Sea ice, especially its types, is a sensitive indicator of climate change. As the dramatic decrease of Arctic sea ice area, it is increasingly dominated by thinner and younger first-year ice (FYI) instead of multiyear ice (MYI). Several sea ice type (SIT) products have been produced with microwave satellite observations since the late 1990s. However, they differ significantly in resolution, method and the involved satellite data, resulting in various performances. It remains insufficient in detailed comparison and accuracy evaluation among the products. This paper presents key results of inter-comparison and evaluation of five SIT products (Table 1) covering the winter seasons in the Arctic Basin. The SIT products from 2000 to 2020 was inter-compared with sea ice age (SIA) product and high resolution Synthetic Aperture Radar (SAR) images. Detailed performances of the five SIT products with methods analysis can be summarized as follows:
① The C3S-SIT product is based on a Bayesian algorithm and single radiometric parameter (gradient ratio), it has the longest temporal record however the largest fluctuation in the periphery of the Arctic basin. ② Unlike C3S-SIT, OSISAF-SIT uses radiometer data along with scatterometer data with higher. It provides more details on ice type distribution, however could not avoid the atmospheric influence. ③ The Zhang-SIT product is based on a feature-clustering algorithm and combination of passive and active microwave observations. It exhibits distinguished performance in the two scatterometer period, with good performance in 2000-2009 while prominent underestimation of MYI occurs after 2009. ④ KNMI-SIT employs a modified Bayesian algorithm and uses scatterometer measurements only as inputs. It performs best in the second half of winters, while has large MYI variations in the beginning of winters, especially with C-band scatterometer. ⑤ Similar as KNMI-SIT, IFREMER-SIT uses scatterometer data only and shows obvious underestimation of MYI.
Overall, the results show that the SIT products with combined scatterometer and radiometer data, with QSCAT than ASCAT scatterometer data, and with high resolution than low resolution have overall better performances on sea ice type discrimination. In addition, their performances vary with periods and regions. Their results are the most stable and credible in the Center Arctic in mid-winter, while large variations occur in the beginning and end of winters, mainly in the Beaufort Sea (BS) and East Siberia Sea (ESS).
发表评论