Retracking in the (NW) Mediterranean Sea

Size: px

Start display at page:

Download "Retracking in the (NW) Mediterranean Sea"

Bryce Johnathan Newton
6 years ago
Views:

1 Retracking in the (NW) Mediterranean Sea L. Fenoglio-Marc Institute of Physical Geodesy, Technische Universität Darmstadt, Germany

2 Outline Motivation (RECOSETO) Coastal data Re-tracking Validation - Along-track altimetry - altimetry and tide gauge sea level Conclusion 2

3 RECOSETO Selection criteria for altimetry Effect of land and ocean characteristics Re-tracking Validation 3

4 RECOSETO Data & Methods Altimeter 4

5 Coastal Data from altimetry > 50 km from coast How near coasts can altimetry data be? TX : 0-15 km J1 : 27! km TX : 0-6 km J1 : 15! km n1waveforms 5

6 Coastal Data % % of passes Sea Land < 5 km Envisat 5% more than TP Land Sea < 5 km Envisat 15% passes, no TP Distance to coast < 10 km 100% all passes n1waveforms 6

7 Coastal data How close to coast? - sea-land vrs land-sea : s-l is the most advantageous (Genova : Envisat (s-l) at smaller minimum distance than TP (l-s), Imperia : TP (l-s) nearer to land than Envisat (l-s)). - Retracked data vrs not-retracked : Smaller distance for TP RGDR wrt TP GDR 7

8 Comparison sea level TG and altimetry (2) Coastal data Imperia (n1 216, T/P 009) Genova (n1 801, pass 044) N1 versus TP (same n) - higher correlation - lower RMS Correlation with TG minimum km from TG N1 RADS & GDR very similar. 8

9 Retracking - Non-parametric (empirical methods) - Geometrical shape fitted to the waveforms defined in terms of position of tracking point (e.g. offset center of gravity tracking on board n1, leading edge detection scheme) - Parametric (physics-based) - Parametric of surface and we estimate them from the waveforms (e.g. MLE with Brown (1977) waveform shape offline for J1/n1). 9

10 Retracking: Off-center of gravity (OCOG) - Non-parametric (empirical methods) Based on the definition of a rectangle about the effective center of gravity of waveform, the amplitude and width, the OCOG retracking method uses full waveform samples to locate the half-power point as Where,P i is the sampled power. 10

11 Retracking: Threshold retracking - Non-parametric (empirical methods) Calculate the thermal noise: P N Compute the threshold level: T l =P N +p(a-p N ) The retracking location on the leading edge of the waveform is linearly interpolated between the bins adjacent to T l using Where,G k is the location of the first gate exceeding T l. (Davis, 1995) 11

12 Retracking: Improved Threshold retracking - Non-parametric (empirical methods) Sub-waves 1. sub-waveforms 2. compute the retracking range correction 3. determine the best one. (Hwang 2006) Gao at TUD 2 College of Environment and Resources, Fuzhou University, China 12

13 Retracking: Model of Beta-5 parameter function Parametric A 5-beta-parameter function is used to fit the single-ramp return waves, with the least squares method or the maximum likelihood method. Where,y(t) is the returned power, 13

14 Retracking Leading edge detection 14

15 Retracking Leading edge detection 15

16 Validation - Along-track altimetry Comparison/validation at TGs 16

17 Validation Along-track altimetry Comparison/validation at TGs 17

18 Validation Along-track altimetry Comparison/validation at TGs 18

19 Conclusions Distance to coast < 15 km - Usable data for 90% of all passes in MED for both satellites (sea land ) - Satellite : Envisat data perform better than TP Up to 15% more data (1Hz) (land sea) in last 5 km Better correlation with TG Re-tracking vrs not : Improvement by Re-Tracking Re-Tracked data not Re-Tracked data More data available / Better correlation with TG Waveform Re-Tracking Sea-land vrs land-sea OCOG less good, Improvement by retracking (Improved Threshold, Beta-5) 19

Tracking and Retracking

Tracking and Retracking W.H.F. Smith With contributions from P. Thibaut and P. Berry, L. Fenoglio-Marc Concurring: Birkett, Callahan, Dorandeu, Lambin, Martin, Raney, Rodriguez, Zanife CIOSS/NOAA Coastal