WP 5000 Impact Assessment

hesitantdoubtfulΤεχνίτη Νοημοσύνη και Ρομποτική

29 Οκτ 2013 (πριν από 3 χρόνια και 8 μήνες)

70 εμφανίσεις

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

1

-



WP 5000


Impact
Assessment




S. Labroue, T. Moreau, P. Thibaut (CLS)

N. Picot, F. Boy (CNES)


CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

2

-


Round Robin
exercise


Collaborative
context

:


In the consortium :
iteration

loops

and
review

processes

will

be

set up


Outside

the consortium : open to
other

data set
evaluation

(contributions
from

NOAA, TAS, ENSEEIHT, UCL, …)


CLS
should

define

the
metrics

and the
methodology

for the
quality

assessment


With

the
retracking

experts (
special

sensitivity

in SAR
processing

which

are not
present

in LRM
processing
)


Taking

into

account

the WP2000 conclusions/
recommendations


=>
Metrics

and
methodology

has to
be

approved

by the consortium


Comparison

of one
retracking

taking

another

one as
reference


Importance to all
agree

on the
reference


Comparison

focused

on SLA and
sea

state (SWH
and sigma0 when available in RDSAR)


Detection

of
differences

between

the
algorithms


Work

on
metrics

able to
quantify

the relative
advantages
/drawbacks of
each

of
them
.


Feedbacks to the WP4000
retracking

experts


Possibly

recommend

one
algorithm
?

WP5000: Impact
Assessment

-

Methodology

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

3

-


Objectives of
this

assessment

for open
ocean

1.
Detect

correlated

errors

for
scales

beyond

150 km

2.
Confirm

that

the SAR
processing

allows

retrieving

smallest

spatial
scales

(20
-
70 km)
thanks

to 20 Hz noise
reduction

in
the
along

track

direction


The objective of WP5000
is

to compare the
algorithms

on
metrics

agreed

by
everyone
.


The
metrics

are not able to
detect

everything

=>
some

subtle

processing

differences

might

not
be

detectable

by the
proposed

metrics

(impact of a few mm
difficult

to
separate

from

the
oceanic

signal)


The multiplication of
different

metrics

is

important


WP5000: Impact
Assessment

-

Methodology

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

4

-


Each

processing

has been
validated

individually

by
each

responsible

of WP4000
task


The objective of the WP5000
is

to go
further

in the
geophysical

validation of
these

algorithms

for
each

sub
-
team
with

a focus on
retracking

methods

(
assessment

of the
geophysical

correction
is

not
foreseen
)


Product Validation Reports are
provided

for
each

algorithm

in WP4000


Data sets are
produced

by WP4000
contributors

on
agreed

areas and
periods


Analysis

performed

at

1 Hz and
at

20 Hz
depending

on the objective

=> Data sets
with

20 Hz
sampling

would

be

preferred

(compression
at

1 Hz
can

be

performed

by the WP5000)

WP5000: Impact
Assessment

-

Methodology

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

5

-


WP5000

Impact
Assessment

Round Robin
exercise



Overall

impact
assessment

report

WP5000: Impact
Assessment



Work

Plan



WP4400 data set



Data set user
manual



WP4000 Product validation
report



WP4000
ATBDs



CNES/CLS
database

(
other

EO satellite data and
geophysical

corrections)



CNES/CLS
database

(L2 CPP SAR/RDSAR)

I N P U T S



WP2000 recommandations

WP4000
contributors

are
consulted

to check and
agree

the outputs

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

6

-

WP5000: Impact
Assessment

-

Work

Plan


We

propose t he CPP/CNES
ret racki ng

as a
ref erence

f or t he LRM
because

we

have a large data set (
since

January

2011)
available

and
already

validated


We

recommend

t he use of L1 CPP dat a over SAR mode areas as i nput s of SAR
and RDSAR
ret racki ngs

1.
to
perform

direct
differences

between

collocated

SAR and RDSAR
measurements

2.
because

SAR
waveforms

are consistent in time
with

unchanged

IPF data
processing

(no
over
-
sampling

in range and
truncation
, and
no
azimuth

Hamming

weighting

implemented
)



RDSAR
retrackings

should

be

validated

compared

to
the CPP LRM
retracking


SAR
retrackings

should

be

validated

compared

to
RDSAR
retracking

to
be

chosen

(in the
most

intelligent
manner

to
evidence

the best of
each

algorithm
)


No SSB corrections
applied

to SAR and LRM
mode. This
could

potentially

explain

some

observed

discrepancies

between

SAR and LRM SLA

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

7

-


How to
validate

SAR
processing
?


Difficult

because

limited

areas and not a global
coverage

compared

to LRM mode


No
overlap

between

LRM and SAR zones


We

propose a 2
steps

approach


SAR
validated

versus
Reduced

SAR =
Relative Validation


The
easiest

one
since

the
reference

is

collocated


Analysis

of
Cryosat
-
2 data
alone


Reduced

SAR
validated

versus LRM =
Absolute

Validation

(3
means
)


Analysis

of
Cryosat
-
2 data


Cross calibration
with

J2


Analysis

of the LRM/RDSAR transition


WP5000: Impact
Assessment

-

Methodology





Absolute validation

Relative validation

LRM

RDSAR

SAR

RDSAR
-
LRM transition

SAR
-
LRM

transition

SAR
-
RDSAR


This 2
steps

approach

could

be

completed

by
absolute

validation
between

SAR and LRM (
without

using

the RDSAR
comparison
)


CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

8

-


Analysis

of
Cryosat
-
2 data


Maps

of
rejected

data


Spectral
analysis

of SLA, SWH


Analysis

of the
parameters

related

to the
retracking

(
histograms
,
maps
)


Analysis

of the
parameters

related

to the 20Hz =>1 Hz compression


Detection

of
dependencies



Focus on
scales

greater

than

150 km


Sensitivity

to radial
velocity
, SWH,
mispointing
,
etc



Performance
at

cross
-
overs

and SLA


Exploit the change of
geographical

mask

that

provides

a few
crossovers

between

LRM and SAR mode (
before

and
after

the 7
th

May 2012)


Analysis

of
parameters

wrt

to
coastal

distance


Sea

floor

mapping

over the Pacific
ocean

region

operating in SAR mode:


To find new structures (using data where we have never got any ones)


To evaluate the quality of the data (
Cryosat

SAR noise level and its qualities at different wavelengths)
by comparing SAR
retracked

data to mss or mean profiles.


