Objectives Sea ice concentration (SIC), which is defined as the proportion of a given area of ocean that is covered by ice, is a significant parameter for Arctic sea ice and climate change, an important input for regional climate model and numerical weather prediction model. The passive microwave sensors are able to penetrate the atmosphere and clouds, regardless of observation time, and their data is commonly used to retrieve SIC. Passive microwave obtained from the microwave radiation imager (MWRI) aboard on the Chinese Fengyun-3 (FY-3) series satellites can be used to extract polar SIC. However, the study on its accuracy is limited.
Methods This paper focuses on the application of MWRI to Arctic SIC estimation accuracy, and compares the SIC obtained using special sensor microwave imager sounder (SSMIS), MWRI and advanced microwave scanning radiometer 2 (AMSR2) data, and utilizes Bootstrap, NASA Team(NT), fully constrained least squares (FCLS), Enhanced NASA Team (NT2) and Arctic radiation and turbulence interaction study (ARTIST) sea ice (ASI) methods with the in-situ data.
Results Validation results show that MWRI has the smallest root mean square error (RMSE) at different spatial resolution (20.4%-24.4%). FCLS performed better than the other five algorithms for three passive microwave data because the error was considered during the SIC retrieve, the SIC was constrained with non-negative constrain, and numerical optimization was used to solved the SICs. In addition, MWRI performs well in summer (17.9%-23.0%) and winter (11.2%-17.8%).
Conclusions MWRI has relatively stable performance and a great potential for sea ice monitoring. Uncertainty also exists in evaluation, such as the uncertainty of accuracy of in-situ data in Arctic and tie points. Furthermore, scale difference between multi-source data could also induce the uncertainty of validation.