Abstract
Satellite altimetry has enhanced the understanding of ocean dynamics through high-rate sampling and global coverage. However, land contamination and bad reflection effects limit its accuracy. We present a geometrical method for retracking altimetry waveforms in coastal areas. Our method follows a geometrical assumption related to the symmetrical reciprocal motion of the radar pulse. Based on this assumption, the altimetry waveform is modelled as a continuous and differentiable third-order spline function, and the symmetry point of this function is considered as the retracking gate. The spline retracking algorithm is validated against the tide gauges at Onsala, Halmstad, and Muscat stations in Sweden and Oman, and its performance is compared with existing retracking algorithms. Our results showed a remarkable reduction of 50-91% in the unbiased-Root-Mean-Squared-Error (ubRMSE) and an increase of at least 13% in correlation coefficients when compared with other algorithms in Swedish coast. This algorithm presented equivalent results with the threshold and improved threshold retrackings in Muscat station, based on Jason-2 measurements. However, along the Jason-3 pass, our spline method showed a considerable reduction of 80% in ubRMSE and the minimum increase of 42% in correlation coefficients than the empirical algorithms. This method also outperformed the ALES algorithm in most cases.
Disclosure Statement
No potential conflict of interest was reported by the author(s).
Data Availability Statement
The SGDR files are accessible by the following link: https://www.ncei.noaa.gov/products/jason-satellite-products. Tide gauge data are also accessible by the following links: https://www.smhi.se/data/oceanografi/ladda-ner-oceanografiska-observationer and http://uhslc.soest.hawaii.edu/network/. The temporal resolutions of Onsala, Halmstad, and Muscat tide gauge data are 1, 10, and 60 min, respectively. The SSH time series of the ALES method are accessible by the following link: https://openadb.dgfi.tum.de/en/data_access/#.