Abstract: In response to the accuracy issue in tidal signal calculation during Global Navigation Satellite System (GNSS) buoy ocean measurements, particularly in high-noise conditions affecting wave parameter calculations, this paper proposes a novel step detection and restoration method based on the sliding window cumulative sum (CUSUM) algorithm, which has been extensively utilized in non-intrusive load monitoring. The algorithm is applied to detect and correct step discontinuities in the sea surface height (SSH) obtained from precise point positioning (PPP) computations and the significant wave height (SWH) extracted from the SSH time series. The performance of the method is evaluated by comparing it with reference data from tide gauge stations and dedicated wave buoys. The results demonstrate that the proposed method significantly improves the accuracy and reliability of GNSS technology in inverting wave parameters. The root mean square error (RMSE) of SSH is enhanced by 75.5%, and the correlation is increased by 7.46%. Moreover, the RMSE of SWH is improved by 65.22%, and the correlation is boosted by 208.28%. These findings underscore the effectiveness of the proposed method in enhancing the accuracy of wave parameter extraction using GNSS technology. The method's implications for enhancing marine engineering safety and economic benefits are also highlighted, making it a valuable contribution to GNSS step detection and providing valuable insights into the extraction and application of tidal signals using GNSS technology.