Based on the data of altimeter measurements of the Japan and Okhotsk sea surface heights, correlation dependencies are estimated between the sea surface height variations and the variability of geophysical and hydro-physical processes. With the help of cross-correlation analysis, the range of out-timing is determined among the best correlated predictors, and the actual term of sea surface height forecast is assessed. A linear multiple regression model is obtained with R² = 0,72 and calculation error probability that is lower than the significance value for all of the regression equation coefficients.