Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height...

32
Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal Sarabandi

Transcript of Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height...

Page 1: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Quantifying the Effects of Wind on polarimetric SAR amp InSAR

Tree height estimation

Michael L BensonDr Leland E Pierce

Prof Kamal Sarabandi

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Motivationbull InSAR is often used over forested areas to obtain information

on forest structuresbull Repeat-pass InSAR over forests suffers from poor coherence

due to changes in a forests physical attributes (exact position motion of branches leaves etc) and moisture which affects the dielectric constant of the scatterers

bull Is there a way to obtain high-coherence repeat-pass InSAR data using models of these effects

bull Could we then produce reliable high-quality forest structure estimates including canopy height

bull This research presents a detailed model of the effects of wind on SAR image formation coherence and the associated effect on tree height retrieval through the scattering phase center

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

bull Use our existing Lindenmayer system based fractal tree generator [Lin + Sarabandi IGRS 1999]

bull Trees defined by a DNA file

bullConsists of basic parameters such as leaf radius leaf thickness and maximum branch angle

bullDifferent DNA for each species

bullDNA is iterated a set number of times to form a complex semi-random tree realization

bullDNA sub-string re-writing rules are used to generate realistic branching structures with needles leaves

bull Current study uses a deciduous red maple stand only

Tree Generation

Radiation Laboratory

bull Generate both coniferous and deciduous trees including

bull Red Maple

bull Red Oak

bull Red Pine

bull Sugar Maple

bull White Ash

bull White Pine

bull White Spruce

bull DNA files only specify tree structure InSAR parameters are specified elsewhere

Tree Generation (2)

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation LaboratoryModelling Wind

Branchesbull Need to know

bull Mechanical Parameters

bull Length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Parent amp Children

bull Youngrsquos Modulus E

bull Need to Find

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation LaboratoryModelling Wind

Stems amp Leavesbull Need to know

bull Mechanical Parameters

bull Stem length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Family (branch stem leaf)

bull Youngrsquos Modulus E

bull Leaf thickness t

bull Need to Find for each stem-leaf pair

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 2: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Motivationbull InSAR is often used over forested areas to obtain information

on forest structuresbull Repeat-pass InSAR over forests suffers from poor coherence

due to changes in a forests physical attributes (exact position motion of branches leaves etc) and moisture which affects the dielectric constant of the scatterers

bull Is there a way to obtain high-coherence repeat-pass InSAR data using models of these effects

bull Could we then produce reliable high-quality forest structure estimates including canopy height

bull This research presents a detailed model of the effects of wind on SAR image formation coherence and the associated effect on tree height retrieval through the scattering phase center

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

bull Use our existing Lindenmayer system based fractal tree generator [Lin + Sarabandi IGRS 1999]

bull Trees defined by a DNA file

bullConsists of basic parameters such as leaf radius leaf thickness and maximum branch angle

bullDifferent DNA for each species

bullDNA is iterated a set number of times to form a complex semi-random tree realization

bullDNA sub-string re-writing rules are used to generate realistic branching structures with needles leaves

bull Current study uses a deciduous red maple stand only

Tree Generation

Radiation Laboratory

bull Generate both coniferous and deciduous trees including

bull Red Maple

bull Red Oak

bull Red Pine

bull Sugar Maple

bull White Ash

bull White Pine

bull White Spruce

bull DNA files only specify tree structure InSAR parameters are specified elsewhere

Tree Generation (2)

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation LaboratoryModelling Wind

Branchesbull Need to know

bull Mechanical Parameters

bull Length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Parent amp Children

bull Youngrsquos Modulus E

bull Need to Find

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation LaboratoryModelling Wind

Stems amp Leavesbull Need to know

bull Mechanical Parameters

bull Stem length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Family (branch stem leaf)

bull Youngrsquos Modulus E

bull Leaf thickness t

bull Need to Find for each stem-leaf pair

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 3: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Motivationbull InSAR is often used over forested areas to obtain information

on forest structuresbull Repeat-pass InSAR over forests suffers from poor coherence

due to changes in a forests physical attributes (exact position motion of branches leaves etc) and moisture which affects the dielectric constant of the scatterers

bull Is there a way to obtain high-coherence repeat-pass InSAR data using models of these effects

bull Could we then produce reliable high-quality forest structure estimates including canopy height

bull This research presents a detailed model of the effects of wind on SAR image formation coherence and the associated effect on tree height retrieval through the scattering phase center

