AN IMPROVED SYSTEM FOR TROPICAL OCEAN SUBSURFACE TEMPERATURE ANALYSES

This study demonstrates techniques that lead to improved use of ocean thermal information and more useful and informative products for monit oring variability in the tropical oceans. The method is based on stati stical interpolation and is illustrated using analyses for the 20 degr ees C isotherm depth over the Pacific and Indian Oceans. A new monthly climatology is derived by exploiting the statistical interpolation me thod to provide an improved weighted estimate. The new climatology is shown to better represent key aspects of the tropical ocean thermal st ructure. A statistical forecast based on the previous analysis and cli matology significantly improves the analysis product, both in a qualit ative sense and as judged by quantitative measures of the skill of the forecast and of the estimated error of the analysis. Both the new cli matology and the statistical forecasting scheme are interpreted as str ategies for delivering enhanced information to the analysis system. A series of 10-day analyses are presented. It is shown that these analys es retain all the information contained in longer period analyses, at least when used with statistical forecasts, and that in addition they resolve higher-frequency events in the equatorial waveguide. The 10-da y analyses have around the same absolute estimated error as the longer period analyses but substantially higher accuracy due to the larger v ariance of the 10-day field. Some implications of these results are di scussed.