Evaluations of sst climatologies in the tropical pacific ocean

Abstract

Accuracies of a total of six sea surface temperature (SST) climatologies are investigated in the tropical Pacific Ocean, spanning latitudes from 30°N to 30°S. Observation-based climatologies are directly obtained from their original sources and typically formed from satellite measurements, in situ data, or a combination of both. Other SST climatologies are constructed from reanalyses of a numerical weather prediction center. All products have different features, each with its own data sources, grid resolutions, interpolation methods, sampling periods, and biases. Even though time periods during which climatologies were formed differ, all products have good agreement with each other, with basin-averaged mean bias values of almost zero and RMS SST differences typically <0.3°C over the annual cycle. Nondimensional skill scores between the pairs of SST products are close to 1, indicating almost perfect agreement everywhere except in the western equatorial Pacific. Comparisons against an ensemble climatology based on the average of all SST climatologies further reveal the weaknesses and strengths of each product. Additional validations are performed by forming climatologies using SSTs from moored buoys of the Tropical Atmosphere Ocean (TAO) array during 1986-2005. Overall, very small cold biases of ≈0.2°C and near-perfect skill scores are found for all SST climatologies when comparing them with 408-month-long SST time series from 34 buoys. In general, all SST products are generally accurate enough to be used for various climate applications. © 2009.