Marine fishery fishing operations are carried out in different spatial and temporal scales, and the distribution of fish may vary with the spatial-temporal scale. Spatial and temporal scales play a vital role in the study of marine fishery habitat, but there were few researches on the comparison of habitat suitability index models in different spatial-temporal scales. In this study, based on the statistical data of Scomber japonicus from July to September in 2003—2011, combined with the data of sea surface temperature (SST) and sea surface height (SSH) obtained by ocean remote sensing, using arithmetic mean model (AM) and the relationship between SST and SSH to build habitat suitability index model (HSI). The study divided the sea surface temperature (SST) and sea surface height (SSH) into different spatial and temporal scales. Spatial scales included 0.25°×0.25°, 0.5°×0.5° and 1°×1°, temporal scales included weekly, ten days and monthly. A total of 9 HSI models were constructed in different spatial-temporal scales. The results indicated that spatial scale of 1°×1° and temporal scale of monthly were the most optimum spatial-temporal scales. When HSI value was greater than 0.6, the percentage of fishing effort was 75.42%, and when HSI value was less than 0.4, the percentage of fishing effort is 12.93%. According to the optimum spatial-temporal scales, using the data from July to September in 2012 to verify the optimum model, when HSI value was greater than 0.6, the percentage of fishing effort was 60.89%, and when HSI value was less than 0.4, the percentage of fishing effort is 13.88%. In summary, the spatial and temporal scales of fishing and marine environment factors should be considered during the construction of fishery habitat suitability index model.