Time and space scales for sea surface salinity in the tropical oceans

An unprecedented compilation of Sea Surface Salinity (SSS) data collected in the three tropical oceans (30°N–30°S) is examined for the period 1970–2003. The SSS data are derived mainly from 13 trans-oceanic Voluntary Observing Ship tracks and from 35 TAO/TRITON and 13 PIRATA mooring sites. Owing to the climatic importance of SSS, the data are used to present an overview of SSS variability, to evaluate the SSS time and space scales, and to assess the ability of the present in situ (and future satellite) observing system to adequately resolve the variability of interest. After quality control procedures, SSS is analyzed both as gridded fields (2° longitude, 1° latitude, 1 month) and as high-resolution records (0.02° latitude/longitude and one day). The gridded fields are used to describe the mean, standard deviation, and seasonal SSS cycle, as well as to estimate the characteristic scales for SSS. Results show that the monthly standard deviation of SSS mostly ranges within 0.1–0.3, except in regions with strong seasonal cycles (e.g., in the Atlantic and Pacific Inter-Tropical Convergence Zones and south of India) or prominent ENSO (El Niño Southern Oscillation) signals (e.g., in the western Pacific warm pool and in the South Pacific Convergence Zone) where standard deviations can range from 0.4 up to 1.4. The time scale for SSS is generally less than 3 months, while it reaches 4–8 months in regions where the variability is controlled mostly by ENSO. The meridional scale for SSS is generally 2–3° latitude, increasing to 4–7° latitude in the western Pacific warm pool and poleward of 10–20° latitude. The zonal scale for SSS is 15–20° longitude in the equatorial Pacific and in the South Pacific Convergence Zone (SPCZ), and 4–7° longitude along 8–10°N in the Indian Ocean. Decorrelation scales in the western Pacific warm pool tend to be smaller during La Niña time periods, with values approaching those found for the central Pacific. The signal to noise ratios are on average of the order of 1–1.5, meaning that part of the variability is not resolved from gridded fields. High-resolution records are thus used to assess the short time/space SSS variability on time and space scales shorter than the above-noted grid sizes. It is shown that the mean expected variability of SSS over a period of 10 days in a 1° latitude and 2° longitude box (i.e., the resolution requirement for the Global Ocean Data Assimilation Experiment (GODAE)) is about 0.2, with, however, notable regi