Global Navigation Satellite System-Reflectometry (GNSS-R) uses GNSS signals as signals of opportunity as a multistatic radar. Most GNSS-R instruments conduct 1 ms coherent integration, followed by 500 or 1000 incoherent averages, leading to level-1 products [delay Doppler map (DDM)]. For remote sensing missions with higher spatio-temporal resolution requirements, raw data, and fewer incoherent averages are required for DDM computations. Fengyun-3E (FY-3E) GNSS Occultation Sounder II (GNOS-II) payload can acquire reflected signal's intermediate frequency (IF) raw data for specific areas, but there is no channel to record raw direct signals. Obtaining level-1 products DDM from raw data requires tracking the delay and Doppler frequency centroid coordinates, as they change during the incoherent integration time. Otherwise, the level-1 DDMs would appear blurring, which would result in wider DDMs and lower peaks. Besides, the geometry of transmitter-specular reflection point-receiver of GNOS-II is unobtainable, so classical algorithms cannot be used. Therefore, an innovative processing technique is presented, which can estimate the peak coordinates of all individual DDMs by appropriately grouping the individual DDMs and incoherently accumulating within each group. The feasibility of this method is demonstrated with representative data from sea, ice, and soil. Furthermore, incoherent averaging DDMs at a temporal resolution of 200 ms can well detect the boundary between sea, ice, and soil in high-latitude and complex environments while maintaining high quality. This work is an important basis for future analysis of the raw data from GNOS-II and will also inspire other work of individual DDM tracking where geometric information is agnostic.