On the suitability of global algorithms for the retrieval of SST from the north Indian Ocean using NOAA/AVHRR data

Abstract

The errors associated with the SST retrievals from the north Indian Ocean, using global multichannel sea surface temperature (MCSST), nonlinear sea surface temperature (NLSST), and Pathfinder sea surface temperature (PFSST) algorithms are analysed using satellite-sea-truth matchups for NOAA-9 and NOAA-11. The trends associated with the SST retrievals with respect to various independent parameters and simple statistics are analysed to assess the performance of the algorithms. The MCSST and NLSST retrievals have undesirably large trends associated with surface temperature, satellite scan geometry and scan time (day/night); SST retrievals using PFSST show the lowest trends. The simple statistics, mean, and rmsd, are worst for MCSST (0.6 and 1.3 C); they are unsatisfactory for NLSST (0.4 and 0.7 C) and PFSST (0.4 and 0.7 C) also. Considerable improvement in statistics and trends are achievable by regionally optimizing the algorithm coefficients. The mean and RMSD of residuals drop to more than 50% of the values obtained by the use of global coefficients. Regionally optimized NLSST appear to be the better algorithm for the north Indian Ocean in the absence of time-dependent regional coefficients suggested by PFSST.