The multi-scale variability of global sea surface temperature (GSST), which is often dominated by secular trends, significantly impacts global and regional climate change. Previous studies are mainly carried out under linear assumptions, even if the nonlinear trend features are discussed based on annual data, the conclusions of which are still incomplete due to several factors such as the sampling frequency and the time interval selection. Here, based on the Ensemble Empirical Mode Decomposition (EEMD) method, the robustness of GSST trends is further explored. The main features derived from the annual data are still maintained. However, monthly and seasonal data lag and mute the cooling in the equatorial central Pacific and the Southern Ocean in the Pacific sector while emphasizing more warming over most North Pacific in both magnitude and surface area. The results also highlight that early data causes a minimal effect on secular trends except for the portion near the left endpoint of the interval due to the local temporal nature of EEMD. Our research also clarifies that quadratic fitting cannot reveal all the meaningful evolution patterns, even as a nonlinear solution.