Melt pond fraction over Arctic sea ice during 2000-2019@en

This dataset includes melt pond fraction (MPF) over Arctic sea ice during 2000-2019. The spatial coverage of this data is north of 60 °N. The MPF data is projected on a polar stereographic grid with a spatial resolution of 12.5 km and a temporal resolution of 8-day intervals from May 9 to September 6, which is archived in the NetCDF format. This dataset was jointly developed by the Beijing Normal University, University at Albany-State University of New York, and Sun Yat-sen University. Large-scale temporal and spatial distribution of melt ponds over Arctic sea ice have implications for surface albedo, heat and mass balance of sea ice, freshwater in the upper ocean, and primary productivity of ice algae and phytoplankton. We retrieved the MPF data based on a robust ensemble-based deep neural network along with the surface reflectance of 7 bands from MOD09A1 (MODIS surface reflectance 8-Day L3 version 6) as the input and the MPF observations from multiple sources as the target. The validation results show that the retrieved MPF is in good agreement with the in-situ measurements (the details can be found at Ding et al., 2020).本数据集包含2000-2019年北极海冰的融池占比（melt pond fraction, MPF）。该数据的空间覆盖范围为北纬60°以北。 MPF数据以极射赤平投影网格（polar stereographic grid）投影，空间分辨率为12.5 km，时间分辨率为8天间隔，时间覆盖范围为每年5月9日至9月6日，以NetCDF格式存储。本数据集由北京师范大学、纽约州立大学奥尔巴尼分校与中山大学联合研制。 北极海冰融池的大规模时空分布，对地表反照率、海冰的热量与质量平衡、上层海洋淡水储量以及冰藻和浮游植物的初级生产力均具有重要研究意义。本研究基于鲁棒集成深度神经网络，以MOD09A1（MODIS地表反射率8天合成L3级第6版产品）的7个波段地表反射率作为输入，以多源MPF观测数据作为监督标签，反演得到MPF数据。验证结果表明，反演得到的MPF与原位观测值吻合良好，详细内容可参见Ding等人2020年的研究成果。