Ice Sheet Motion Speckle tracking. Velocity measurement techniques Day 1 Day 24 Feature Retracking 100 m Day 1 Day 24 10 m Day 1 Day 24 10 m InSAR Speckle.

Ice Sheet Motion

Speckle tracking

Velocity measurement techniques

Day 1

Day 24

Feature Retracking

100 mDay 1 Day 24

10 m

Day 1 Day 24

10 m

InSAR Speckle Retracking

Speckle Geometry

1

1

2

1

21

x

2/1222

)sin()cos(

))cos()sin()sin()(cos(

)90cos(

)cos(baseline parallel

chosen) slope negative (for the )sin(

)cos(

directionazimuth in the definitionby is Since

ntdisplacememotion ice )(

yzr

rr

r

r

rz

ry

ax

zyx

So

SrSrD

SrD

B

SrD

SrD

and

Range direction only

periodrepeatSpeed

So

Again

Sr

SaBp

Bpce

and

SrSa

This

SrSa

zyx

ary

ax

yxy

yxz

365

)(

)cos(*)sin(

)cos(**

and sin

)cos(**)sin(

nts.displaceme horizontal

the tocompared large is rateon accumulati thewhere

casesin invalid is and assumptionimportant an is

**

slope surface the toparallel flows ice that theAssume

2/1222

1

1r

Preprocessing

Doppler Overlap

(See if you can use the image data to compute doppler spectra)

Velocity Processing

Azimuth Corrections• Azimuth pixel spacing goes as prf as well as orbit height, terrain elevation, look angle

and latitude. Consequently two slightly different orbits will have slightly different azimuth spacings given approximately by

For a purely vertical separation, corresponding to about 14 m in height, the azimuthDisplacement will change about 3 parts in 10^6. There will be an accumulating Error such that after about 500 km, the azimuth will be incorrect by a measurable .3 pixels.

Comparisons

Coherence Streaks

An early incorrect idea

Ionospheric Streaks

• For C-band Radarsat, A SAR footprint is illuminated for about 0.5 sec. The satellite travels about 3.5 km.

• If the integrated electron density changes along the line of sight, while the footprint is illuminated, then the phase history built from about 600 pulses will be corrupted

• The consequence is incorrect azimuth placement.

• The effect is smaller in the range direction.

Streaks

• Idea tested by – Comparing streak density with ionospheric

activity– Simulations using a phase screen– Relationships to streak density and magnetic

poles

Option 1 – azimuth offset, range phase; azimuth offset, range offsets; ascending descending range;

Option 3 – only ascending descending range products.

Calibrating Phase Islands in the

Absence of Local Control

Phase and Range Offset

rr

rr

rr

rr

rr

r

r

Ln

L

RRL

Grouping

RRL

and

LRR

Where

rLRR

rLRR

R

For

4144

)4

(4

constants

)(4

So

rangein pixelnth theisr and size, pixel theisLr pixel,first the torange the tocorrespond R

speckleFor

retracking speckle with found biason registrati the

examplefor toequivalent biases geometric thecontainsIt scene. theacrossuniformly addedconstant some is

island, phase local ain phase unwrapped theis where

)(4

R

InSAR

'

'

02

010

02

010

01

02

0

2022

1011

0

012

Phase reconciliation

Information from Wrapped Phase

• Plug Vr into Vf and differentiate along flow direction

We assume that several terms are small: change in look direction is small; changes in flow directions are small for large glaciers over 0-30 km, Large scale strain rates not effected by surface slope.

Calculated Strain Rates