Post on 23-Apr-2020
Enhancing Envisat ASAR WSM for oil slicks detection 1
Enhancing Envisat ASAR WSM segments to detect oil spills in the framework of the
“20 years of oil routes” project
Serge RIAZANOFF VisioTerra serge.riazanoff@visioterra.frKévin GROSS UPEMLV kgross01@etudiant.univ-mlv.fr
Enhancing Envisat ASAR WSM for oil slicks detection 2
Agenda
• Presentation of VisioTerra– Optical / Radar experience - Software development– Collaboration with ESA– The “20 years of oil routes” project
• The Envisat ASAR WSM product– Acquisition geometry– Normalized Radar Cross Section (NRCS)– Gain at sub-swath junctions– Homogenization of mean (m0) and standard deviation (σ0)
• C-Models– From Digital Number (DN) to sigma0– CMOD4, CMOD5, CMOD5.N– Retrieving V and φ
• Sub-swath junctions– Highlighting the defects– Piecewise adjustment
• Conclusions
Enhancing Envisat ASAR WSM for oil slicks detection 3
Presentation of VisioTerra
VT Satellite Tracker• Optical / Radar experience– Image processing (training)– Cartographic production– Viewing geometry– Orbit propagation
• Software development– Virtual globes– Stand-alone / servers
Enhancing Envisat ASAR WSM for oil slicks detection 4
Presentation of VisioTerra
• Collaboration with ESA– GAEL Consultant - Since 1994– Quality control of ESA products– Envisat MERIS– VTGoce
Enhancing Envisat ASAR WSM for oil slicks detection 5
Presentation of VisioTerra
• Diversity of data providers– Huge amount of data
Globcover (ESA)– More data providers– Free data policy– Interoperability
OGC standards– Consumers may
produce data
VT Escape
Enhancing Envisat ASAR WSM for oil slicks detection 6
The Envisat ASAR WSM product
• Acquisition geometry
S(t)sensor position
position du capteur
sensor velocityvitesse du capteur
azimuthazimut
swath fauchée
azimuth beamwidthlargeur du faisceau
haltitude
nadirtrace au sol
elevation beamwidthhauteur du faisceau
near rangeligne la plus
proche du nadir Boresight beamcentral axisaxe central du faisceau
rangedistance au capteur
pulse repetition period
far rangeligne la plus éloignée du nadir
RADAR ↔ RAdioDetection And Ranging
Enhancing Envisat ASAR WSM for oil slicks detection 7
The Envisat ASAR WSM product
• To normalize the mean (m0) and the standard deviation (σ0)
Enhancing Envisat ASAR WSM for oil slicks detection 8
The Envisat ASAR WSM product
• Normalized Radar Cross Section (NRCS)
– Mean mr(j)
– Standard deviation σr(j)
α(i,j)
satellite S
),(),(.),(
1)(1
01
0
jirjiji
jmM
iM
i
r ×= ∑∑
−
=−
=
δδ
( )221
01
0
)(),(),(.),(
1)( jmjirjiji
j r
M
iM
i
r −
×= ∑∑
−
=−
=
δδ
σ
Enhancing Envisat ASAR WSM for oil slicks detection 9
C-Models
• The backscattering coefficient (σ0) as a function of the wind speed (V) and azimuth (φ)
azimuth axis (trajectory)
V3
φ3
V2 φ1=90°
V1 φ1=0°
a
b
Enhancing Envisat ASAR WSM for oil slicks detection 10
C-Models
• From Digital Number (DN) to sigma0 (M(j))
• CMOD4
( ) ))(sin()()()(2
jK
jmjbjM rlin θ×== Moyenne, écart-type des CN et Sigma0 par colonne
0
1000
2000
3000
4000
5000
6000
7000
0100 200 300 400 500
600 700 800900
1000
1100
1200
1300
1400
1500
1600170
0180
01900
2000
2100
2200230
0240
0250
0260
0270
0280
0290
0300
0310
0320
03300
3400
3500
3600370
0380
03900
4000
4100
4200
4300
4400
4500
4600
4700
4800
4900
5000
5100
5200
5300540
0550
05600
Index de colonne
Moyenne, écart-type des CN
0
0,2
0,4
0,6
0,8
1
1,2
1,4
1,6
Sigma0
CNMean CNStdev Sigma0
( ) 6.123104 2costanhcos1),,( φφθφ bbbbVMCMOD ++=&
)(.0
110 βγα ++= VfRbb
>≤<
≤−= −
−
5,2.3/510,)log(
10,1010
10
1
ssss
sf
231201 PcPcPc ++=α
261504 PcPcPc ++=γ
291807 PcPcPc ++=β
( ) )(. 213012110101 xfVcPcVcPcb +++=...
