An Iterative Coastel Altimetry Retracking Based on Fuzzy Expert System for Improving Sea Surface Height Estimates
Idris, Nurul; Deng, Xiaoli
The University of Newcastle, AUSTRALIA
This paper presents an iterative waveform retracking system to improve sea surface height (SSH) estimates in coastal regions. In recent years, there is a healthy diversity of waveform retracking algorithms that has been beneficial to the coastal community. However, there is a lack of clear recommendation and guideline on which retracker should be used under various conditions. The iterative waveform retracking system based on the fuzzy approach has been updated and improved for selecting optimal resulted SSH from Jason-2 satellite in the area of Great Barrier Reef, Australia. The main principle of the system is to reprocess the altimeter waveforms using the optimal retracker, which is assigned base on the analysis from fuzzy expert system. When selecting the optimal retracker, the system considers both the physical features of waveforms and the statistical features of retracking results. This is to minimise the risk of assigning a wrong retracker when classification procedures are unable to precisely classify the corrupted waveforms. By including additional information in the fuzzy expert system to support the selection criteria of the optimal retracker, this system reduces the uncertainty on the value of altimeter-derived SSHs.
Results from this study have been compared with those from existing retrackers provided in the Sensor Geophysical Data Records and PISTACH products. It has been found that the retracked SSHs from this study archive higher improvement of percentage than those from the existing retrackers in almost all cases. Comparison with independent in-situ data has shown that there is a good agreement between sea levels from the retracking system and the tide gauge. The temporal correlation coefficients >0.8 are observed around Townsville and Bundaberg stations, suggesting that the derived sea level from this study can explains more than 80% of the total variance of sea levels. It suggests that the retracking system developed in this study is practical for improving SSH estimates near coasts by combining results from optimal retrackers selected through an iterative retracking strategy. This approach provides robust solution, especially when dealing with highly complicated waveforms near coasts. Therefore, results obtained in this paper significantly improve the precision of the estimated SSHs and efficiently reduce the altimetry no-data gap in coastal regions.