The application of statistical computation for fitting the global sea temperature

We present the statistical fittings for global sea temperatures using the data of the World Ocean Atlas (WOA). We have gridded fittings for annual, monthly, and seasonal means of temperatures on standard levels (typically at 33 depths). To decrease bias, we apply the statistical regression models which combine the high-order polynomials with the mutually orthogonal polynomials. The comparison of the fitted results with the data of WOA are given, which demonstrate the accurateness of the fitted results.