THE GOCE ESTIMATED MOHO BENEATH

Transcription

1 HE GOCE EIMAED MOHO BENEAH HE IBEAN PLAEAU AND HIMALAYA D. ampietro (1), M. Reguoni (1), C. Braitenberg (2), R. Barzaghi (1) 1) DIIAR, Politecnico di Milano, Italy (2) Department of Geosciences, University of rieste, Italy he present research has been partially funded by EA through the GEMMA project and by AI through the GOCEI ALY project.

2 HE GOAL Afterjusttwoyearsinorbit,EA's GOCE satellite has gathered enough data to map Earth's gravity with unrivalled precision. he solution of inverse gravimetric problems can benefit from GOCE. he main objective of the GOCE EXPLOIAION FOR MOHO MODELING AND APPLICAION (GEMMA) project is to improve the knowledge of the crust mantle discontinuity surface by using data coming from satellite mission.

3 HE PROJEC he GEMMA project has been funded by EA and Politecnico di Milano and has a duration of 2 years. WP0000 Project management WP1000 GENERALIZAION OF HE PROPOE APPROACH O PHERICAL APPROXIMAION. WP2000 AEMEN OF HE COVARIANCE FUNCION AND OF HE EIMAION ERROR. WP3000 IOLAION OF HE GRAVIMERIC IGNAL ICE HEE EIMAION WP4000 ON LINE OFWARE AND MAP GENERAION WP5000

4 Based on two layer model; HE PROPOED APPROACH Input: grids of GOCE data (potential and second radial derivative) with their full covariance matrices; A priori information: densities (lateral and vertical variations), approximated moho model); Inversion method: collocation in the frequency domain (Wiener deconvolution)

5 Reduced GOCE data grids pherical domain Normal potential removal δ δ Linearization + ( x ) = ( x ) ( x ) = k ( x ξ) ( ξ) Fourier transform + dξ ( x ) = ( x ) ( x ) = k ( x ξ) ( ξ) δˆ δˆ () f = kˆ ( f ) ˆ ( f ) + ˆ ( f ) 0 υ 0 opography/isostatic effects removal () f = kˆ ( f ) ˆ ( f ) + υˆ ( f ) convolution ediments effects removal dξ λ = γ = error spectrum ε, (f ) MIO Wiener filter prediction ˆ, ( f ) = λˆ 0 + γˆ 0 υ υ = Inverse Fourier ransform Estimated Moho υ υ υ + kˆ 2 + kˆ υ kˆ kˆ υ υ + kˆ υ 2 υ + kˆ + kˆ 2 2 υ υ 2 υ + kˆ 2 υ υ υ Error cov matrix Collocation solution

6 pace-wise solution INPU he main idea behind the space wise approach is to estimate the spherical harmonic coefficients of the geo potential model by exploiting the spatial correlation of the Earth gravitational field. Intermediate results that can be used for local applications: filtered data (potential and gravity gradients) along the orbit grid values at mean satellite altitude

7 INPU pace-wise solution GOCE data grids 400 Monte Carlo samples... 1

8 REDUCION OF HE OBERVAION Numerical integration to obtain the effect (at satellite level) of the topography (from etopo1) and of sediments (from CRU2.0 model).

9 MODELING LOCAL INFORMAION Corrections to the reduced observations can be considered: i.e. in the case of Moho Beneath the ibetan Plateau and Himalayas the density and the thickness of sediments from CRU2.0 has been corrected according to the local information given in Braitenberg et al., 2003.

10 MODELING MOHO COVARIANCE FUNCION Errors spectra are estimated from Montecarlo samples. ignals spectra are computed from the Moho covariance function. he theoretical covariance function (different models: Exponential, Gaussian, table, Matern) is computed by fitting with a least squares principle an experimental anisotropic covariance obtained from a priori Moho model. Distance [ ] Distance [ ] Distance [ ] Distance [ ] A priori Moho model [km] Empirical covariance matrix heoretical covariance matrix

11 EIMAED MOHO (GOCE ONLY) Estimated (propagated from the observations) error standard deviation = 1.6 km GOCE only Moho model [km] Error map from MonteCarlo samples [km]

12 EXPLOIING ADDIIONAL INFORMAION PROFILE 1 (WE EA) PROFILE 2 (NORH OUH) Crustal thickness estimated from wide angle profile (from Zhongjie Zhang and imon L. Klemperer, 2005). Accuracy of the profile = 5 km. Crustal thickness estimated from wide angle P wave reflections (from ai Lin seng et al., 2009) and by Gaussian beam migration of direct P to wave conversions (Nowack et al., submitted manuscript,2009). Accuracy of the profile = 3 km.

