An evaluation of satellite and in situ based sea surface temperature datasets in the North Indian Ocean region

Abstract

Satellite based daily fields of Pathfinder SST (PFSST) and blended-analysed fields like National Center for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) and Reynolds weekly SST data were compared with the in situ measurements obtained from several drifting buoys and a moored buoy in the north Indian Ocean. The mean differences (bias) and scatter (root mean square deviation (RMSD)) were higher than the ideally expected values of < 0.01° C and 0.5° C for all three datasets. There was very little difference in the error statistics from one region to another. The error statistics differed significantly from year to year. The PFSST fields reported cooler SSTs ~0.5° C, during August 1991 to April 1992, due to the increased input of aerosols in the atmosphere caused by the eruption of Mount Pinatubo. Being a high resolution dataset, the PFSST is best able to mimic the sudden changes in the in situ SSTs that happen over a couple of days to a couple of weeks. Neither NCEP/NCAR data nor the Reynolds data were able to mimic the changes in SST that occur over a period of a few days. All three datasets are useful for the analysis of low frequency oscillations like annual and semi-annual cycles. PFSST and NCEP/NCAR SSTs are also useful for the analysis of higher frequency oscillations in the SST fields such as the 26-day period oscillations in the equatorial Indian Ocean.