PHY-04 Cross-scale interactions: mesoscale and smaller
Satellite-observed time and length scales of global sea surface salinity variability: a comparison of three satellite missions
Daling Li Yi* , Sun Yat-sen University School of Atmospheric Sciences

Sea surface salinity (SSS) observations from Aquarius, Soil Moisture and Ocean Salinity (SMOS), and Soil Moisture Active Passive (SMAP) satellite missions are compared to characterize the time and length scales of SSS variability globally. Overall, there is general agreement between the global patterns of the time and length scales of SSS variability estimated from the three satellite missions. The temporal scales of SSS variability vary from more than 90 days in the tropics to ~15 days in the Southern Ocean. The very short temporal scales (close to Nyquist period) in some parts of the ocean are probably due to the high level of noise in the satellite data or high noise-to-signal ratio. The longest temporal scales are observed along the South Pacific Convergence Zone (SPCZ) and in the central and western tropical Pacific. These areas are also related to the strongest ENSO-related signal in SSS. Dominant spatial scales of SSS variability are generally the longest (up to 150 km) in the tropics and the shortest (<60 km) in the subpolar regions. The distribution of the dominant spatial scales is not simply latitudinal but exhibits a more complex spatial pattern. In the tropics, there is slight east-west and inter-hemispheric asymmetry observed in the Pacific but absent in the other two oceans. In particular, the length scales are long (up to 150 km) in the eastern tropical Pacific (ETP), but relatively short (~100-120 km) in the western basin, contrary to the distribution of temporal scales. The analysis also reveals that the length scales of SSS variability are highly anisotropic. Except for the ETP and western tropical Atlantic, the meridional scales in the tropical belt are typically longer than the zonal ones and become more isotropic towards higher latitudes. Regional differences in the estimates of the scales from the three satellite SSS datasets are discussed and may arise due to differences in observation duration, spatial resolution and/or different levels of noise.