High-frequency (HF) ocean radar has the potential to observe sea ice, but this application has not been thoroughly investigated. In this article, we explore sea ice observation with HF radar (HFR) and develop a corresponding signal processing procedure. We derived sea ice radial velocities from raw HFR data and determined angular positions using the MUltiple SIgnal Classification (MUSIC) algorithm. Sea ice velocities were evaluated by comparison with measurements from moored acoustic Doppler current profilers (ADCPs) and a drifting buoy. The root-mean-square (rms) differences between the sea ice velocities obtained by HFR and the ADCPs decreased with increasing radar signal-to-noise ratio (SNR), implying that a high SNR was associated with accurate measurement of sea ice velocity. These results agree with previous validations of HFR ocean current observations. The sea ice pattern mapped by HFR is consistent with the simultaneous C-band microwave radar observations. In particular, the HFR captured the approach of the sea ice edge toward the coast. Sharp attenuation of the HF signal over high concentration sea ice resulted in missing and/or low-SNR data at ranges beyond the front edge of pack ice, in qualitative agreement with model estimates of attenuation.

Sea surface salinity (SSS) data derived from the Aquarius mission were compared with global observation by Argo floats to evaluate the quality of the SSS and assess the error structure. The Aquarius SSSs agreed well with the Argo near-surface salinity with a root-mean-square (rms) difference of 0.42 psu (version 2.10.1 standard product). The residual of the salinity observation showed a significant correlation with the sea surface temperature (SST) and wind speed. SSS bias caused by inaccurate corrections for the galactic radiation was also discernible. Similar results were obtained for the Aquarius SSSs retrieved by the different algorithms. Aquarius monthly SSS averaged over 1° × 1° was also compared with outputs from a global ocean optimal interpolation system. The rms difference, calculated over 40°S-40°N, was 0.27 psu (version 2.10.1 standard product). The rms difference was reduced to 0.21 psu, which is very close to the goal of the Aquarius mission, for the product retrieved by an alternative algorithm (CAP version 25.1).

Sea surface salinity data derived from Aquarius and Soil Moisture and Ocean Salinity (SMOS) missions were compared with observations by Argo floats in the global oceans and outputs from a global ocean data assimilation system to evaluate the quality of the salinity measurements and assess the error structure in the data. Under the condition of sea surface temperature (SST) higher than 5°C and wind speed lower than 15 ms-1, salinity observed by Aquarius agreed well with the Argo data with an rms difference of 0.6 psu. If we applied spatial and temporal averages over 1° × 1° and one month, the rms difference was reduced to 0.3 psu. The residual of the salinity observation depended significantly on the SST and wind speed, as expected from the principle of salinity measurement. Also bias caused by inaccurate corrections for the galactic radiation and radio frequency interference were considered to contribute largely to differences with in-situ observations.

Sea surface salinity data derived from Aquarius on the SAC-D satellite were compared with observations by Argo floats in the global oceans to evaluate the quality of the Aquarius salinity measurement and assess the error structure in the data. In tropical and mid latitudes, salinity observed by Aquarius agreed well the Argo data with an rms difference of 0.6 psu. In high latitudes where the sea surface temperature is low, a large scatter of data with a deviation up to 2 psu was discernible. The residual of the salinity observation depended significantly on the sea surface temperature, wind speed and relative wind direction, as expected from the principle of salinity measurement.