RADARSAT-2 Image Quality and Calibration...

www.mdacorporation.com

RADARSAT-2Image Quality and Calibration Update

by Dan Williams, Yiman Wang, Gordon Fitzgerald, Ron Caves, Marielle Chabot, Neil Gibb,

Yan Wu, Alan Thompson, Peter Allan, Cathy Vigneron

November 7, 2017

Use, duplication or disclosure of this document or any of the information or images contained herein is subject to the restrictions on the title page of this document. Copyright ©2017 MDA.

Restrictions, Credits and Disclaimer Language

2

RESTRICTION ON USE, PUBLICATION OR DISCLOSURE OF PROPRIETARY CONTENTThis presentation includes content that is proprietary to MDA Systems Ltd. (“MDA”), its subsidiaries, and third parties. Do not disclose, use, or duplicate this document or of any of its content.

MDA provides this presentation for general information purposes only, and this presentation does not constitute an offer, promise, warranty or guarantee of performance. MDA and its licensors do not authorize, and disclaim all liability for, any actions taken in reliance on this presentation.The products depicted are subject to change, and are not necessarily production representative. Actual results may vary depending on certain events or conditions.

COPYRIGHT © 2017 MDA Systems Ltd., and third parties whose content has been used by permission. All rights reserved.

RADARSAT-2 Data and Products © Maxar Technologies Ltd (2008-2017). All Rights Reserved. RADARSAT is an official mark of the Canadian Space Agency.

GENERAL ACKNOWLEDGEMENTSCertain images contained in this document are property of third parties:Images on P. 30, 31, 32 COPYRIGHT © Google and Copernicus Sentinel data for Sentinel data.Image on P.35 COPYRIGHT © Canadian Crown Copyright

Use, duplication or disclosure of this document or any of the information or images contained herein is subject to the restrictions on page 2 of this document. Copyright ©2017 MDA.

RADARSAT-2 Commercial SAR Modes

3

Ref: http://mdacorporation.com/docs/default-source/technical-documents/geospatial-services/52-1238_rs2_product_description.pdf


RADARSAT-2 Commercial SAR Modes

4

• C-Band

• 20 Beam Modes– 4 ScanSAR

(2 to 8 Beams)

– 15 Stripmap(Single Beam)

– 1 Spotlight (Steered Beam)

• Swath Widths – 20-500 km

• Nominal resolutions– Range: ~3 - 100 m

– Azimuth: ~0.8 – 100 m

• Commercial operations since April 24, 2008– Launched December 2007

Ref: http://mdacorporation.com/docs/default-source/technical-documents/geospatial-services/52-1238_rs2_product_description.pdf

Use, duplication or disclosure of this document or any of the information or images contained herein is subject to the restrictions on page 2 of this document.Copyright ©2017 MDA.

Outline• RADARSAT-2 Mission Status• RADARSAT-2 Image Quality Monitoring and Calibration

– Overview– Geolocation– Resolution– Radiometric Accuracy– Beam Pointing– Inter-Wing Phase Balance– Polarimetric Balance– Noise Levels– Mutual Interference with Other C-Band SAR Satellites– Other

• New beam / mode development• Conclusion

5


RADARSAT Portal

6

https://gsiportal.mdacorporation.com (can log in as guest)

https://gsiportal.mdacorporation.com/


RADARSAT-2 System Performance• Availability approaching 99%

– Over 600,000 successful acquisitions

• No degradation in performance – All 512 transmit/receive modules still

operating– Only 2 failed components, both no longer

used thanks to redundant units• Antenna column #12 (since 2012)• Antenna column heater #3 (since 2008)

– Product quality remains excellent

• Several initiatives underway to extend the mission life – Ground System upgrades– Spacecraft risk mitigation strategies– Image quality monitoring and calibration

