SWARP KO Bergen, 4 Feb. 2014 WP4 : Satellite remote sensing of wave in ice .

SWARP KO Bergen, 4 Feb. 2014

WP4 : Satellite remote sensing of wave in ice

ww

w.p

hoto

from

thew

orld

.com


OceanDataLab (ODL)

• IFREMER spin-off incorporated in Brest, april 2013.• Principal objective: Develop tools for multimodal

synergy analysis (multisensors, models, in-situ)• 4 persons : 3 research engineers and 1 software

engineer.• Role in SWARP : provide SAR wave spectra retrieval in

the marginal ice zone and participate to the validation effort.


WP4 OutlineObjective

To develop and implement remote sensing methods for observation and model validation of waves in the MIZ.

Tasks

1. Ice type recognition from coarse resolution scatterometry (Ifremer)2. Ice type recognition from high resolution SAR and optical images (NIERSC) 3. Waves-in-ice retrieval methodology review and implementation (ODL)4. Acquisition and analysis of collocated SAR, optical and CryoSat altimeter

data (NERSC) 5. Analysis of observed waves-in-ice evolution relative to sea ice type (ODL)

Task 4.1 : Ice type recognition from coarse resolution scatterometry

Fanny GIRARD ARDHUIN, IFREMER

Sea ice roughness from scatterometer sensors

Canada

18-02-2007

Grid. resolution : 12.5 km, 25 km

Daily (weekly for ERS)

Since 1991

QuikSCAT

for both Arctic & Antarctic areas

ERS-1&2 : 1991-2001

NSCAT : 1996-1997

ASCAT : 2007-present

QuikSCAT : 1999-2009

Ifremer/CERSAT unique time series

Low res. scatterometer data :

daily maps for both pole areas

Robinson (2004)

Multi year ice (MY)First year ice (FY)

18-02-2007QuikSCAT

FY MY

Roughness is linked with sea ice age

MY/FY ice can be detected

Ice type from scatterometer

OCT MAYMARDEC

Example of backscatter time series during a winter

MY extent time series

Backscatter values over sea ice depend on frequencies, polarisation, incidence angle but also on ice type, salinity in the ice, etc...

QuikSCAT method to adapt to ASCAT data for recent period (since 2009) for SWARP project and to validate → need SAR ice type detection

Example of MY area time series with QuikSCAT data (1999-2009) (consistent with Kwok's results)

Swan & Long, 2009

Example of QuikSCAT backs. time series and ice type classification

FY and MY areas can be quantified applying a moving back. threshold value


Task 4.2 Sea ice classification from SAR data

Nansen International Environmental and Remote Sensing Centre, St.Petersburg, Russia

Vladimir Volkov

Natalia Zakhvatkina


Objective

To develop sea ice classification algorithm using high-resolution SAR images in order to classify the MIZ in selected test areas

To map ice edge and the details of the ice cover like ice types, open water and various stages of new and first-year ice, leads, polynyas, and others;

Implement the developed technique for the determination of the zone of broken-up floes in the MIZ, which will be used to validate the floe size distribution given by sea ice models


Data

• I. Envisat’s ASAR (Advanced Synthetic Aperture Radar)– ASAR operated in the C band in 5 modes; we worked mostly with

Wide Swath mode images of 405 km swath and 150 m resolution.– ESA announced the end of Envisat's mission on 9 May 2012.

• II. Radarsat-2 SAR– multiple modes of operation, – HH, HV, VV and VH polarized data can be acquired, – its highest resolution is 3 m in Very High Resolution mode,– we work with data of ScanSAR Wide Beam mode that has a

nominal swath width of 500 km and an imaging resolution of 100 m.

• III. Optical data ?????????????


Radarsat-2 ScanSAR Wide mode images• We use RADARSAT-2 data received in ScanSAR Wide (SCW) mode at HH

(horizontally transmitted and horizontally received) and HV (horizontally transmitted, vertically received) polarizations. This mode assembles wide SAR image from several narrower SAR beams, resulting to an image of 500 × 500 km with 100 m resolution.

HH HV

Fram Strait, 20/02/2012

OWr

OW

Ice OWr

OW

Ice

Ice and water pixels can be separated

OW roughOW calmSea Ice


Sea ice classification using SAR images

Two RADARSAT-2 SAR ScanSAR Wide images: HH and HV polarizations

SAR images calibration, angular dependence correction

Noise correction of HV dual-polarization SAR image

Image features calculation: mean backscatter, texture characteristics

Image classification using Support Vector Machines technique

Sea ice charts


Noise reduction in HV polarization• The effect is reduced by

subtracting the noise floor level from the HV image values.

• Left image - raw HV polarization image,

• right image – noise reduced image.

• Blue curve shows the sigma0 value profile of the raw HV channel image over the horizontal line, the red curve depicts the noise floor level and the green curve is the result of subtraction


Support Vector Machines algorithm TEACHING

(klusterization)


Support Vector Machines CLASSIFICATION


Automated classification of Radarsat-2 data

• The described technique has been used in the development of automated sea ice / water classification.


18 Jan 2014


Vilkitskiy Stright, 23, 26 Aug 2013

HH HV SVM


28 Aug 2013


Task 4.3 Waves-in-ice retrieval methodology review and

implementation

Fabrice Collard (OceanDataLab)


Wave modulation modification

Incident swell

MIZ



Incident swell

MIZ



Incident swell

MIZ


MIZOPEN OCEAN

ASAR data © ESA 2011



MIZMIZ



MIZMIZ


Work to be done

• Update Modulation transfer functions to cope with ice roughness and dynamical properties

• Validate retrieved SAR wave spectra using wave buoys in the MIZ.


Task 4.4 Acquisition and analysis of collocated SAR, optical and

CryoSat altimeter data (NERSC)

SMOS ice Thickness2013 02 19


Task 4.5 Analysis of observed waves-in-ice evolution relative to

sea ice type (ODL)


Attenuation analysisWARNING !!! : Uncalibrated retrieved wave height in the MIZ

MIZ

OPEN OCEAN

SWARP KO Bergen, 4 Feb. 2014 WP4 : Satellite remote sensing of wave in ice .

Documents

sea ice agemyfy ice

ice types

ice cover

sea ice roughness

ice type classification

tasks ice type recognition

marginal ice zone

ice evolution relative