Enhancing Envisat ASAR WSM for oil slicks detection 11
C-Models
• CMOD5
• CMOD5.N
( ) 6.12105 2coscos1),,( φφθφ bbbVM CMOD ++=&
25/)40( −= θx
( )γ020 ,.10 10 sVafb Vaa +=
( )
≥<
=0
0000 ,)(
,)(/),(sssgsssgssssf
α
sesg
−+=
11)(
[ ])(34.0
171615141 181
)].(4tanh[5.0.)1(cVe
VccxxVcxcb −+
++−+−+=
[ ])(1. 00 sgx −=α
34
23210 xcxcxcca +++=
xcca 651 +=
xcca 872 +=2
11109 xcxcc ++=γxccs 13120 +=
...
Enhancing Envisat ASAR WSM for oil slicks detection 12
C-Models
• Best fit with one of the 3 models
))(,,(),()( mod jVMVBjM θφφ &×≈
[ ]∑∑
−
=−
=
×−××=1
0
2mod1
0
))(,,()()()(
1),,,,(N
jN
j
k jVMBjMjj
VBMSE θφδδ
φ &
2mod
1
0
mod
1
0
))(,,()(
))(,,()()(),(
jVMj
jVMjMjVB N
j
N
j
θφδ
θφδφ
&
&
∑
∑−
=
−
=
×
××=⇒
[ ] [ ][ ] [ ]2modmod
1
0
21
0
1
0modmod
),())(,,()()()(
),())(,,()()(),(
φθφδδ
φθφδφ
VMjVMjMjMj
VMjVMMjMjVr
N
j
N
j
N
j
&&
&&
−××−×
−×−×=
∑∑
∑−
=
−
=
−
=
Enhancing Envisat ASAR WSM for oil slicks detection 13
C-Models
• Best fit results
Enhancing Envisat ASAR WSM for oil slicks detection 14
C-Models
• Applying a model-based equalization
raw model-based observation-based
Enhancing Envisat ASAR WSM for oil slicks detection 15
Sub-swath junctions
• Observed defect
Enhancing Envisat ASAR WSM for oil slicks detection 16
Sub-swath junctions
• After C-MOD5 centering-reduction
• After differentiation
• After Gaussian convolution
Enhancing Envisat ASAR WSM for oil slicks detection 17
Sub-swath junctions
• Semi-linear model
+≤<′′
+−′′
+−
×′′−′
+
′′≤<′′′+′
′≤≤′
×′′−′
−
=
22
22
)()()(
2)()(
02
)()()(
)(
211
211
21
11
111
11
11
1
jjjjsijjj
jjjjejeje
jjjsijeje
jjsijjjejeje
jc
Enhancing Envisat ASAR WSM for oil slicks detection 18
Sub-swath junctions
• Inter-swath correction
Enhancing Envisat ASAR WSM for oil slicks detection 19
Sub-swath junctions
• Applying a model-based equalization
raw model-based observation-based
Enhancing Envisat ASAR WSM for oil slicks detection 20
Sub-swath junctions
• Applying a model-based equalization and inter-swath correction
raw model-based observation-based
Enhancing Envisat ASAR WSM for oil slicks detection 21
Conclusions
• To help the photo-interpretation– 20050714_110323
in Google Earth
• To make easier theoil slicks classification
Enhancing Envisat ASAR WSM for oil slicks detection 22
Thank you for your attention
Questions ?
Serge RIAZANOFF VisioTerra serge.riazanoff@visioterra.frKévin GROSS UPEMLV kgross01@etudiant.univ-mlv.fr