7


Overview of Image Quality and Calibration

8

Ron Caves,

• Ongoing monitoring & calibration program– Acquisitions over calibration sites

– Analysis and trending of quality measures

– Calibration adjustments as warranted

• Ongoing work to enhance accuracy and minimize artifacts

• Creation of new modes – (e.g. Dual-Pol Extended Low, Maritime Surveillance ScanSAR, Extra-Fine,

Wide Quad Pol Modes)

Corner Reflector and Antenna Dish Point Target Measurements

(Resolution, sidelobe ratios, geolocation)

Amazon Rainforest Distributed Target Measurements

(Radiometric accuracy, beam patterns, beam pointing, polarimetriccorrection matrices, inter-wing phase balance)

8

Other

(Noise levels, antenna

verification, local oscillator

frequency, new mode

evaluation, issue investigations)


Point Target Monitoring• MDA owned, precision surveyed corner reflectors in Canada

– 2 in Vancouver, 2 in Quebec City

• Corner reflectors in other countries

– JPL (Rosamund, California)

– JAXA (Tomakomai, Japan)

– Simon Fraser University (Bolivia)

– CONAE (Argentina)*

– DTSO (Adelaide Australia)*

– U of Zurich (Dubendorf, Switzerland)*

– * past deployments

• Antenna dishes in Canada

– Gatineau, Prince Albert, Saskatoon, St-Hubert, Aldergrove, Masstown, Inuvik

– Higher radar cross-section than corner reflectors, but less accurate

• For geolocation, resolution and sidelobe measurement

• Results are filtered to eliminate non-representative measurements

– Dish not tracking sensor

– Contamination from surrounding clutter

– Snow on reflectors in winter

– Ground truth accuracy limitations

9


Geolocation Measurement Results(with Downlinked Orbit data, over Corner Reflectors)

Excellent accuracy at downlink thanks to precision orbit determination on-board spacecraft

10


Geolocation Measurement Results(with Definitive Orbit data, over Precision Corner Reflectors, with Atmospheric Correction,

since release of enhanced accuracy Definitive Orbit on 30-June-2015)

Notes:

- Atmospheric correction was done in post-processing as a slant range correction for the dry part of a standard atmosphere = 2.3m * (1/cos(incidence angle) – 1/cos(35o))

- Geolocation calibration is anchored at 35 incidence angle

- Precision surveyed corner reflectors: MDA and JPL

11


0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

10.0

0 10 20 30 40 50 60

Cir

cula

r Lo

cati

on

Err

or

(m)

Incidence Angle (deg)

Measured Circular Location Errors in Stripmap & Spotlight Modes

CE90

RMS

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

10.0

0 10 20 30 40 50 60

Cir

cula

r Lo

cati

on

Err

or

(m)


Measured Circular Location Errors in Stripmap & Spotlight Modes

CE90

RMS

Stripmap & Spotlight modes, Definitive Orbit, Precision Corner Reflectors,

with Atmospheric Correction, Results since Jan 1st 2016

Geolocation Measurement Statistics• Measured location errors with Downlinked orbit data:

– < 6 m RMS in most Single-Beam and Spotlight modes – <10 m RMS in Extended Low Incidence mode– <20 m RMS in ScanSAR modes

• Measured location errors with Definitive orbit data (available with ~1-2 days latency):– <2 m RMS in most Single-Beam and Spotlight modes after atmospheric correction in post-processing

• Better than the conservative accuracy values given in RADARSAT-2 Product Description documentation– Much better than original mission performance goals– Improved during mission thanks to orbit accuracy, calibration and SAR processing refinements

Stripmap & Spotlight modes, Downlinked Orbits, Corner Reflectors, Results since Jan 1, 2012

Note: Accurate geolocation requires accurate terrain height knowledge. These geolocation measurements are for calibrated corner reflectors of known elevation.

