Presently: DownUnder Geosolutionshttp://www.dugeo.com

Previously: Kyoto Universityhttp://earth.kumst.kyoto-u.ac.jp/~adam

adamaki@hotmail.com

Landstreamersand rapid Vs imagingwith surface waves

Adam O’Neill

1. Surface wave overview

2. Landstreamer QC tests

3. Field data inversion

4. Synthetic modelling

Contents

Petroleum seismicNoise !!!

‘Ground-roll’

or

‘Source generated

noise’

Yilmaz (2001)

Signal !!!

1. Acquisition

2. Processing

3. Inversion

Dispersion curves

Flat-layeredVS model

Engineering geophysics

Plane-wavetransform

Iterativeoptimisation

Shot gather

Civil / mining / environmental / transportation / petroleum

Civil engineering

Surface wave types

Water

Solid

Water

Stiff layer

PPP-wavemultiples

‘‘Leaky’ or ‘Guided’Leaky’ or ‘Guided’

Solid

SV

P

ScholteScholte

**

Air

Solid

P

SVRayleighRayleigh

*

Air

Solid

Stiff layer

SH

SH-wavemultiples

LoveLove

Rayleigh wave motion

http://www.kettering.edu/~drussell/Demos/waves/wavemotion.html

Counterclockwise ellipticalmotion at surface

De

cre

ase

s w

ith

dep

th

Pure vertical motionat about 1/5 wavelength Clockwise motion at depth

Layering effects

http://www.oyo.co.jp/product/1-geo_survey/6-surface_wave/surface_wave1.html

Pulse dispersion

…pulse changes shape

As

dis

tan

ce

inc

reas

es

…

*

Dispersion method (f-k)

λ = c/f

z≈λ/2.5β≈1.1c

(a) Off-endgather

(b) Transform and pick ridge

(c) Phase velocity curve

(d) Velocity-depth

Aliaswrap !

c(f)=f/k

Phase velocity relations

Normal

Inverse

Irregular

f (Hz)

f (Hz)

f (Hz)c

(m/s

)c

(m/s

)c

(m/s

)

β (m/s)

β (m/s)

β (m/s)

z (m

)z

(m)

z (m

)

Frequencies and depths

Earthquake

seismology

Engineering

geophysics

Frequency Depth Scenario

mHz 100's km Mantle

sub Hz 10's km Crust

Hz 100's metres Basin

Hz 10’s metres Deep

10’s Hz metres Shallow

kHz centimetres Road

MHz millimetres Materials

Surface wave benefitsProperty… Advantage…

Estimate shear-wave velocity For stiffness (Gmax) estimate

Detect stiffness reversals Caprock thickness and geo-hazards

High signal to noise Survey urban areas / long offsets

Low coupling dependency Use landstreamers / over roads

Survey rubble and waste landfill Where penetrometers not possible

Model velocity gradations More uniquely than refraction

In-field processing Rapid, cost-effective results

Non-destructive test Provide average, in-situ properties

Conventional vs ‘New’ modelling

Plane-wave matrix methods Full-wavefield P-SV reflectivity

Idealised model:

-Plane wavefronts only

-No acquisition-processing effects

-Pure surface wave modes only

-Smooth elastic contrasts only

Problem:

Failed for many difficult field sites

e.g. stiff / compacted surface layers

Realistic field test simulation:

-Spreading wavefronts

-Source-receiver effects

-Body wave contributions

-Stable for all elastic contrasts

- And mode identification-free !

Outcome:

Accurate results at nearly all field sites !

Low Velocity Layer

Miss soft layer!

Higher frequency modes not modelled…

Fundamental-mode, plane-wave modelling

Low Velocity LayerFull-wavefield modelling

Soft layer detected

Higher modes all fitted…

High Velocity LayerFundamental-mode, plane-wave modelling

Stiff layer poorly estimated

Low frequency mode(s)not fitted…

High Velocity LayerFull-wavefield modelling

Stiff layer recovered

Higher mode(s) fitted well…

Remaining problems

1D inversion only

- Wavefield scattering ‘corrupts’ dispersion curve

Wavefield discrimination

- Overlapping components (Love / Rayleigh / guided / reflected etc. )

Geological interpretation

- Relate stiffness model to lithology and significance

1. Surface wave overview

2. Landstreamer QC tests

3. Field data inversion

4. Synthetic modelling

Contents

Landstreamer specsOYO Japan – Geometrics USA

24 channel

4.5 Hz vertical geophones

Flat baseplates

Twin rope fasteners

Mueller clip takeouts

Landstreamer photosLong spread on road / short spread on sand

Some notesFrom our experience with flat baseplates

At 5 stacks, up to 1 shot per minute = maximum 400 per day

But comfortably get 200 shotpoints per day when off-road

24 channels at 2 m spacing easily pull by one person - on road or sands

Flat baseplates – easier to pull through mud and around corners – but can rock on gravelly/pebbly base

Tripod baseplates – maybe more resistance through soft material – and hard to slide laterally – but better coupling and less rocking no doubt (?)

