Analysis of the high frequency content of Jason-1, Topex and Envisat data

1
Analysis of the high frequency content of
Jason-1, Topex and Envisat data
Y. Faugere (CLS) A. Ollivier (CLS) P. Thibaut
(CLS)
2
Introduction

• Purpose of the study
• To compare the high frequency content of Jason-1,
  Topex and Envisat GDR products
• The analysed signal is SLAOrbit-range-MSS.
  This signal includes instrumental noise,
  processing noise, correction noise, residual
  geophysical signals
• The 1Hz and the 20Hz signals have been analysed
• Two types of method have been used to analyse
  such signals a spectral analysis, and a
  filtering technique
• Plan
• Method and data used
• Cross comparison of Jason-1 and Envisat HF
  content
• Improvement of the Envisat/Jason-1 consistency
  with MLE4
• Impact of mispointing and SWH selection on the
  high frequency content
• First results on the cross comparison of Jason-1
  and Topex RGDR HF content
• Conclusion

3
Method and data used

• First method spectral analysis of the SLA signal
• SLA(t) s(t) ?(t) , where s(t) is the
  geophysical signal and ?(t) is the noise

A plateau on a power spectrum can be the
signature of a white noise.
Plateau ?

• 1Hz spectra are computed from 10 days of data
• 20Hz spectra are computed from 2 days of data

4
Method and data used

• Second method Filtering technique

SLAOrbit-Range-MSS
High-pass filter (20km cut-off)
HF(SLA)
Standard deviation
sHF(SLA) in 2x2 boxes
5
Method and data used

• 3 Missions with several retrackings

• Jason-1
• MLE3 (Cycles 128-135) and MLE4 (Cycle 172, 20)
• Parameters are averaged from 20Hz estimations

• Envisat
• MLE3 (Cycle 39, 51)
• Parameters are averaged from 18Hz estimations

• Selection
• Calval editing
• Ocean data where
• Bathymetry lt -1000m

• Topex (Cycle 363)
• LSE
• range is averaged from 10Hz estimations
• Other parameters are estimated at 1Hz
• MAP
• Idem as LSE

6
Method and data used
Calval editing criteria for Jason-1
7
Cross comparison of Jason-1 and Envisat

• Improvement of the Envisat/Jason-1 consistency
  with MLE4 spectral analysis

Legend EN J1 GdrA J1 GdrB
1Hz Data

• The strange shape of 1Hz Jason-1 spectra in the
  0.1-0.4 Hz bandwidth disappears with the use of
  MLE4 retracking. 1Hz Envisat and Jason-1 spectra
  are superimposed.

8
Cross comparison of Jason-1 and Envisat

• Improvement of the Envisat/Jason-1 consistency
  with MLE4 spectral analysis

Legend EN J1 GdrA J1 GdrB
1Hz Data
20Hz Data

• The strange shape of 1Hz Jason-1 spectra in the
  0.1-0.4 Hz bandwidth disappears with the use of
  MLE4 retracking. 1Hz Envisat and Jason-1 spectra
  are superimposed
• At 20Hz the higher energy in the 0.1-0.4 Hz
  bandwidth is reduced with the use of MLE4
  retracking. Jason-1 HF energy is now lower than
  Envisat at all frequencies

9
Cross comparison of Jason-1 and Envisat

• Improvement of the Envisat/Jason-1 consistency
  with MLE4 Filtering technique

s²HF(EN) s²HF(J1 GDRa)
-2cm² 2cm²

• the higher Jason-1 energy in the wet areas (blue
  color) is the signature of the higher energy in
  the 0.1-0.4 Hz bandwidth seen on the 1 Hz spectra
  with J1 MLE3 retracking.

10
Cross comparison of Jason-1 and Envisat

• Improvement of the Envisat/Jason-1 consistency
  with MLE4 Filtering technique

s²HF(EN) s²HF(J1 GDRa)
s²HF(EN) s²HF(J1 GDRb)
-2cm² 2cm²
-2cm² 2cm²

• the higher Jason-1 energy in the wet areas (blue
  color) is the signature of the higher energy in
  the 0.1-0.4 Hz bandwidth seen on the 1 Hz spectra
  with J1 MLE3 retracking.
• The wet areas are no more visible when using MLE4
  j1 retracking