RMS = root mean square; CE90 = circular error, 90th

percentile

12


0.5

1

2

4

8

16

32

1/1/2010 1/1/2011 1/1/2012 1/1/2013 1/1/2014 1/1/2015 1/1/2016 1/1/2017 1/1/2018

Gro

un

d R

ange

Re

solu

tio

n

No

rmal

ize

d t

o 3

5o

Inci

de

nce

(m

)Wide

Standard (Far)

Extended High

Standard Quad (Far)

Standard (Near)

Extended Low

Standard Quad (Near)

Fine Quad

Fine

Multi-Fine

Wide Fine

Extra-Fine

Ultra-Fine

Spotlight

Ground Range Resolution (SGX Products)

13

Very stable (when normalized by incidence angle)Normalized resolution = resolution * sin(incidence angle) / sin(35o)


0.5

1

2

4

8

16

32

1/1/2010 1/1/2011 1/1/2012 1/1/2013 1/1/2014 1/1/2015 1/1/2016 1/1/2017 1/1/2018

Azi

mu

th R

eso

luti

on

(m

)Wide

Standard

Extended High

Extended Low

Standard Quad

Fine Quad

Fine

Multi-Fine

Wide Fine

Extra-Fine

Ultra-Fine

Spotlight

Azimuth Resolution (SGX Products)

14

Very stableNote: Wide mode (W1 beam) azimuth resolution was refined in 2014


-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

30 31 32 33 34 35 36 37

Elev

atio

n Pa

tter

n G

ain

(dB)

Elevation Angle (deg)

Measured Pattern Derivedfrom Amazon Scene

Reference Pattern fromCalibration Parameters

-15-14-13-12-11-10

-9-8-7-6-5

10 20 30 40 50 60

Gam

ma0

(d

B)


Amazon Reference Backscatter vs Incidence Angle

Co-Pol

Cross-Pol

Radiometric Accuracy Monitoring (1)

• The main method for monitoring absolute radiometry is through antenna elevation pattern analysis in Single Beam modes:– Measure backscatter profile as a function of range over a

homogeneous area of the Amazon rainforest

– Convert into a measured elevation pattern by subtracting noise, scaling by the assumed mean backscatter function of the Amazon, and backing out the elevation pattern correction applied during processing

– Align the measured pattern with the reference pattern from the calibration parameters, take the power ratio between the 2 patterns, and track it over time

15


Mean Radiometric Offset

16

Overall results (for all beams) generally stable since routine operations began in April 2008, with slight changes due to release of new beam modes, a subtle seasonal pattern, and a slight gradual decline overall

Mean per-scene differences from Amazon reference are typically within +/- 1 dB (std. dev. = 0.3 dB)

Overall mean difference is small (currently ~ -0.15 dB on Ascending passes, 0.0 dB on Descending passes)

Occasional calibration adjustments are made on a per-beam-mode basis to keep the results centered near zero


OSVN and DVWF Modes Calibration Update (Before)

• Ongoing calibration monitoring measurements showed that refinements to the Elevation Beam Patterns for the OSVN and DVWF modes were feasible

• The image to the left was an example in OSVN mode (Congo, Africa) – Shown at high contrast stretch (-10 to -4

dB gamma HH, -16 to -10 dB gamma HV)

Gam

ma0

(d

B)

-6

-8

-12

-14

HHHV

15


OSVN and DVWF Modes Calibration Update (After)

• Elevation beam pattern calibration update was released January 9, 2017– Applies to new OSVN and DVWF mode

products generated from data acquired since Sept 1st 2013

• The image to the left is the same Congo image after the updates – Shown at high contrast stretch (-10 to -4

dB gamma HH, -16 to -10 dB gamma HV)

Gam

ma0

(d

B)

-6

-8

-12

-14

HHHV

16


Beam Pointing Monitoring

• Elevation beam pointing is monitored using elevation pattern analysis over the Amazon in Single Beam modes

– The shifts needed to align the measured and reference elevation patterns are recorded and trended over time

• Azimuth beam pointing is monitored by comparing pitch and yaw measured on-board with Doppler centroid frequency estimated adaptively during SAR processing