Landstreamer vs spikesData and dispersion images

4.5 Hzlandstreamer

28 Hzspikes(plantedgeophones)

Higher mode transition – less clear with landstreamer

Landstreamer vs spikesDispersion and power curves

However, fundamental mode dispersion is equivalentOnly slight low-freq power loss with 28 Hz geophones

Moral – Don’t need to buy low frequency phones for surface wave surveys if you already have reflection ones!

Landstreamer resultsTie to downhole Vs log

Model:Soft clays detected

Field and synthetic dispersion curves

Landstreamer vs spikesNormalised waveforms

Surface wave pulse differs later in train – lower frequency portionNonetheless, phase velocity dispersion is equivalent

Landstreamer vs spikesAGC shot gathers

Air wave

28 Hz spikes

4 Hz landstreamer

Landstreamer more affected by early time noise and air-waveRefracted arrivals harder to pick

Maximum offsetsData and dispersion image

96 channels - 1 m near offset - on asphalt - 4-spread walkawayUpper frequency limited to about 70 Hz

Maximum offsetsAGC shot gather

Strong ground-roll – and air wave - to 100 m offsetWeaker first arrivals - possible reflections 40-60 m offset

Note: Low cut filter was turned off here (usually set at 3 Hz / 6 dB/octave) thus some DC shifts remain in raw data

Asphalt vs grassData and dispersion images

On asphalt

On grass

220 m/stop-mute

Poor coupling – ‘floating’ up to 2.5 cm on grass/sticks

Asphalt vs grassDispersion curves

Higher mode above 25 Hz not seen in grass data – no stiff surface

Lower frequency portion is similar shape – but offset parallel by 5 m – so difference within acceptable lateral variation limits

Positional repeatabilityData and dispersion images

Seemingly minor variations

Day 1

Day 3

Positional repeatabilityDispersion curves

Most likely reason for difference: Geophone re-positioning errorShotpoint relocated to within 10 cm

But streamer was dis- and then re-assembledPossible spacing differences and/or rope stretch

1. Surface wave overview

2. Landstreamer QC tests

3. Field data inversion

4. Synthetic modelling

Contents

Field tests

Test site 1

Niigata, JapanObjective: Locate extent of rising ‘mud volcano’ plumesSealed asphalt surface

Test site 2

Osaka, JapanObjective: Sediment mapping around faultDry, sandy surface

Rayleigh and Love landstreamer applications

Mud volcanoLocation map

Overpressured mud formations at depth

Surface via diapirs / conduits

No magmatism

Methane gas expelled (+CO2+N)

Can range from 0.5 m to 800 m high

Thousands worldwide, mostly offshore

Associated with petroleum systems

Also an engineering hazard e.g. offshore platforms, onshore infrastructure

Mud volcanoSite map and photo

A

B

Mud volcanoExisting data

Low resistivity= Mud plume

Higher resistivity= Weathered bedrock

Mud volcanoSeismic line location and parameters from walkaway test

Zone of most mud emanation

Best surface-wave / reflection surveyparameter selection

24 channels2 m geophone spacing10 m near-offset2 m shot spacing2048 samples at 0.5 ms5 stacks wooden mallet on road

Reasoning

Resolve surface wavelengths up to 20 m

Achieve maximum frequency up to 70 Hz

Possible reflections at 40-60 m offset

Mud volcanoShear wave velocity and resistivity images

Coarse models12 layers

0.5 – 2.5 m thickLow damping

Fine models24 layers

0.25 – 1.25 m thickHigh dampingLateral 5-pointmedian filtered

Two mud plumes

connecting at surface?

Higher resist.= gas or sands?

Mud volcanoMidpoints with Vs over 200 m/s

Scatteringeffects?

Indicates no mud

Possibly fresh or weathered basement

Mud volcanoMidpoints with Vs under 200 m/s

Possible mud plume ?

Or zone of scattering…

Scatteringeffects?

