Abstract

The recently developed Bristol passive-microwave-satellite algorithm for deriving sea-ice concentration has possible theoretical and practical advantages over the widely used NASA/Team and Comiso algorithms. It was applied for the first time here to derive a long-term sea-ice series. The Bristol algorithm was reparameterized by manually tuning its brightness temperature tie-points both seasonally and interannually to systematically account for changing environmental conditions, satellite radiometer drift and differences in calibration and observing time between the three Special Sensor Microwave/Imager (SSM/I) sensors. In addition, an automatic algorithm was developed to remove residual noise in SSM/I images. The reparameterized Bristol algorithm performed well against the others tested in an Advanced Very High Resolution Radiometer (AVHRR) thermal infrared case-study validation. The SSM/I sea-ice extent was compared with Geosat radar altimetry and showed good agreement in winter; reasons are proposed to explain the summer difference. The optimized Antarctic sea-ice record spans the whole of the available SSM/I period (July 1987 to December 1997) and reveals a 3±0.3 (3±1.5) increase in extent (area); it has been used to study Antarctic sea-ice/atmosphere/ocean interactions and climatic couplings.