– For image quality monitoring products over the Amazon

• This shows trends in how the beam pointing varies with respect to nominal

Actual beam centre

Nominal beam

centre

Elevation Pointing Error

AzimuthPointing Error

19


Elevation Beam Pointing Results

20

Overall results are stable since initial operations in April 2008

Mean measured pointing error overall ~= 0Typically within +/- 0.1o, std. dev. ~= 0.02o

The measurements contain a few outliers, mainly due to variations in scene content (Amazon scenes not perfectly uniform)


Azimuth Beam Pointing Results• Mean azimuth beam pointing from antenna boresight is centered close to zero• Follows seasonal trends, due to thermal effects

– Generally within +/- 100 Hz ~= 0.02o

– Compensated by adaptive Doppler centroid estimation in SAR processing and inter-wing phase calibration adjustments– Influenced by occasional refinements in solar array tracking and star tracker operations (summer 2013 and fall 2016)

21

(Sin

gle

Bea

m M

od

es)

Occasional problems with Left-looking slews (a few in fall 2015 and one in 2016), were addressed by refining star tracker operations


Inter-Wing Phase Balance Monitoring

• RADARSAT-2 can operate with dual receive apertures (wings)– Doubles the effective pulse repetition frequency

– Used in Multi-Fine, Extra-Fine, Ultra-Fine, Spotlight, and DVWF modes

• A phase difference between fore and aft wings causes an azimuth beam pointing shift (Doppler shift) from boresight

• The main impacts of uncompensated phase differences are subtle radiometric residual ripples in Spotlight images

– Not typically visible except in scenes of uniform ground cover

– Not a significant issue in other commercial modes

• We use Spotlight images over the Amazon to monitor and correct for this:

– Observed radiometric errors are used to estimate inter-wing phaseimbalances, which drive regular seasonal calibration adjustments

range

AftWing

Fore Wing

22


-10

-8

-6

-4

-2

0

2

4

6

8

10

12

14

1-J

an

-12

1-A

pr-

12

1-J

ul-1

2

1-O

ct-1

2

31

-De

c-1

2

1-A

pr-

13

2-J

ul-1

3

1-O

ct-1

3

31

-De

c-1

3

2-A

pr-

14

2-J

ul-1

4

1-O

ct-1

4

1-J

an

-15

2-A

pr-

15

2-J

ul-1

5

2-O

ct-1

5

1-J

an

-16

1-A

pr-

16

2-J

ul-1

6

1-O

ct-1

6

1-J

an

-17

2-A

pr-

17

2-J

ul-1

7

2-O

ct-1

7

1-J

an

-18

2-A

pr-

18

Inte

rwin

g P

hase D

iffe

ren

ce f

rom

C

urr

en

t E

qu

ilib

riu

m (

deg

)

Inter-wing Phase Calibration Summary Chart

H Receive Estimates

V Receive Estimates

H Receive Calibration

V Receive Calibration

1 year ago

2 years ago

3 years ago

4 years ago

5 years ago

Inter-Wing Phase Balance Results and Seasonal Calibration Adjustments

• Seasonal trends resemble those of the mean Doppler Centroid frequency delta from boresight• Changes are compensated through adaptive processing and calibration adjustments to maintain

optimal image quality– Calibration adjustments are made at discrete intervals in both H and V receive polarizations

23


Expected covariance matrix for Amazon rain forest

C=

𝑎 0 0 𝑑0 𝑏 𝑏 00 𝑏 𝑏 0𝑑 0 0 𝑎

• Polarization distortion is characterized over homogeneous regions of the Amazon rain forest:– Supports calibration over the entire swath– Known, stable, polarimetric signature:

• Reciprocity• Azimuth symmetry

– High signal to noise ratio– No ground infrastructure cost

• Input:– Quad-Pol (HH+HV+VH+VV) SLC products with HH absolute radiometric calibration

applied

• Procedures:– The homogoneous area of each product is identified (regions to exclude are selected

manually if necessary)– This area is partitioned into range sections, each spanning a fixed elevation angle (~0.2°)– An average 4 x 4 covariance matrix is calculated over each section, representing the

observed polarimetric signature– Imbalances on TX and RX and cross-talk are estimated from each covariance matrix– These are tracked and trended over time