11
Cross comparison of Jason-1 and Envisat

• Impact of mispointing selection on the high
  frequency content

Envisat
Jason-1
Small Mispointing (X² - biaslt0.02deg2)
-0.15 0.15
-0.15 0.15
deg2
deg2
0.3 0.15 -0.15
Mispointing over 1 Jason-1 pass (deg2)
-20 -10
0 10
20
Latitude
12
In red small mispointing only
In black no selection
0.1 0 -0.1
Mispointing (deg2)
2 0
Water liquid content (Kg/m3)
-1 -2 -3
Uncorrected SLA (m)
13
Cross comparison of Jason-1 and Envisat

• Impact of mispointing selection on the high
  frequency content

Legend EN (all data) J1 (all data) Low
Mispointing EN Low Mispointing J1
1Hz Data

• 1Hz Envisat and Jason-1 are consistent in both
  cases. The HF energy is very sensitive to the
  mispointing selection

14
Cross comparison of Jason-1 and Envisat

• Impact of mispointing selection on the high
  frequency content

Legend EN (all data) J1 (alla data) Low
Mispointing EN Low Mispointing J1
1Hz Data

• 1Hz Envisat and Jason-1 are consistent in both
  cases. The HF energy is very sensitive to the
  mispointing selection
• At 20Hz the plateau is not impacted by the
  selection on both satellites. However the energy
  between 0.1-0.4Hz is strongly reduced when
  selecting low mispointing

15
Cross comparison of Jason-1 and Envisat

• Impact of SWH selection on the high frequency
  content

Envisat SWH histogram
Jason-1 SWH histogram
0
11
0
11
m
m
Small SWH (0.5ltSWHlt1.5m)
High SWH (3mltSWHlt11m)
Small SWH (0.5ltSWHlt1.5m)
High SWH (3mltSWHlt11m)
16
Cross comparison of Jason-1 and Envisat
Impact of SWH selection on HF content
Legend High SWH EN High SWH J1 Small SWH
EN Small SWH J1
1Hz Data

• 1Hz Envisat and Jason-1 are superimposed in both
  cases. The noise level increases with the wave
  height. Moreover a sort of pseudo-plateau is
  visible on 1Hz spectra at 1Hz only for high waves

17
Cross comparison of Jason-1 and Envisat
Impact of SWH selection on HF content
Legend High SWH EN High SWH J1 Small SWH
EN Small SWH J1
1Hz Data

• 1Hz Envisat and Jason-1 are superimposed in both
  cases. The noise level increases with the wave
  height. Moreover a sort of pseudo-plateau is
  visible on 1Hz spectra at 1Hz only for high waves
• At 20hz Envisat and Jason-1 spectra are closer
  for small waves (plateau almost superimposed).
• The pseudo-plateau visible at 1Hz on high waves
  is not the signature of a instrumental white
  noise. It is the signature of the energy between
  0.1-0.4Hz on the 20Hz spectra

18
First results on the cross comparison of Jason-1
and Topex RGDR HF content
Variance difference of HF content J1 MLE4 (cycle
20) - TP LSE (Cycle 360)
TP MAP TP LSE J1 MLE4
-1cm² 1cm²

• TP LSE and J1 MLE4 1Hz spectra are very
  consistent
• The Geographical distribution of the difference
  of HF content is not as homogeneous as for
  Jason-1/Envisat

19
Conclusion

• Cross comparison of Jason-1 and Envisat HF
  content
• The energy at 0.1-0.4Hz is reduced with the use
  of MLE4 on Jason- at 1Hz and 20Hz. GdrB enables
  then to Improve the Envisat/Jason-1 consistency
  at 1Hz and 20Hz.
• BUT
• the perfect superimposition of 1Hz Envisat and
  Jason-1 spectra is with the lower energy on
  Jason-1 20Hz spectrum. Why?
• There is a remaining suspicious high energy in
  the 0.1-0.4Hz bandwidth on 1Hz spectra
• At 20Hz, in the case of small waves selection,
  the energy is reduced at all frequencies whereas
  only the remaining energy in the bandwidth
  0.1-0.4Hz is reduced in the case of the low
  mispointing selection
• gt the remaining suspicious high energy in the
  0.1-0.4Hz bandwidth is due to perturbed data with
  high mispointing values
• gt The pseudo-plateau visible at 1Hz on high
  waves is not the signature of an instrumental
  white noise. It is linked to the energy between
  0.1-0.4Hz on the 20Hz spectra
• First results on the cross comparison of
  Jason-1and Topex RGDR HF content
• Good consistency between Topex LSE and Jason-1
  MLE4