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

bull Use our existing Lindenmayer system based fractal tree generator [Lin + Sarabandi IGRS 1999]

bull Trees defined by a DNA file

bullConsists of basic parameters such as leaf radius leaf thickness and maximum branch angle

bullDifferent DNA for each species

bullDNA is iterated a set number of times to form a complex semi-random tree realization

bullDNA sub-string re-writing rules are used to generate realistic branching structures with needles leaves

bull Current study uses a deciduous red maple stand only

Tree Generation

Radiation Laboratory

bull Generate both coniferous and deciduous trees including

bull Red Maple

bull Red Oak

bull Red Pine

bull Sugar Maple

bull White Ash

bull White Pine

bull White Spruce

bull DNA files only specify tree structure InSAR parameters are specified elsewhere

Tree Generation (2)

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation LaboratoryModelling Wind

Branchesbull Need to know

bull Mechanical Parameters

bull Length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Parent amp Children

bull Youngrsquos Modulus E

bull Need to Find

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation LaboratoryModelling Wind

Stems amp Leavesbull Need to know

bull Mechanical Parameters

bull Stem length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Family (branch stem leaf)

bull Youngrsquos Modulus E

bull Leaf thickness t

bull Need to Find for each stem-leaf pair

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 4: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

bull Use our existing Lindenmayer system based fractal tree generator [Lin + Sarabandi IGRS 1999]

bull Trees defined by a DNA file

bullConsists of basic parameters such as leaf radius leaf thickness and maximum branch angle

bullDifferent DNA for each species

bullDNA is iterated a set number of times to form a complex semi-random tree realization

bullDNA sub-string re-writing rules are used to generate realistic branching structures with needles leaves

bull Current study uses a deciduous red maple stand only

Tree Generation

Radiation Laboratory

bull Generate both coniferous and deciduous trees including

bull Red Maple

bull Red Oak

bull Red Pine

bull Sugar Maple

bull White Ash

bull White Pine

bull White Spruce

bull DNA files only specify tree structure InSAR parameters are specified elsewhere

Tree Generation (2)

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation LaboratoryModelling Wind

Branchesbull Need to know

bull Mechanical Parameters

bull Length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Parent amp Children

bull Youngrsquos Modulus E

bull Need to Find

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation LaboratoryModelling Wind

Stems amp Leavesbull Need to know

bull Mechanical Parameters

bull Stem length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Family (branch stem leaf)

bull Youngrsquos Modulus E

bull Leaf thickness t

bull Need to Find for each stem-leaf pair

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 5: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

bull Use our existing Lindenmayer system based fractal tree generator [Lin + Sarabandi IGRS 1999]

bull Trees defined by a DNA file

bullConsists of basic parameters such as leaf radius leaf thickness and maximum branch angle

bullDifferent DNA for each species

bullDNA is iterated a set number of times to form a complex semi-random tree realization

bullDNA sub-string re-writing rules are used to generate realistic branching structures with needles leaves

bull Current study uses a deciduous red maple stand only

Tree Generation

Radiation Laboratory

bull Generate both coniferous and deciduous trees including

bull Red Maple

bull Red Oak

bull Red Pine

bull Sugar Maple

bull White Ash

bull White Pine

bull White Spruce

bull DNA files only specify tree structure InSAR parameters are specified elsewhere

Tree Generation (2)

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation LaboratoryModelling Wind

Branchesbull Need to know

bull Mechanical Parameters

bull Length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Parent amp Children

bull Youngrsquos Modulus E

bull Need to Find

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation LaboratoryModelling Wind

