Istituto Nazionale di Geofisica e Vulcanologia Regione autonoma Valle D Aosta Università degli Studi di Genova (Dip.Te.Ris., Geofisica) CORSO DI AGGIORNAMENTO PER GEOLOGI 18-19 Ottobre 2011 STUDI DI MICROZONAZIONE SISMICA: TEORIA E APPLICAZIONI Gli effetti di amplificazione topografica: aspetti teorici e casi studio Marco Massa (*) (*) Istituto Nazionale di Geofisica e Vulcanologia, Milano
Summary 1 TOPOGRAPHIC EFFECTS: STATE OF THE ART Primary topographic effects Induced (or secondary) topographic effects The Italian seismic rules for building (NTC 2008) 2 A DETAILED CASE STUDY : NARNI (CENTRAL ITALY) Seismic monitoring of the topography Results from recorded data Comparison with numerical simulations Evaluation of NTC 2008 3 EXAMPLES FROM OTHERS ITALIAN SITES
Seismic local site response All factors able to modify a seismic signal in amplitude, duration and frequency content during its propagation from the source to the recording site A1 (ω)( = S(ω) ) P(ω) s1(ω) ) t1(ω) A2 (ω)( ) = S(ω) ) P(ω) s2(ω) ) t2(ω) F For sites 1, 2 (and also site 3) S(ω) and P(ω) are common For sites 1 and 2 s1(ω) = s2(ω) (same lithology) For sites 1 and 2 t1(ω) t2(ω) For site 2 t2(ω) = 1 (flat surface) A1(ω)/A2(ω) = t1(ω)
Primary Topographic effects: causes 1) The free oscillation of the topography can have significant effects on seismic waves when the incident wavelengths are comparable to the size of the topographic features and the topographic slopes are relatively steep; Homogeneous body Resonance frequency f 0 = n V/L (from Geli et al., 1988) L = width of the topography V = shear wave velocity n= 1 fundamental mode 2) The variations of seismic motion, related to isolated reliefs, are due to different physical phenomena such as the focusing of seismic waves near the crest, because of the reflection on free surface and/or the interaction between incident and diffraction waves.
Geometrical factors : SR and WIDTH 1) The shape ratio (SR) H/L 2) The total width L H (from Lanzo and Sivestri, 1999)
Geometrical factors : vertex angle Homogeneous and elastic material φ = 90 Surface motion equations: v= v 0 e[i ω (t+z/ β)] = v 0 eikz ei ω t (Sanchez-Sesma, 1990) v= v 0 (eikz +e-ikz)+ v 0 (eikx +e-ikx) v= 2v 0 (coskx + coskz) On vertex, x=0 and z=0 v(0,0)= 4 v 0 In general v (0,0)= v 0 360 / φ For φ = 120 v (0,0)= 3 v 0 For φ = 180 v (0,0)= 2 v 0 For φ = 270 v (0,0)= 1.3 v 0 270 120
Effects induced by topography : sub-blocks blocks Marzorati et al., 2010
Effects induced by topography : landslides (Burianek et al., 2009) Stable area Unstable area Frequency [Hz] Frequency [Hz]
Time domain : crest vs. base Transversal section of Narni ridge, NRN4 (top), NRN1 (base) PGV (%) = 118
Time domain : NS component vs. EW Narni,, station at the top Mw 4.2, R 5.5 km No filtered data PGA (%) = 103 Other stations on topographies
Frequency domain : polarization MATLAB Handle Graphics Topography Polarization effects in EW direction H/V difference = 4 H/V=4 Absence of topography H/V difference = 0.3 H/V=0.3
Frequency domain : different phases Agreement between different phases
Italian seismic rules for buildings (NTC 2008, D.M. 14/01/2008)... per configurazioni superficiali semplici si può adottare la seguente classificazione Le suesposte categorie topografiche si riferiscono a configurazioni geometriche prevalentemente bidimensionali, creste o dorsali allungate, e devono essere considerate nella definizione dell azione sismica se di altezza maggiore di 30 m
Techniques for experimental site response evaluation Horizontal to Vertical Spectral Ratio (Lermo and Chavez-Garcia, 1993) Standard Spectral Ratio (Borcherdt, 1970) Directional Analysis A ij (f) = S i ( f ) *P ij (f)*g j (f)*i j (f) A ij H(f) / A ij V(f) A ij (f) = S i ( f ) *P ij (f)*g j (f)*i j (f) A ij (f) / A ik (f)= [S i ( f )*P ij (f)*g j (f)*i j (f)]/[s i ( f )*P ik (f)*g k (f)*i k (f)] G j (f) / G k (f) F A ii H(f) A ii V(f) A ik (f) A ij (f) Example of directional SSR at Narni site NON REFERENCE SITE TECHNIQUE REFERENCE SITE TECHNIQUE
THE CASE STUDY OF NARNI RIDGE (A-T3)
Morphological setting Azimuth dorsale: N 31 W - max H ~ 220 m Scarpata Est SE Scarpata Sud-Ovest 22 450 m1300 m 870 m 35 NW Scarpata Nord-Ovest Lato Nord Lato Nord
Geological setting and temporary velocimetric network Geology : Massive limestone Network : 10 surveyed sites from March to September 2009 3 white triangles : base 6 black triangles : crest 1 grey triangle : middle 1 white square : DPC station Sensors : velocimeters Lennartz LE3D-5s (flat instrumental response 0.2-40 Hz) Recording systems: 24 bits Reftek 130/01 and 20 bits Lennarts Mars-Lite