24

Polarimetric Calibration Monitoring

Co-pol g0 backscatter: 𝑎 ~ − 6.5 dBX-pol g0 backscatter: 𝑏 ~ − 12.5 dBCo-pol g0 product: 𝑑 ~ − 9.5 dB

Excluded regions


Polarimetric Balance Results• Following corrections, balance accuracy for all beams is well within performance goals:

– TX imbalance: mean ~= 0, std. dev. = 0.1 dB, 1° phase, span = +/- 0.25 dB, +/- 2° phase– RX imbalance: mean ~= 0, std. dev. = 0.1 dB, 2° phase, span = +/- 0.25 dB, +/- 4° phase

25

Recent results show less outliers than in previous

years thanks to refinement of monitoring sites and

analysis methodsPhase of transmit imbalancecalibration was recently adjusted by 1o to correct for slight long-term drift in measured values

Small seasonal variation


-60

-50

-40

-30

-20

2006 2008 2010 2012 2014 2016 2018

Me

an S

qu

are

Rat

io (

dB

)

Year

C12/C11 C13/C11 C42/C11 C43/C11

-60

-50

-40

-30

-20

2006 2008 2010 2012 2014 2016 2018

Me

an S

qu

are

Rat

io (

dB

)

Year

C12/C11 C13/C11 C42/C11 C43/C11

Polarimetric Cross-Talk Results• Measured cross-talk levels remain excellent and well within performance

goals• < -30 dB before and < -50 dB after cross-talk correction

– Based on yearly averages of Amazon rainforest results over all beams

26

C11 = spatial averaged magnitude of the HH and HV complex scattering amplitudeC12 = spatial averaged Hermitian product of the HH and HV complex scattering amplitudesC13 = spatial averaged Hermitian product of the HH and VH complex scattering amplitudesC42 = spatial averaged Hermitian product of the VV and HV complex scattering amplitudesC43 = spatial averaged Hermitian product of the VV and VH complex scattering amplitudes

After CorrectionBefore Correction


Influence of Solar Beta Angle • Solar beta is the angle between solar illumination and the orbital plane• Its seasonal variation throughout the year is closely mirrored by some of

the quality measures we track– Inter-wing phase balance (addressed by calibration updates)– Azimuth beam pointing (compensated by adaptive Doppler centroid

estimation)– Polarimetric receive phase balance (impact on quality is very small, well

within performance goals)

• Spacecraft component temperatures show similar trends and are the likely mechanism

b

Solar Beta Angle (deg)Inter-Wing Phase

BalanceAzimuth Beam Pointing Variation from Boresight

Polarimetric Receive Phase Balance

Ho

tter

Co

ole

r

27


Noise Level Monitoring

• Data collected in “receive-only” modes

– Antenna receives but does not transmit

– Processed image contains only noise

– Acquisitions cover all pulse types used in commercial modes

• Processed image levels are analyzed and converted to estimates of noise levels in the raw data

• These estimates are compared to the calibrated noise level associated with each pulse

28


Noise Level Measurement ResultsResults are stable since initial operations in April 2008

Measurements in all polarizations remain within ~1 dB of calibrated levels

29

Vertical scale is equivalent to the sensor noise power in digital counts, for fully-focused beams, before front-end attenuations

(Influences noise equivalent sigma-naught, which varies locally within each product as reported in the product metadata)


Mutual Interference with Other C-Band SAR Satellites• Mutual interference occurs when pulses transmitted by antenna

on one spacecraft are received by antenna on the other spacecraft– Sentinel-1 operates at similar C-Band frequencies as RADARSAT-2– Main concern is the bistatic case where both satellites illuminate the

same area on the ground at the same time• Temporary rise in noise level

• MDA is taking a pro-active approach– Discussing with ESA approaches to mitigate the problem through orbit

control adjustments– Monitoring the known orbit crossing points where interference may

occur and issuing quality notices to users when necessary – Developing a software tool to predict future interference events with S-1

and other missions– Considering future missions in collaboration with Canadian Space Agency– Raising awareness

