Site effect characterization of the Ulaanbaatar basin THÈSE DE DOCTORAT DE L’UNIVERSITÉ STRASBOURG Site effect characterization of the Ulaanbaatar basin Chimed Odonbaatar Supervised by: Luis Rivera & Antoine Schlupp

Plan of presentation Context of the study Site effect characterization of the Ulaanbaatar basin: Plan of presentation Context of the study Seismic activity Weak motion and microtremor recordings Amplification and amplified frequencies Seismic wave modeling Duration variation versus frequency Conclusions

MONGOLIA: Population - 2.75 million Area - 1.56 million km2 Context of the study: Population - 2.75 million Area - 1.56 million km2  MONGOLIA:

Ulaanbaatar Context of the study: Seismic activity 05/01/1967 Mw=7.2 D=260 km 4/12/1957 Mw=8.1 D=550 km 24/11/1998 ML=5.4 09/01/2010 ML=5.1 D=(190 km)

Context of Ulaanbaatar: Seismic activity (after Ulziibat et al 2010) Seismic crisis starts -2005

Context of the study: Seismic activity Several earthquakes has been felt recently (1998 – 2010) The number of buildings are rapidly increasing Ulaanbaatar urban area Ulaanbaatar is the capital of Mongolia where 1/2 (more than 1.2 million) of the total population of country is living Ulaanbaatar is located on the Tuul river basin (4 km x 25 km)

Mogod earthquake (05 Jan 1967, Mw=7.2), distance 260 km Context of the study: Seismic activity Mogod earthquake (05 Jan 1967, Mw=7.2), distance 260 km (after Medvedeva C. V. 1971).

104 sites Location of measured sites Context of the study: Weak motion and microtremor recordings Location of measured sites :104 sites :32 sites 8 short period 3 component station 104 sites

Amplified signal at sediment Context of the study: Weak motion and microtremor recordings Amplified signal at sediment

Plan of presentation Context of the study Site effect characterization of the Ulaanbaatar basin: Plan of presentation Context of the study Amplification and amplified frequencies Spectral ratio methods and analysis HV noise stability ? Results Seismic wave modeling Duration variation versus frequency Conclusions

Spectral ratio procedure Amplification and amplified frequencies : Spectral ratio methods and analysis SSR (Standard spectral ratio) is the spectral ratio between Reference and Sedimentary sites on S waves Spectral ratio procedure HVevent Is the spectral ratio between Horizontal and Vertical components on S waves HVnoise Is the spectral ratio between Horizontal and Vertical components on microtremors

Reference site qualification Amplification and amplified frequencies : Spectral ratio methods and analysis Reference site qualification Reference -> rock site without specific amplification Site to Reference distance<< site to event distance HV noise on Reference SSR between Reference and ALFM (permanent station at rock)

Comparison HVnoise, HVevent and SSR results Amplification and amplified frequencies : Spectral ratio methods and analysis Comparison HVnoise, HVevent and SSR results - Same amplified frequency - Different amplification factor - HV noise gives minimum and SSR gives maximum amplification

Comparison HVnoise and HVevent with SSR results Amplification and amplified frequencies : Spectral ratio methods and analysis Comparison HVnoise and HVevent with SSR results A-clear peak, B-wide peak, D-very wide or flat peak

Comparison HVnoise and HVevent with SSR results Amplification and amplified frequencies : Spectral ratio methods and analysis Comparison HVnoise and HVevent with SSR results

Influence of noise level on HVnoise ratio Amplification and amplified frequencies : HV noise stability ? Influence of noise level on HVnoise ratio

Influence of local noise sources on HVnoise ratio Amplification and amplified frequencies : HV noise stability ? Influence of local noise sources on HVnoise ratio

Influence of local noise sources on HVnoise ratio Amplification and amplified frequencies : HV noise stability ? Influence of local noise sources on HVnoise ratio Day is higher noise level Z noise level less increased than H

Polarization of the amplified peak Amplification and amplified frequencies : HV noise stability ? Polarization of the amplified peak 135

Azimuth dependence of H/V ratio Amplification and amplified frequencies : HV noise stability ? Azimuth dependence of H/V ratio

HVnoise ratio polarization / known noise sources Amplification and amplified frequencies : HV noise stability ? HVnoise ratio polarization / known noise sources day time Night time

Map of SSR amplification factor Amplification and amplified frequencies: Results Map of SSR amplification factor

Map of HVnoise (or SSR) amplified frequencies Amplification and amplified frequencies : Results Map of HVnoise (or SSR) amplified frequencies Figure 2.23. HV peak frequency distribution across Ulaanbaatar basin

Plan of presentation Context of the study effect characterization of the Ulaanbaatar basin: Plan of presentation Context of the study Amplification and amplified frequencies Seismic wave modeling Basin properties 1D and 2D modeling Comparison of modeling with observations Amplified frequency zoning Duration variation versus frequency Conclusions

Topography Geology Gravimetery Drillings Sediment thickness Seismic wave modeling: Basin properties Topography GIS DATABASE Geology Gravimetery Drillings 54 reached base rock Sediment thickness

Available data => 3D basin model Seismic wave modeling: Basin properties Available data => 3D basin model

Microtremor array measurements to estimate S-wave velocity Seismic wave modeling: Basin properties Microtremor array measurements to estimate S-wave velocity

Results of the high resolution FK analysis Seismic wave modeling: Basin properties Results of the high resolution FK analysis Vs~570 m/s H=30m Vs~520 m/s H=28m Vs~600 m/s H=37m

Input of the 3D model for 1D and 2D wave modeling Seismic wave modeling: 1D and 2D modeling Input of the 3D model for 1D and 2D wave modeling --------------------------------------------- 3D model + 2D horizontal grid => 1D vertical modeling at each node of the grid (4000 Nodes) Shear wave velocity of each node is interpolated from nearest array velocity structure. Use of Shake91, a program for equivalent linear seismic response analyses 3D model => 2D modeling (cross section) Average shear wave velocity is used for all cross sections sediments Use of Mka3D, a code using discrete element method, developed by Christian Mariotti,( CEA-DASE)

Seismic wave modeling: Comparison of modeling with observations Map of amplified frequencies using 1D modeling: comparison with observed amplified frequencies model accuracy (3D model , velocity)?

