Doppler slope correction outside ocean surfaces

Doppler slope correction
outside ocean surfaces
Fabien Blarel and Benoit Legresy
CTOH/LEGOS, Toulouse
QWG#18 – Propriano – May 2012
ICE Validation
●
●
The ice validation showed an
issue when using the Doppler
slope correction over the
cryosphere.
The RMS of the altitude is
bigger when this correction is
applied
QWG#18 – Propriano – May 2012
Doppler slope correction versus surface slope
Doppler slope correction versus along track slope
From the ENVISAT Handbook, the equation shows a linear relationship between surface slope and Doppler slope
correction ...
C0
δ h Doppler _ Slope ku=−
∗Surface _ height _ Rate
λ ku Chirp _ SL
The maps show the dependence between the slope along track and the Doppler slope correction with opposite sign. They look like
ok.
It appears from the plot that the linear relation is very noisy and there is an amount of data having Doppler slope correction equal to
●
zero despite the slope value.
The DEM used to calculated this correction in the GDR seems to be quite different to the altimeter observation and explain the
●
strong noise in this linear relationship.
The DEM is suspected to be the problem (probably in areas where it's incorrect or absent).
➔
●
QWG#18 – Propriano – May 2012
Doppler effect
The measurement includes 2 Doppler effects : The orbit rate and the surface height
rate (surface slope).
➔
Orbit rate
-----> Given from DORIS in GDR as instant_alt_rate parameter
Doppler correction
● Surface Slope
-----> Given from DEM
Doppler slope correction
Range rate
Orbit rate
Re a
l or b
from DORIS
it fo
l l ow
ed
range
H sat
Earth surface
geo
pote
nt i a
Surface Slope
l or
bit
ellip
soid
●
We would like improve this correction by using directly range rate to avoid error in
the DEM to enter the radar measurement
QWG#18 – Propriano – May 2012
Range : Doppler corrections
Here is an example of Doppler
correction track 834 over sea and
continental surface.
● δh
ku Doppler correction
from orbit rate in black
●Doppler correction from range
rate in red. The range rate is
smoothed over 5 km (~footprint
range of the radar echo).
●
Antarctica
Australia
Doppler _ Orbit
The Range contains the 2
Doppler effects : The orbit
rate and the surface height
rate (surface slope).
➔
QWG#18 – Propriano – May 2012
North America
Doppler slope correction
The Doppler slope
correction is :
●
Antarctica
Australia
In blue the Doppler
slope correction
●
δ h Doppler _ Slope ku=δ h Doppler _ range ku−δ h Doppler _ Orbit ku
QWG#18 – Propriano – May 2012
North America
In green the Doppler
slope correction from
GDR
●
●
QWG#18 – Propriano – May 2012
Doppler slope correction over Antarctica
In red the Doppler
slope correction
from the GDR
(from DEM).
●
In blue the
Doppler slope
correction
recalculated from
the ICE2 range
rate (smoothed on
5km).
●
It appears phase
differences
between GDR and
recalculated data.
➔
The radar does
not « see » the
same as the DEM.
There's may be an issue on how the slope / relocation is taken into account in the GDR correction.
➔
QWG#18 – Propriano – May 2012
Doppler slope correction
GDR (from DEM)
Recalculated from range rate
Very similar but... Quite different...
QWG#18 – Propriano – May 2012
Doppler slope correction versus along track slope
320 MHz
80 MHz
Crossover difference between
dopp_slp_ku recalculated and GDR.
In the sloppy areas, it appears that the
difference is strong.
In black the recalculated correction. It is more
linear as for the GDR and it is now defined in any
area.
Due to the bandwidth mode (320 MHz, 80 MHz
and 20 MHz) at the coast, the Doppler slope
correction is defined by different slope lines (yellow
lines).
It appears that the DEM and altimeter slope are
very different. The corrections are strongly different
where the DEM is not well defined. Especially in
coastal and mountain areas.
QWG#18 – Propriano – May 2012
Distribution of Doppler
slope correction anomaly
for validated crossover
data over Antarctica.
●
GDR (red):
● RMS= 3 mm
● Recalculated (black):
● RMS= 9 mm
●
The re-calculated data
have more extreme
values than in the GDR.
It is due to the strong
range variability,
especially in coast and
mountain areas. And it is
also due to the switching
bandwidth mode at the
coast.
The RMS of the re-calculated correction is bigger than GDR may be because the along
track slope computed from the DEM is smoother (or resolution?) than the re-calulated
correction (5km).
QWG#18 – Propriano – May 2012
Conclusion
Some improvements remain:
●
We have to asses the noise reduction by using
various linear smoothing (what is the right horizontal
scale to compute this rate 5 km, more, less ?).
But we can already see:
●
●
●
●
It appears that new Doppler slope correction does not
affect the RMS of crossover height difference. Like
almost all corrections. (need to do it after echo and
geo correction)
The range rate is a better estimator of the surface
slope “as seen by the altimeter” than probably any
DEM.
This new Doppler slope correction will be fully
validated for Cryosphere after this tunning
Other continental surfaces would easily get a reliable
doppler slope error correction using this method.
Range rate => valid everywhere when
the altimetric measurement is valid
➔
QWG#18 – Propriano – May 2012