A Global Analysis of Sea Surface Temperature for Numerical Weather Prediction

Abstract

A global analysis of in situ observations of sea surface temperature (SST) developed for use at the Canadian Meteorological Centre is described. The analysis is done on the anomaly, the departure from climatology. The anomaly plus climatology, or resulting SST, is used as the lower boundary condition by the numerical weather prediction model. Since there is no ocean model to provide a background or first-guess field for the analysis, and since anomalies are observed to persist over long periods, the background field is obtained essentially by assuming persistence of the previous anomaly. The analysis algorithm is statistical interpolation. Attention is focused on techniques to control the quality of the observations, including a technique to remove systematic errors from ship observations. The analysis resolution is 0.9° and the correlation e-folding distance is 212 km. Verification of the analysis is presented using independent data from buoys and expendable bathythermographs for a one-year period. Verification is also presented for the National Centers for Environmental Prediction (NCEP, Washington) weekly analysis and for climatology. Results indicate that the analysis has skill over climatology in all regions and skill over the NCEP weekly analysis in the North Atlantic. In the rest of the Northern Hemisphere, analysis error estimates for the two analyses are similar, while in the Southern Hemisphere the NCEP analysis is superior, probably due to its use of satellite data. It is intended that this analysis will be an essential component of a debiasing algorithm for satellite SST observations.