Examining the performance of along-track multi-mission satellite altimetry – A case study for Sentinel-6

• Published in: Journal of Geodetic Science
• Main Authors: Heinoja, Lenne-Liisa, Delpeche-Ellmann, Nicole, Ellmann, Artu
• Format: Article in Journal/Newspaper
• Language: English
• Published: Walter de Gruyter GmbH 2023
• Subjects: Sea ice
• Online Access: http://dx.doi.org/10.1515/jogs-2022-0159; https://www.degruyter.com/document/doi/10.1515/jogs-2022-0159/xml; https://www.degruyter.com/document/doi/10.1515/jogs-2022-0159/pdf

Abstract Satellite altimetry (SA) is one of the most valuable techniques that measure the sea level data at both the near-coast and offshore. There exists, however, multiple challenges and hindrances in determining and using accurate sea level data. The most pertinent is that evaluation of SA performance requires that all data sources (such as tide gauges (TG) and hydrodynamic models (HDMs)) refer to the same vertical datum. Thus, knowledge of the geoid (equipotential surface of the earth) is essential in linking different sources of sea level. Accordingly, this study examines performance of along-track data for three satellite missions (Sentinel-3A, Jason-3, and Sentinel-6A) to obtain realistic sea level variation and to determine the accuracy of the various missions in the complex area of the eastern Baltic Sea. The methodology consisted of utilizing SA, HDM, and TG data and a high-resolution geoid model. Results show that root-mean-square error (RMSE) varied for Jason-3 within a range of 1.68–50.14 cm, Sentinel 3A with a range of 2.8–46.27 cm, and Sentinel 6A with a range of 3.5–43.90 cm. Sentinel 6A was determined to be the most accurate and reliable satellite mission. Results also showed higher RMSE (15.7–46.2 cm) during (i) the seasonal sea ice month (e.g. March 2018); (ii) at locations of several islands (e.g. eastern section of Gulf); and (iii) at locations where rivers discharged into the Gulf (e.g. Nava, Kemi, Luga, and Neva rivers). These features tended to show up as peaks in the final results even though robust data processing for outliers were undertaken. These results suggests that improvements can still be made in the SA retrackers and also in the data-processing techniques utilized.