Stems amp Leavesbull Need to know

bull Mechanical Parameters

bull Stem length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Family (branch stem leaf)

bull Youngrsquos Modulus E

bull Leaf thickness t

bull Need to Find for each stem-leaf pair

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 6: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

bull Generate both coniferous and deciduous trees including

bull Red Maple

bull Red Oak

bull Red Pine

bull Sugar Maple

bull White Ash

bull White Pine

bull White Spruce

bull DNA files only specify tree structure InSAR parameters are specified elsewhere

Tree Generation (2)

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation LaboratoryModelling Wind

Branchesbull Need to know

bull Mechanical Parameters

bull Length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Parent amp Children

bull Youngrsquos Modulus E

bull Need to Find

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation LaboratoryModelling Wind

Stems amp Leavesbull Need to know

bull Mechanical Parameters

bull Stem length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Family (branch stem leaf)

bull Youngrsquos Modulus E

bull Leaf thickness t

bull Need to Find for each stem-leaf pair

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 7: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation LaboratoryModelling Wind

Branchesbull Need to know

bull Mechanical Parameters

bull Length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Parent amp Children

bull Youngrsquos Modulus E

bull Need to Find

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation LaboratoryModelling Wind

Stems amp Leavesbull Need to know

bull Mechanical Parameters

bull Stem length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Family (branch stem leaf)

bull Youngrsquos Modulus E

bull Leaf thickness t

bull Need to Find for each stem-leaf pair

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 8: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation LaboratoryModelling Wind

Branchesbull Need to know

bull Mechanical Parameters

bull Length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Parent amp Children

bull Youngrsquos Modulus E

bull Need to Find

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation LaboratoryModelling Wind

Stems amp Leavesbull Need to know

bull Mechanical Parameters

bull Stem length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Family (branch stem leaf)

bull Youngrsquos Modulus E

bull Leaf thickness t

bull Need to Find for each stem-leaf pair

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 9: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation LaboratoryModelling Wind

Stems amp Leavesbull Need to know

bull Mechanical Parameters

bull Stem length l

bull Center (x y z)

bull Orientation (ΘΦ)

bull Family (branch stem leaf)

bull Youngrsquos Modulus E

bull Leaf thickness t

bull Need to Find for each stem-leaf pair

bull Moment of Inertia I

bull Resonant Frequency fr

bull Deflection angle α

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 10: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Modelling Winda single branch

m

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 11: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Modelling Windmultiple branches

W

Θ = 80deg

Θ = 45deg

Θ = 90deg

z

y

x

WaN = 0889

WaN = 05

WaN = 10

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 12: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Modeling Windbull A steady wind force on the branch causes a vibration with

frequency bull Moment of Inertia I is determined by the mass distribution in

the branch as well as the mass of branches attached to its end

bull Youngs Modulus E is a measure of the stiffness of the

branch measurements of E for different species are availablebull Tree motion composes each branch motion using movement

of lower branches to alter locations of upper branches

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 13: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Modelling Wind Branch Motionbull A steady wind force on the branch causes a vibration with

frequency

bull How far should each branch movebull Depends on the wind velocity and the branchrsquos physical

parametersbull Assuming T = 25 C and average moisture content in each

branch calculate the maximum deflection for each branch as

bull Under SHM approximation branch will be directly moved along the direction of the wind field a maximum of frac12 max in any direction However branches may move more than frac12 max relative to their original (rest) position as a result of their parentsrsquo motion

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 14: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Modelling Wind Branch Motion (2)bull Where does Φmax come from

bull Pressure due to wind is [httpwwwvent-axiacomknowledgehandbooksection1windflowasp ]

bull Branch surface area (SA) presented to wind is ~

bull Maximum deflection for a cantilever is defined as [httpdarkwinguoregonedu~structcourseware461461_lectures461_lecture40461_lecture40html ]

bull Use of simple trigonometric relationships yield Φmax as

ymax

Φmax

L L

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 15: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation LaboratoryModelling Wind Constant

Breeze Movie

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 16: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 17: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