13 Normal potential removal Reduced GOCE data grids opography/isostatic effects removal ediments effects removal MIO Wiener filter Additional information e.g. seismic profiles δ δ Linearization + ( x ) = ( x ) ( x ) = k ( x ξ) ( ξ) Fourier transform + dξ ( x ) = ( x ) ( x ) = k ( x ξ) ( ξ) convolution dξ prediction ˆ δ δ D = μ + δ Gridded data λ Inverse Fourier ransform eimic profiles o δˆ δˆ ( f ) = kˆ ( f ) ˆ ( f ) + ˆ ( f ) 0 υ 0 ( f ) = kˆ ( f ) ˆ ( f ) + υˆ ( f ) Estimated Moho Error cov matrix

14 PARIIONED COLLOCAION OLUION he collocation system can be partitioned as: C C C C δ ˆ δ = μ + λ = C C C C δ,,,, o,,,, δ C, C D, C, C δ, o -1 1 C, C, C, C, z μ= λ C, C, C, C, z 1 1 C,, C C Do, C 1, C Do, C, C Do, Do, C, C Do Do, C, C Do, C, C, Do C C, C,,, D Do C δ λ = C his matrix is large (many gridded data) BU it has a oeplitz structure his matrix is small (few sparse points) Correction to the GOCE only Moho model [km] he system can be efficiently solved! (Reguoni and ampietro 2009)

15 EXPLOIING ADDIIONAL INFORMAION PROFILE 1 (WE EA) (from Braitenberg et al. 2009) Resolution (km) PROFILE 2 (NORH OUH) PROFILE 1 (WE EA) Mean value correctly recovered (difference between GOCE and eismic = 0.6km); moothing effect due to the particular Moho configuration; PROFILE 2 (WE EA) Mean value correctly recovered (difference between GOCE and eismic = 1.2km); Very large differences (about 10km) between 300 and 450 km;

16 EXPLOIING ADDIIONAL INFORMAION PROFILE 2 (WE EA) Very large differences (about 10km) between 300 and 450 km in corrispondence of the collision between Indian and Eurasian plates. DICREPANCIE BEWEEN GOCE AND EIMIC CAN BE UED O DEEC ANOMALIE INO HE MOHO GOCE only Moho GOCE + eimic profiles Moho

17 HE EIMAED MODEL GOOD AGREEMEN BEWEEN GOCE MOHO MODEL AND PRECEDEN MODEL. GOOD AGREEMEN WIH HE MAIN KNOWN ECONIC LINE (WHIE DOED LINE). FURHER UDIE ARE REQUIRED IN COLLIION AREA, WHERE GOCE MODEL AND EMIC PROFILE DEEGREE. MOOHING EFFEC DUE O HE PREENCE OF A VERY DEEP MOHO WIH HIGH VARIABILIY. Final estimated Moho model [km]

18 CONCLUION A method based on collocation and FF (in spherical approximation) has been implemented to evaluate the contribution of GOCE data in the Moho estimation. he effects of sediments (from the CRU2.0 model) in terms of gravitational potential and its second radial derivatives at satellite altitude has been computed worldwide. he possibility to add local geological/geophisical information (e.g. local sediment model, seismic profile) has been studied and implemented.

19 CONCLUION General good agreement between GOCE Moho model and seismic profiles at low frequencies. Discrepancies between GOCE model and seismic profiles can be used to detect anomalies into the Moho. moothing effect due to the presence of a very deep Moho with high variability. Better results are expected in other regions of the world.

20 HE PROJEC he GEMMA project has been funded by EA and Politecnico di Milano and has a duration of 2 years. WP0000 Project management WP1000 GENERALIZAION OF HE PROPOE APPROACH O PHERICAL APPROXIMAION. WP2000 AEMEN OF HE COVARIANCE FUNCION AND OF HE EIMAION ERROR. WP3000 IOLAION OF HE GRAVIMERIC IGNAL ICE HEE EIMAION WP4000 ON LINE OFWARE AND MAP GENERAION WP5000