The satellite-borne synthetic aperture radar (SAR) has been proven to be a valuable tool for high resolution ocean surface wind measurements. However, oceanic surface phenomena observed by SAR and oceanic processes which can cause the change of backscatter in SAR imagery will influence the SAR wind retrieval. Upwelling is one of the main factors, and it is prevalent in the East China Sea. It smooths the sea surface which results in a lower backscatter cross section in SAR imagery. In this study, using sea surface temperature and chlorophyll2-a data derived from Earth Observing System (EOS) MODIS, the low backscatter features in ENVISAT advanced synthetic aperture radar (ASAR) imagery are analyzed. A CMOD4 algorithm is adopted to retrieve the sea surface wind speed from SAR imagery. Results show that the wind speed is negatively biased due to the low normalized radar cross section associated with the upwelling. In order to resolve the impact of coastal upwelling on SAR wind retrieval, combined with high resolution numerical meteorological model wind field data, a wind speed correction method is proposed by using linear robust regression. To demonstrate the applicability of this method, underestimated wind speeds retrieved from ENVISAT ASAR images in the upwelling areas of Zhejiang coast and the sea area near northeast of Taiwan are corrected. Results show that the accuracy of upwelling region SAR wind retrieval data has been improved.