors ( RMSEs) ranging from 13. 7% to 33. 8% . For the Landsat-7 ETM + sea ice type classification
the l argest errors occurred in areas of new ice; in such areas
new ice was recognized as seawater by AMSR-E owing to its emission of passive microwave radiation
which is different to that emitted from other first-year ice.