428 IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 10, NO. 3, MAY 2013 Assessment of Satellite-Derived Sea Surface Salinity in the Indian Ocean Smitha Ratheesh, Bhasha Mankad, Sujit Basu, Raj Kumar, and Rashmi Sharma Abstract—The study has been motivated by the desire to assess the performance of sea surface salinity (SSS) from the Soil Mois- ture and Ocean Salinity (SMOS) satellite launched by the Euro- pean Space Agency. Daily Level 3 product on a 0.25 ◦ × 0.25 ◦ grid for the year 2010 has been used for this assessment in the Indian Ocean. Various data sets, like the in situ data sets available from the Research Moored Array for African–Asian–Australian Monsoon Analysis and Prediction (RAMA) buoys and Argo floats and also the data sets from modular ocean model version 3 sim- ulations, have been utilized for this purpose. Comparison made at two buoy locations suggests good quality of SMOS SSS with root-mean-square error of the order of 0.36 and 0.34 psu. The triple collocation method, which explicitly takes into account the error characteristics of the SMOS, Argo, and model data sets, has been used for further validation of the SMOS data. Since the Indian Ocean exhibits characteristically different patterns of SSS in its different subregions, the study area has been divided into different such subregions. The SMOS-derived SSS appears to be of very good quality in the equatorial Indian Ocean and southern Indian Ocean, while the data are of poorer quality in the Arabian Sea and the Bay of Bengal possibly because of the errors in SSS retrieval due to the land contamination and strong winds. Index Terms—Argo floats, functional relationship (FR), modu- lar ocean model (MOM) version 3 (MOM3) ocean model, Research Moored Array for African–Asian–Australian Monsoon Analysis and Prediction (RAMA) buoys, Soil Moisture and Ocean Salinity (SMOS) salinity, triple collocation. I. I NTRODUCTION O CEAN salinity, along with temperature, controls the oceanic thermohaline circulation. Its knowledge is also essential for studying the variations of the mixed layer depth [1]. As far as the Indian Ocean is concerned, ocean salinity plays a vital role in ocean-atmosphere coupling, an important manifestation of which is the Indian Ocean dipole [2], [3]. Moreover, in the Indian Ocean, the steric part of the sea level rise associated with climate change is crucially dependent on the SSS variability [4]. The Global Ocean Data Assimilation Experiment specifies an accuracy requirement of 0.1 psu on a spatial resolution of 2 ◦ × 2 ◦ and a temporal resolution of ten days for large-scale circulation studies [5]. Recognizing the im- portance of salinity in the sea in general and sea surface salinity (SSS) in particular, two satellite missions have been proposed for measuring this parameter from space. The first of these mis- Manuscript received April 19, 2012; revised June 1, 2012 and June 14, 2012; accepted June 25, 2012. Date of publication August 13, 2012; date of current version November 24, 2012. The authors are with the Atmospheric and Oceanic Sciences Group, Space Applications Centre, Ahmedabad 380015, India (e-mail: smitha@sac.isro. gov.in; bhasha_mankad@yahoo.com; rumi_jhim@yahoo.com; rksharma@sac. isro.gov.in; rashmi@sac.isro.gov.in). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LGRS.2012.2207943 sions is the Soil Moisture and Ocean Salinity (SMOS) satellite of the European Space Agency (ESA), and the second one is the Aquarius satellite of the National Aeronautics and Space Ad- ministration. SMOS was launched in November 2009 and is the first satellite to measure SSS from space [5], [6].The repeat cy- cle of the satellite is 149 days, with 18 days as “almost repeat.” Three days is the period necessary for a full Earth coverage. In this letter, preliminary measurements of SSS from SMOS in the Indian Ocean are investigated. Our study closely follows a similar previous study where SSS from SMOS was investi- gated in the Atlantic [7]. However, in that paper, SMOS data of only one month, namely, September 2010, were investigated, whereas we have used data for the full 2010. Another major distinction of our study is the use of triple collocation functional relationship (FR) method, unlike the simple linear regression (LR) used in the mentioned study. In LR, it is tacitly assumed that the dependent variable (e.g., Argo-measured SSS or nu- merical model-generated SSS), against which the measured data (SMOS-measured SSS) are being compared, represents the “physical truth,” without any error. This is, of course, far from reality. On the other hand, the triple collocation FR method [8], [9] takes into account the different error characteristics of all the data sets in a very efficient manner. II. DATA AND METHODOLOGY ESA reprocessed SMOS-derived L1A data for the year 2010 based on up-to-date algorithm. Taking these data as input, the Centre Aval de Traitement des Donnees SMOS (CATDS)–CATDS Expertise Center- Ocean Salinity has pro- duced one year of composite Level 3 SSS data. These so-called “CATDS/CEC-OS SMOS Level 3 SSS research products” are available as monthly, ten-day, and daily composites. The Level 3 product version is V01. We have used the daily composites available as ten-day running averages. These are gridded at 0.25 ◦ × 0.25 ◦ resolution (http://www.catds.fr). The validation data set consists of the data available from the Research Moored Array for African–Asian–Australian Mon- soon Analysis and Prediction (RAMA) buoys and Argo floats. The RAMA buoys were deployed for the improved description, understanding, and prediction of east African, Asian, and Aus- tralian monsoon systems. The data have been obtained from the site www.pmel.noaa.gov/tao. The data were averaged in the same way as the SMOS data (as running ten-day averages com- puted daily) to facilitate proper comparison. Argo is an array of profiling floats reporting temperature and salinity from 2000 m to surface (4–5 m), and the Argo SSS fields were obtained from the official Argo site (http://www.argo.net). Although Argo does not measure salinity exactly at the sea surface, it is possible to use the shallowest Argo measurements as proxy for SSS. This is not a source of large error in regions with a 1545-598X/$31.00 © 2012 IEEE