| Sea ice is an extremely sensitive factor in climate system.It is closely correlated to atmospheric circulation,ocean hydrodynamic circulation and the equilibrium of temperature,salinity,and heat throughout complex physical feedbacks.In recent years,driven by rapid variation of climate,the pace of thinning in sea ice fastened.Accurate observation of sea ice thickness will not only affect sea ice modeling,but also will have an impact on society and economics such as long-term response of climate and sea aviation.Due to the lack of datums for the thickness of sea ice based on measurement,continuous observation in wide range is still based on satellite datums and products of modellings.In this paper,we made a quality control on sea ice thickness data from CS2 SMOS which is an integration of satellite datums,APP-x which is inversed from satellite datums and PIOMAS which is a modeling product based on ULS,IMB and Ice Bridge measurement datums.Combined with knowledge-driven and data-driven scheme we sift out typical ocean-atmospheric factors which influence the thickness of sea ice.Then we improved the correction of deviation in the thickness of sea ice by combining the sifted factors,timed geological information,and machine-learning algorithm.For the first time we realize the separated recurrence of thick ice and thin ice considering the feature of set ice thickness products,which can efficiently promote the quality of APP-x datum.In order to obtain high-resolution sea ice thickness datasets,we integrate datums from multiple satellites based on Diva algorithm.For the first time we acquired daily sea ice thickness data with 5km resolution,amending the flaws of former coarse products.Finally,we discussed the application of refined product in climate analysis for arctic,risk assessment and regionalization of up-floating and ice-breaking of under-ice voyagers and ice-breaking of ships.Results show that refined products have a bright future in climate analysis and navigational security. |