30


2017-01-27 over Amazon Rainforest • Three of nine RADARSAT-2 scenes (circled in yellow below) showed image artifacts deemed due to interference with simultaneous

Sentinel-1B EW mode acquisition (represented by the green rectangle)

• Satellite orbits are estimated to have crossed (i.e. were aligned along-track) around 09:56:30, a few seconds after the acquisitions ended

31


2017-01-27 over Amazon Rainforest • Sentinel-1B image also showed evidence of interference (in the 5th beam of EW mode)

• Note: Images are contrast enhanced in order to visualize the artifacts more clearly

32

Sentinel-1 image from Sentinel Data Hub


33

Sentinel-1 Interference Monitoring• C-Band interference events with significant impact on RSAT-2 image quality observed since beginning of

Sentinel-1 operations– < 20 events , << 0.1 % of acquisitions since August 2015, none since Feb 20th 2017

• Satellite orbit crossing points are currently located South of the equator– Coverage is less frequent in these zones so conflicts between missions are currently minimal– They continue to migrate south at a rate of 4-5 deg/month and then are expected to start heading North again

Sentinel-1A

Sentinel-1B


Other Aspects of RADARSAT-2 Image Quality Monitoring

Type of Monitoring Description

Point target sidelobe ratios Stable (within specifications where clean measurements are possible), <= -18 dB peak SLR, <= -14.9 dB integrated SLR

Azimuth pattern shape Stable since initial operations (correlations of measured patterns with reference patterns > 98%)

Ambiguity levels Stable (results of occasional spot checks in selected modes are consistent with models)

Chirp replica coefficients Stable

BAQ table usage Stable (good use of available dynamic range)

Payload local oscillator Stable (based on timing analysis of raw echo data)

Antenna verification Nominal (all T/R modules are operating well; results show minor seasonal magnitude and phase variations within design expectations)

Product Quality Control at Gatineau HQ

Each product is checked for major artifacts – any problems are reported and tracked

34


Pulse 1

Pulse 2

Pulse 3

Pulse 4

H

H

V

V

H

H

H

H

H

H

H

H

3.75 m3.75 m

New Experimental MODEX Modes• Moving Object Detection Experiment (MODEX) beam mode

– Developed for Defence Research and Development Canada (DRDC)• The research agency for Canada’s Department of National Defence

– ~60 non-commercial beam/mode combinations– Stripmap (Single Beam) and ScanSAR up to 150 km swath width– Dual-aperture operations on receive, and optionally on transmit– Successfully used by DRDC for Ground Moving Target Indication and

vessel motion estimation research and development

• Currently developing new dual-polarized MODEX beam/mode combinations– Transmit H and V polarizations alternately– Receive on fore and aft wings simultaneously– Single Beam and ScanSAR– For DRDC research in enhanced moving target detection (e.g. vessels at

steep incidence angles, ship-iceberg discrimination)

35

Image by Defence Research & Development Canada


Conclusion

• RADARSAT-2 image quality remains excellent– High system availability and stability

– Ongoing improvements continue through software updates and calibration refinement

– Occasionally affected by mutual radar interference with Sentinel-1

• Calibration adjustments are applied as needed to maintain operating objectives– As per measured seasonal fluctuations in inter-wing phase

– As per long-term gradual changes in other quality monitoring measures (e.g. radiometric levels, swath uniformity, polarimetric balance)

• Modes continue to be expanded according to client needs– Currently expanding the MODEX experimental beam mode for dual-pol moving object detection

36

Use, duplication or disclosure of this document or any of the information or images contained herein is subject to the restrictions on the title page of this document. Copyright ©2017 MDA.

THANK YOU!

37

Date post:	02-Jun-2020
Category:	Documents
Upload:	others
View:	19 times
Download:	0 times

RADARSAT-2 Image Quality and Calibration...

Documents