High-resolution ocean remote sensing imaging is of vital importance for research in the field of ocean remote sensing. However, the available ocean remote sensing images are often averaged data, whose resolution is lower than the instant remote sensing images. In this paper, we propose a data transformation method to process remote sensing images in different locations and resolutions. We target satellite-derived sea surface temperature (SST) images as a specific case-study. In detail, we use a modified very deep super-resolution (VDSR) model as our baseline model and propose a data transformation method to improve the robustness of the model. Furthermore, we also illustrates how the degree of difference in the data distribution influences the model’s robustness and also, how our proposed data transformation method can improve the model’s robustness. Experiment results prove that our method is effective and our model is robust.