Page 196 - NGTU_paper_withoutVideo

Modern Geomatics Technologies and Applications

re-tracking has 15 cm lower RMSE at pass 595, 28 cm at pass 709 but 9 cm higher RMSE at pass 154 comparing to Level-2 data
which make it inappropriate in these passes. This method also shows 4 cm lower RMSE for pass 709 but 4 cm higher RMSE for
pass 595 and 3 cm for pass 154 comparing to original waveform re-tracking.
The table also shows that the mean waveform re-tracking approach has better results in pass 595 and 709 by 34 and 24cm
lower RMSEs but 20 cm higher RMSE for pass 154 comparing level-2 data. However this re-tracking approach shows 15 cm
better RMSE for pass 595, same results as pass709 but 14cm higher RMSE for pass 154 comparing to original waveform re-
tracking. This method also shows 19 cm lower RMSE for pass 595 but 4 cm higher RMSE for pass 709 and 11 cm for pass 154
comparing to first meaningful re-tracking approach.
Proposed re-tracking approach in this study shows 34 cm lower RMSE for pass 595 and 29 cm for pass 709 but 4 cm
higher for pass 154 comparing to Level-2 data. Comparing to original waveform re-tracking it shows 15, 5 cm and 2 cm better
RMSE for pass 595, 709 and pass 154. Comparing to first meaningful sub-waveform re-tracking it shows 19, 1 and 5 cm better
results for pass no. 595, 709 and 154 also 5 and 16 cm for pass 709 and 154 but same results for pass 595 comparing to mean
waveform re-tracking.
An overall looking confirms that for pass 709, proposed re-tracking approach had the best results than the other re-tracking
approaches comparing to level-2 data. For pass 595 the proposed re-tracking approach and mean waveform re-tracking showing
better results in comparison with level-2 data although the proposed re-tracking approach showed the most improvement for
pass 154 than other re-trackers. According to table-7 final SST TS evaluation shows 17 cm lower RMSE of Proposed re-tracking
approach comparing to level-2 data over the study area.

6. Summary and Conclusion

In this study sea surface topography (SST) monitoring over the Strait of Hormuz as one of the world’s famous economical
zones has been done during 2016/06/07 to 2019/06/11 using Sentinel-3A SRAL altimeter which is working in SAR mode with
high spatial resolution (~300m along track resolution). To study the waveforms 6 available passes which lay over the study area
have been chosen and to study the distorted waveforms and overcome to waveform contamination issues, 3 passes with highest
waveform corruption have been selected (pass no. 154, 595 and 709). Level-1 and Level-2 altimetry data processed to estimate
the SST over this area by different re-tracking algorithm including, original waveform re-tracking, first meaningful re-tracking,
mean waveform re-tracking and developed re-tracking called maximum correlation to mean waveform re-tracking using
threshold algorithm. The analysis of SST time series comparing to TG data records have been evaluated using RMSE and the
results are as follow:

 Better performance by Ocean and OCOG re-trackers showing that the waveform of the selected passes are coincide
better with these two re-trackers algorithm comparing to the others in level-2 data.
 The proposed method leads to better results than other re-tracking approaches (pass 709 and 154) and leads to better
results at least in two passes (595 and 709 by 34 and 29 cm lower RMSE) comparing to level-2 data.
 All the other re-trackers in pass 154 than level-2 data were unsuccessful to estimate the SST, but the proposed method
showed the minimum deviation comparing to the other algorithm in level-1 data.
 According to the previous studies threshold algorithm mostly represent better results comparing to the other re-trackers
and that is the reason, it has been used in this study. However, there could be more improvements using mathematical
re-trackers for pass 154.
 Re-tracking waveform in proposed method indirectly effected by all waveforms and instantaneous water level (unlike
mean waveform re-tracking which directly effected by distorted waveforms).
 The proposed method had much better results in term of RMSE (17 cm) comparing to Level-2 data by TG data records
evaluating.

9