InSAR Simulator (1)bull Divide stand into several horizontal slabsbull In each slab estimate the mean field using Foldys

approximation [Lin + Sarabandi IGARSS 1996]bull Also estimate attenuation through each slabbull For each branch or leaf in the tree calculate the backscattered

field as Escat = Einc Seiɸ where S is the scattering matrix of the object and ɸ is the relative phase of the scattering due to the relative position of this object in the tree

bull The scattering matrix is estimated using four (4) scattering mechanismsndash Direct Scattering St ndash Ground-Object scattering Sgt

ndash Object-Ground scattering Stg

ndash Ground-Object-Ground Sgtg

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 18: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

InSAR Simulator (2)bull Use Δk Approach [Sarabandi TGRS 1997]bull Approximate InSAR baseline with a small change in frequency

bull Measured phase can be calculated as

bull For each polarization calculate a scattering phase center as

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 19: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 20: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Combined InSAR and Windbull Produced 5 instances of a single red maple without the influence of

wind and placed them in a 625 m2 regionbull Applied wind to these trees and saved all geometries for each time

step in the simulationndash Δt = 002s total time = 1s

bull Use InSAR simulator to produce a single-look complex (SLC) image of a one-pixel forest stand for each geometry at each time step This includes the no-wind case

bull Now can produce a coherence estimate between pairs of SLC images

Where u1 is the no-wind tree and u2 is one sample from a wind-blown tree sequence

bull Can plot this as a function of time

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 21: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 22: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Forming a SAR Imagebull A SAR image formed with multiple looks in practice will be

collected over a period of time often under 1 secondbull Wind Speeds below vary from a strong breeze to a sustained

hurricane force wind

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 23: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Overview

bull Motivationbull Tree Modelbull Wind Modelbull InSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Future Work

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 24: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Coherence and Height Estimationbull This Coherence measurement can also be thought of as a

measure of similarity between two SAR images taken at different times The only difference between the two images is a wind induced motion

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 25: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Coherence and Height Estimationbull At L-band the principal contributor to the VV-polarization from

the target will be the tree trunks As the wind height increases the mean scattering phase center height decreases at a nearly consistent rate

mean SPCbull Strong Breeze 39777mbull Stronger Breeze 35758mbull Storm Gust 29520m

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 26: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Coherence and Height Estimationbull The physical description of the windrsquos effect on higher frequency

(C-band) tree height estimate is not as straight forward as the L-band VV case

bull We have observed that the co-polarized mean SPC increases with an increasing wind force while the mean cross-polarized SPC decreases

Mean Scattering Phase Center Height [m]

Strong Breeze Storm Gust

VV C-Band 65587 69315

HH C-Band 27823 28971

VH C-Band 87781 72317

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 27: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Overview

bull Introductionbull Tree Modelbull Wind Modelbull ifSAR Simulatorbull SAR Image Coherencebull InSAR Height Estimatebull Conclusions and Future Work

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 28: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Conclusions and Future Workbull Developed a realistic wind model for trees including branches stems

leaves and needles and demonstrated it on a red maple stand

bull Applied SAR and InSAR model to the stand with and without wind

bull Calculated coherence between wind no-wind cases to simulate repeat-pass InSAR

bull Showed poor coherence for both L-band and C-bandndash This is only due to the movement of the branches and leaves

bull Showed wind effect on IfSAR Tree Height estimation

bull Future Workndash Different moisture conditions in branches and leavesndash Large database generation

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 29: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Questions

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 30: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Extra Slides

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 31: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Coherence due to Windinc angle = 436deg

bull L-band coherence drops below 07 at 239s 101sbull C-band coherence drops below 07 at 16s 07sbull C-band has significantly more scattering from the trees upper

branches than does L-band and so their movement will create greater decorrelation at C-band than L-band

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0

Page 32: Radiation Laboratory Quantifying the Effects of Wind on polarimetric SAR & InSAR Tree height estimation Michael L. Benson Dr. Leland E. Pierce Prof. Kamal.

Radiation Laboratory

Windrsquos effectσ0 comparison

bull At both L and C bands variations in σ0 are minimalbull σ0 is relatively unaffected by wind motion

L-Band C-Band

σ0