Location of the cross section used in the 2D simulation Seismic wave modeling: 1D and 2D modeling Location of the cross section used in the 2D simulation

ref sediment Base rock 2D simulation Seismic wave modeling: 1D and 2D modeling 2D simulation ref Base rock sediment

-1D modeling predicts well the amplified frequencies, Seismic wave modeling: Comparison of modeling with observations -1D modeling predicts well the amplified frequencies, -2D modeling predicts better the shape of wide peaks

Data “layers” used to build amplified frequency zoning map Seismic wave modeling: Amplified frequency zoning Data “layers” used to build amplified frequency zoning map ------------------------------------------------- 1D modeled amplified frequencies + Observed amplified frequencies + Geological Map + Topography and Satellite image = Amplified frequency zoning map of the Ulaanbaatar basin

Amplified frequency zoning map of the Ulaanbaatar basin Seismic wave modeling: Amplified frequency zoning Amplified frequency zoning map of the Ulaanbaatar basin

Plan of presentation Context of the study Site effect characterization of the Ulaanbaatar basin: Plan of presentation Context of the study Amplification and amplified frequencies Seismic wave modeling Duration variation versus frequency Duration based on the coda Duration based on arias intensity Conclusions

Plan of presentation Context of the study Site effect characterization of the Ulaanbaatar basin: Plan of presentation Context of the study Amplification and amplified frequencies Seismic wave modeling Duration variation versus frequency Duration based on the coda Duration based on arias intensity Conclusions

Location of selected sites for the duration study Duration variation : Data qualification Location of selected sites for the duration study Sites that recoded events with signal to noise ratio higher than 3.

Duration estimation based on the coda: measurement procedure Duration variation : Duration based on the coda Duration estimation based on the coda: measurement procedure

Duration estimation based on the coda: Comparison with SSR Duration variation : Duration based on the coda Duration estimation based on the coda: Comparison with SSR * Lengthened frequency = amplified SSR frequency related to amplification factor

Duration variation : Duration based on the coda Impact of the analysis window length and the sediment thickness to the duration ratio

Duration variation : Duration based on the coda Response of 1D modeling

Plan of presentation Context of the study Site effect characterization of the Ulaanbaatar basin: Plan of presentation Context of the study Amplification and amplified frequencies Seismic wave modeling Duration variation versus frequency Duration based on the coda Duration based on arias intensity Conclusions

Duration variation : Duration arias

Response of 1D modeling Arias duration does not include 1D amplification => Study of others lengthening sources

H<15m H~30 m UB9 Duration variation : Duration arias 5 7 6 8 3 5 5 7 6 8 5 8 3 5 2

Arias duration variation from 2D simulation Duration variation : Duration arias Arias duration variation from 2D simulation SSR The duration lengthening is at higher frequency than the amplified frequency

Site effect characterization of the Ulaanbaatar basin: Conclusions 1/2 Amplified frequency is independent from the spectral ratio method (HVnoise, HVevent, SSR). Amplification factor of HVnoise ratio is influenced by the noise level and local noise source azimuth. The amplified frequency of the HVnoise ratio can be shifted in particular situations. SSR gives higher amplification factor than HV ratio. We observe an amplification factor between 3 and 10 at the Ulaanbaatar basin with SSR on weak motions. A 3D geometrical model was produced using available data. Velocity structure determined locally gives a shear wave velocity of ~ 550 m/s at sediment. The amplified frequencies at Ulaanbaatar basin vary mainly between 2 and 5 Hz. An amplified frequencies zoning was deduced from observations and 1D modeling.

Site effect characterization of the Ulaanbaatar basin: Conclusions 2/2 The lengthening duration, deduced from the coda wave, shows an absolute duration lengthening It is the maximum at the same frequency that the amplification frequency. The observed lengthening duration up to 100%. The lengthening duration, deduced from normalized Arias intensity, is due to basin effects outside the 1D amplification factor. We observe locally lengthening between 5 to 13 seconds for 40 to 80 seconds weak motion signals. The most lengthened frequency is always higher than the amplified frequency. The main duration lengthening, observed with weak motion at Ulaanbaatar basin, is due to 1D amplification.

Site effect characterization of the Ulaanbaatar basin: Next … The results of this study should be included in the Ulaanbaatar seismic hazard assessment, capital of Mongolia. The strong motion site effect will be modeled, beyond the thesis, with a 3D modeling within the framework of the collaboration between the RCAG, the CEA-DASE and the EOST. The GIS data base (model, measurements, modeling results) built during this PhD will be extended with additional data as geophysical exploration data, new drillings, new records, etc.

Thank you for your attention Site effect characterization of the Ulaanbaatar basin: I would like to thank The CEA-DASE for its assistance and collaboration, specially for the instruments that have been provided and the 2D modeling. The French Embassy for they financial support of my stays at Strasbourg. Thank you for your attention