Inference on subsurface sources from multiparametric numerical analysis of geodetic data José Fernández and Antonio G. Camacho Institute of Geosciences.

Inference on subsurface sources from multiparametric numerical analysis of geodetic data

José Fernández and Antonio G. Camacho

Institute of Geosciences (CSIC, UCM)

Fac. C. Matemáticas, Plaza de Ciencias, 3

28040-Madrid, Spain

([email protected]; [email protected])

Collaborators

INGV-OE, ItalyA. Bonforte, F. Cannavò, D. Carbone, G. Puglisi, F. Guglielmino, M. Matia

INGV-OV, Italy G. Berrino

IREA-CNR, ItalyF. Casu, S. Pepe, E. Sansosti, P. Tizzani

WU, CanadaP. J. González, K. F. Tiampo

CCRS, CanadaS. V. Samsonov

JPL, USAP. Lundgren

Outline

> Introduction>Methodology>Simulation tests>Application test cases>Future developments

Introduction

> Changes in gravity and/or surface deformation are often associated with volcanic and seismic activity.

> Usually displacement and gravity changes are not simultaneously inverted.

> Normally deformation sources with a simple and a priori defined geometry are used, considering for volcanic activity mass and/or pressure changes.

> New methodology which try to skip these limitations.

Introduction

Methodology

Camacho, A. G., P. J. González, J. Fernández, and G. Berrino (2011), Simultaneous inversion of surface deformation and gravity changes by means of extended bodies with a free geometry: Application to deforming calderas, J. Geophys. Res., 116, B10401, doi:10.1029/2010JB008165.

We assume:

> Surface deformation and gravity changes are due to

changes in density and/or pressure and/or slip

allocated within extended source structures.

> Homogeneous elastic medium.

> Anomalous density and/or pressure changes, and/or

slip, are nearly homogeneous within the causative

bodies according to some prescribed values for

these magnitudes.

Methodology

> Simultaneous non-linear inversion of gravity and displacements

as produced by extended bodies with a free geometry.

> Assuming simple homogenous elastic conditions, the

approach determines general geometrical configurations of

pressured, density and slip source structures corresponding to

prescribed values of anomalies.

> These source bodies are described as aggregation of elemental

point sources for pressure, density and slip, and they fit the

whole data (keeping some 3D regularity conditions).

> The approach works in a growth step-by-step process that

allows very general geometrical configurations.

Methodology

> Solution for 3-D model space

> Semi-automated data inversion routine.

> Acceptance of non-gridded non-planar imprecise data.

> Simultaneous inversion for both positive and negative density contrasts, positive and negative pressure sources, and dislocation sources (slip).

> Simultaneous inversion for gravity and deformation data (levelling, three GPS components, InSAR LOS ascending and descending,…).

> Inversion for irregularly shaped structures composed of several individual bodies.

> Very short running time allowing quite real time computations.

Methodology

3D grid of (empty) point sources

Gravity data dg+ Covariance matrix Qg

Deformation data dx, dy, dz -levelling -GPS -InSAR

+ Covariance matrix Qd

Model equations

……………………….………….…………………………………..

gok

m

kk

ik

ki

m

kkk

ik

ki

ikik

kici

nidzFdgd

ZzF

d

Zz

d

Fg

d

ZzGdg

,...,1,pv34

1

v)(

)1(24

01

3

12

20

3

Fit conditions

Regularization conditions

Mixed condition

.min11 PP

TPGG

TG εQεεQε

.min1 mQm MT

, = f + -1M

T2T min-1D mQmεQε

Elastic parametersDensity and pressurecontrasts

Growth process (scale factor)

Extended 3D models for structures of anomalous density and pressure, defined as aggregation of (filled) point sources

Nearly automatic inversion process

Methodology

Simulation tests

Simulated anomalous structure composed by:- Vertical ellipsoid with anomalous pressure- Horizontal parallelepiped with anomalous mass

Simulated data:

- Gravity and levelling changes.- InSAR (LOS) for ascending and

descending passes.

Simulation tests

Modelled anomalous structure

Simulation tests

Simulation tests

Some vertical and horizontal cross-sections of the 3D modelled structures

> Fit between simulated and modeled structures is good.

> Magnitude, location, depth and geometry of the modeled structures approach the original structure.

> The inversion approach tends to generate rather rounded bodies.

> Sizes and depth are quite good. However, for the case of the depressurized ellipsoid we observe some expansion in its bottom that exceeds the original contour. It is produced because the bottom of the ellipsoid is too deep with respect to the survey diameter.

> In the example, for very peripheral or very deep areas in the model, some distortions can be observed.

Simulation tests

Application test cases

(G. Berrino, ING-O.V.,

P. Tizzani, CNR-IREA)

Application test cases: Campi Flegrei

- Gravity changes at 15 benchmarks

- Levelling changes at the same 15 benchmarks

Data:

- SBAS DInSAR data (LOS) for ascending and descending passes.

Period: 1992-2000

Observed and modeled

Gravity , Elevation changes, Ascending and Descending LOS data

Application test cases: Campi Flegrei

Some central cross-sections of the 3D model for anomalous pressure

No significant anomalous mass detected

Application test cases: Campi Flegrei

> Data (gravity, GPS, InSAR LOS) from 1995-2000 are inverted.

> Data are organized on annual periods (1995-1996, 1996-1997,…).

> For each annual period we carry out an inversion fit modelling sources of pressure and mass changes and sliding.

> First results assuming no particular hypothesis about relative weighing for the data sets have been obtained.

> Input data from SAR are composed by an arbitrary selection of pixels (for computation agility).

> SAR data is amplified to approximate the GPS scale.

95-96 96-97 97-98 98-99 99-00 Mass + - - - + + : increasePress + - + - : decreaseSliding + + +

Application test cases: Etna

A. Camacho, D. Carbone, A. Bonforte, F. Guglielmino, J. Fernández, and G. Puglisi (2013). Simultaneous inversion of ground deformation and gravity changes using bodies with free geometry. Application to data from Mt. Etna (Italy). EGU General Assembly, Vienna, April 7-12, 2013.

First r

esults

First r

esults

Bonaccorso , A. Bonforte, G. Currenti, C. Del Negro, A. Di Stefano, F. Greco (2011) Journal of Volcanology and Geothermal Research 200, 245–254.

Mass increase/decrease at 4 km depth

Press. increase at 5-10 km

First r

esults

Comparison with Bonaccorso et al. (2011)

Future developments

I. Testing for several sites (CF, LV, ETNA, Hawai’i,…) with more in depth studies/interpretation and new data sets (under development in the frame of MED-SUV project and other collaborations)

II. Similar modeling [3D extended structures with free geometry] for surface deformation produced by faulting (in validation phase)

III. Combination of different sources [mass changes + pressure changes + faults] (in validation phase)

IV. Test and validation of methodology running in automatic mode for real time monitoring purposes (under development in the frame of MED-SUV project and other collaborations)

V. Others (e.g, improvement of running time of the inversion code, consideration of viscoelastic properties, …)