WP5000: Impact
Assessment



Absolute

Validation

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

9

-


Cross calibration
with

J2


Qualitative by
comparing

2
along

track

profiles
by
selecting

parallel

tracks


Quantitative
with

statistics

at

crossovers



Analysis

of the LRM/RDSAR transition


Important to
insure

RDSAR validation


Histograms

of
parameters

in LRM mode
at

the transition location


Histograms

of
parameters

in SAR mode
at

the transition location


Analysis

of the
difference

RDSAR
-
LRM
at

the transition location

WP5000: Impact
Assessment



Absolute

Validation

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

10

-


Analysis

of
differences

between

2
retrackings


Some

of the
metrics

should

certainly

be

analysed

by the WP4000
validation reports


Maps

of
gained

or
rejected

data
compared

to the
reference

based

on
our

experience

in data
editing


Spectral
analysis

of
differences

of SLA, SWH (cancels out the
physical

signal)


Analysis

of the
parameters

differences

(
histograms
,
maps
)


Detection

of
dependencies

in the
difference



Focus on
scales

greater

than

150 km


Sensitivity

to radial
velocity
, SWH,
mispointing
,
etc





WP5000: Impact
Assessment



Relative Validation

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

11

-

WP5000: Impact
Assessment

-

Metrics


Problem

of multi
-
degrees

of
freedom
:


Data analyses
will

focus on the
dependencies

wrt

radial
velocity

and off
-
nadir
mispointing

angle of the satellite (
that

may

impact the
estimates
)





Need

to
separate

ascending

and
descending

tracks

that

are
related

to
different

radial
velocity

and off
-
nadir
angle values
at

the
same

point
location.


We

mi ght

have t o
consi der

al l SAR
mode areas
t o
cover

t he
l argest

range of val ues as possi bl e for
al l owi ng

robust

st at i st i cal

anal yses
and effi ci ent
assessment

of t he
i mpact of
each

new
product s


CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

12

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

13

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

14

-


Equatorial Pacific


Acquisition
since

7 May 2012 (and
reduced

area
since

01/10)


Brown
-
like

zone (few
rain
/blooms, SWH close
to 2 m,
low

oceanic

variability

stable in time)


Med
Sea


Well

known

region


But
with

calm

seas

(bloom
events
)


Coastal

region


North

Atlantic


Seasonal

variation (
with

bloom
events

in
summer

time)


High
waves

in
winter

time


Agulhas

Current


Zone
studied

in PISTACH => J2
high

resolution

data sets
available

(20Hz)
with

optimised

processing


High
waves

WP5000: Impact
Assessment



Zones &
Period

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

15

-

WP5000: Impact
Assessment



Zones &
Period


What

is

the
recommended

geographical

coverage
:
Global


to
separate

the
errors

linked

to radial
velocity
,
mispointing
, SWH


to
cover

most

of the range of radial
velocity
,
mispointing
, SWH values

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

16

-

WP5000: Impact
Assessment



Zones &
Period


What

is

the
recommended

time :


At

least 2
months

time
series

(1 cycle in
summer

and 1 cycle in
winter
) to
provide

significant

variation of SWH and
mispointing

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

17

-

WP5000: Impact
Assessment

-

Zones &
Period


To
assess

the quality of the sea level spectrum at all spatial scales, t
he

spectral
analysis

should

be

able:


To detect noise level @ high frequency


To
identify

correlated

errors

for
scales

between

10 and 80km


To check
consistency

of the
oceanic

signal @
high

wavelength

1 cycle over SAR
Pacific zone

1
day

over SAR
Pacific zone

3 cycles over SAR
Pacific zone

SAR CPP CS2

LRM J2

RDSAR CPP CS2

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

18

-


The
different

metrics

listed

will

help to
answer

to the 4
sub

themes

Themes

Metrics

Open

Ocean

Coastal

Ocean

Sea Floor Mapping

Polar Ocean

All

Distance

to coast

SLA from multi
-
tracks tightly spaced

All

WP5000: Impact
Assessment

-

Synthesis

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

19

-


Inputs


WP1000 user
requirements

synthesis


WP2000
synthesis



WP4000 validation reports



Deliverables


Validation report for
each

algorithm


Synthesis

report



Risks


Not
enough

data to
assess

each

algorithm

(open
ocean
)


No
clear

conclusion
coming

out
from

this

assessment


WP5000: Impact
Assessment

-

Synthesis

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

20

-

WP5000: Impact
Assessment



Tools


Comparisons

will

be

performed

in
terms

of:


Cartographies (to
visualize

geographyically

correlated

mean

error
)


Histograms


Spectral
analysis

(
allowing

to
identify

the
energy
/
error

levels

at

different

spatial
wavelengths
)


Time
series

analysis


Depedencies

analysis

(
correlations

between

parameters
)




CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

21

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

22

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

23

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

24

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

25

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

26

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

27

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

28

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

29

-

CP4O Progress meeting 1
-

Frascati

-

21
-
23 November 2012

-

30

-