Three-Dimensional Thermohaline Reconstruction Driven by Satellite Sea Surface Data Based on Sea Ice Seasonal Variation in the Arctic Ocean

Abstract

This study investigates and evaluates methods for the three-dimensional thermohaline reconstruction of the Arctic Ocean using multi-source observational data. A multivariate statistical regression model based on sea ice seasonal variation is developed, driving by satellite data, and in situ data is used to validate the model output. The study indicates that the multivariate statistical regression model effectively captures the characteristics of the three-dimensional thermohaline structure of the Arctic Ocean. Areas with large reconstruction errors are primarily observed in the salinity values of ice-free regions and the temperature values of ice-covered regions. The statistical regression experiments reveal that salinity errors in ice-free regions are caused by inaccuracies in the satellite salinity data, while temperature errors in ice-covered areas mainly result from the inadequate representation of the under-ice temperature structure of the reanalysis data. The continuous and stable thermohaline data produced using near real-time satellite data as input provide an important foundation for studying Arctic marine environmental characteristics and assessing climate change.

Keywords

• sea ice seasonal variation
• statistical regression model
• idealized experiment
• three-dimensional thermohaline structure
• reconstruction