SH-wave source and landstreamer

River sands

Data and dispersion images

River sands

Love

(landstreamer)

Rayleigh

(planted

geophones)

Inversion results and interpretation

River sands

0-5 m = Post-fault cover - positive anisotropy (Vsh > Vsv)

>5 m = Faulted sediments - reverse anisotropy (Vsh < Vsv)

Vsh <> Vsv

Transverse

isotropy

1. Surface wave overview

2. Landstreamer QC tests

3. Field data inversion

4. Synthetic modelling

Contents

Synthetic modelling

Procedure

1. Use 2D numerical code to simulate full-wavefield2. Apply rollalong 1D inversion, as per field test

Test cases

(i) Soft pinchout(ii) Sinkhole

(iii) Fault

To verify 1D inversion reliability over 2D structures

Modelling methodElastic 2D Finite-Difference (4th order)

Receivers48 and/or 96 channels1 m geophone spacing

SourceVertical impact at surface2 m shot spacing

GeometryOff-end shots2.5 m near offsetBoth pushing (from left) and pulling (to right)

Imaging1D models plotted at spread midpoint

Soft pinchout2DFD model

Soft pinchoutInverted Vs image

96-channel shot pushing from left

Soft pinchoutInverted Vs image

96-channel shot pulling from right

Synthetic vs field imagesInverted VS images with shot pushing from left

Syntheticdata

96 channels1 m spacing

Fielddata

24 channels2 m spacing

Common features:

- Zone of anomalous dispersion around pinchout

- Covers about 20% of spread length, mostly beyond pinc

hout

- Pinchout location possibly overestimated by up to 10% o

f spread length

- When pushing spread off end of an LVL, prefer to plot mo

dels nearer to shot

(Or take average model between reciprocal shots)

SinkholeLimestone dissolution

SandSand

LateriteLaterite

SinkholeModelling inspiration

Hyden fault scarp field data (actually laterite)… Soft zone?

Sinkhole2DFD model

SinkholeInverted Vs image

96-channel shot pushing from left

SinkholeInverted Vs image

96-channel shot pulling from right

2DFD model

Fault

Genetic Algorithm inverted Vs image48-channel shot pushing from left

Fault

Raw CMPCC

CMP cross-correlation processing

Fault

Scatter !Smooth !

Dispersion curves and 1D misfits

Important observations

Scatter when 1/5 to 2/5

of spread is over fault

Smooth when spread is

midway over fault

RMS misfit not indicate

1D inversion breakdown

Conclusions•Landstreamer highly suitable for surface waves – shear wave velocity profiling

•Easily 200 shotpoints per day – more on smooth, straight road

•Coupling and/or geophone frequency not a concern for surface waves – 28 Hz are fine

•Only difference is with higher mode transitions – less clear with landstreamer

•Strong response at far offsets – to even 100 m or more

•Refractions slightly harder to pick at far offsets with landstreamer – and flat baseplates more susceptible to air wave

•2D profiling a shallow LVL pinchout shows scattering – but rollalong 1D inversion provides an accurate image

•Shorter spreads show clear forward/reverse shot differences – longer spreads show broader scattering but less shot geometry dependence

•Plotting model location at spread centre may not be best option – response appears dominated by material below the shot / nearer offsets

If time……show basement reflection synthetics

Basement below sinkhole2DFD model to depth

Surface waveproblem

Reflection problem

Basement below sinkholeP-SV shot gathers

1D reflectivity (viscoelastic) 2D Finite Difference (elastic)

-Strong surface and guided wave noise due to near-surface waveguide

-Basement P-P reflection only at far offset

-Viscoelastic has lower frequency content

Rayleigh modes-Fundamental-Higher

Guided wave (multiply reflected P-waves

P-P basement reflection

Basement below sinkholeCDP versus zero-offset

Surface seismic96 channels1 m spacing2.5 m near offset2 m shot spacingConventional CDP flow

Exploding reflectorSources every 10 cmat basement interface96 receivers at 1 mat surfaceLow-velocity pull-down

CDP processing gives pull-downs either side of sinkhole

*

Basement below sinkholeFar-offsets and static issues

…thus, apparent pull-down for CDP’s where source is over sinkhole due to static from soft material

* *RecSrcNo static here

With thin surficial waveguide, strong ground-roll only allows far-offset reflections to be identified and imaged…

Basement below sinkholeP-SV and SH synthetic shot gathers

P-P reflection-Far offset (stretched)-Noise shrouded

SH reflection-All offsets-After noise-Better image?