Knowledge of particulate organic carbon (POC) export flux (EP) is the key to understanding the marine biological carbon pump, which is an important process in ocean carbon sequestration. In this paper, a satellite remote sensing algorithm based on the food-web model established by Siegel et al. (2014) was used to estimate EP in the northern South China Sea (NSCS), which consists of the direct sinking flux of large phytoplankton and associated aggregates by gravity and flux of zooplankton feces by grazing. This is the first time to unveil the fine spatiotemporal variation of satellite derived EP in the NSCS. Compared with the results on 1° spatial scale, the model results based on the satellite products with 1/3° or 1/12° spatial scales showed better consistency with the observations. Validation with in situ EP showed that the model exhibited a fairly good performance in the NSCS basin, but the predicted EP was much smaller than the measurements in the shelf regions. The model-predicted annual mean EP in the shelf areas was 8.47 mmol C·m-2·d-1 and that in the NSCS basin was 5.56 mmol C·m-2·d-1, on the 1/3° spatial scale, with the relative difference of about 50% and 15% for the shelf and basin, respectively. The flux of feces from zooplankton grazing, especially that from micro-zooplankton grazing, might account for the major fraction of EP in the NSCS. Accurate products of OPP and phytoplankton carbon content and size distribution would be helpful in improving the performance of the satellite-based EP model.