Data set and processing 702 events (about 10.000 waveforms) Local Magnitude : 1.5 5.3 642 from April 2009 L Aquila sequence Epicentral distance : 5 100 km Analyses HVSR (Single station spectral analysis ) SSR (Standard spectral ratio ) Mean removal and baseline correction; Butterworth filter 0.2 Hz - 25 Hz; FFT on different windows (S-phase and coda); Smoothing (Konno Omachi, b=20); rotations of NS component (0-175, step 5 )
The target event: 16 th December 2000, Mw 4.2 R 5.5 km Est of Narni (depth 9.8 Km) NRN station 10s of S phase
Microtremor survey (results on EW component) NRN2 (base) NRN1 (middle) Top 5 Top 4 Top 1 Top Top Top 2 3
Non-reference site technique (HVSR) NRN7 NRN1 NRN2
SSR : dependence on source to site direction NRN2 NRN7 All azimuths (R<30 Km, 1.5 M 3.6) azimuths between 60 and 120 NRN4
SSR : dependence on epicentral distance NRN2 NRN7 R < 30 Km 60 < R < 80 Km NRN7/NRN2 NRN7/NRN2
SSR : dependence on phase NRN2 events R < 30 Km, 1.5 M L 3.6 NRN7 NRN4 S-phase coda S-phase coda
SSR : dependence on components NRN2 Horizontal Vertical NRN7 NRN4
Wrong evaluations : HVSR vs. SSR An example from L Aquila aftershocks 4.7 ML 5.3 HVSR NRN1 base middle top SSR NRN2 NRN7 NRN1 NRN1 middle top
Wrong evaluations : soil-structure structure interaction NR10 NRN4 Noise measurement in the tower (2nd floor) NRN4 SSR between top and base SSR between two stations at the top not polarized not polarized
Considerations about experimental results For the stations installed at the top the amplification effects involve frequencies between 3 Hz and 5 Hz, with an amplification factor up to 9 (with respect to the reference station); The peak between 3 and 5 Hz shows a clear polarization effect: the highest amplifications are detected for direction perpendicular to the main axis of the ridge; The amplification factor increases with increasing difference of quota between top and bottom; The amplification factor increases with respect to the source to site direction, showing the highest values for direction perpendicular to the main elongation of the ridge; The same result is obtained if different phases of signal are considered (noise or coda); Amplification peak at frequencies between 4 Hz and 10 Hz are detected also on vertical component; HVSR results generally well agree, in terms of frequencies, with those obtained from SSRs, even if the case of L Aquila highlights the possibility of wrong interpretations in absence of SSR; Noise measurements, being a fast and cheap tool often used in site response analyses, appear to be suitable but only for very preliminary considerations.
2D and 3D modelling: : input parameters 3(z) 2(y) 1(x) K P - SV Method: boundary elements (BEM) Codes: 2D : HYBRID (Kamalian et al., 2003) 3D : BEMSA (Sohrabi et al., 2009) Domain: elastic, homogeneous and isotropic Input parameters: γ = 23.5 KN/m 3 θ = 0.37 Vs = 1000 m/s Vp = 2210 m/s Input at bedrock: Ricker wavelet (fc=3hz) Investigated frequencies 1-8 Hz Transverse sections : P1 (NRN7), P2 (NRN4). DEM resolution 20 m f ( t) A max 1 2 ( f p ( t t 0 )) 2 exp ( f p ( t t0 )) 2
2D results: NRN7 NRN2 NRN7 NRN4 SSR H/V
2D results : NRN4 NRN2 NRN7 NRN4 SSR H/V
3D results: NRN4 SSR H/V
Considerations about numerical simulations At least in terms of frequencies, 2D and 3D analyses agree with the experimental ones: a constant amplification between 4 and 5 Hz was found. As highlighted in many studies, the models are not able to reproduce the amplification estimated by the experimental analyses: an underestimation of a factor up to 3 was found; Having the Narni ridge a clear 2D configuration, the 3D model does not improve the final results; Considering the discrepancies, in term of amplification factors, between experimental results and modeling, it is possible to suppose that the observed amplification between 3 and 5 Hz derives from a coupling of different factors (e.g. topography, rock weathering and structural anisotropies).
Evaluation of Italian predictive models Characterization of anomalies for the single station Calculation of normalized residual (SA, 5%) between observed and predicted values Reference Italian model : Bindi et al., 2010 (M >4; R<200; rock site NTC A category) H (WE) H (NS) V
Topography corrective coefficients Calibration of empirical predictive models, in terms of SA (5%) up to 1s, for NRN2 (base) and NRN7 (top) stations. Log10 ( Y ) = a + ( b*m ) + ( c * Log ( R ) ) + σ a, b, c coefficients for NRN2 (59 near field earthquakes) Log10 ( Y ) = a + ( b*m ) + ( c * Log ( R ) ) + ( St * 1 ) + σ St corrective coefficients for NRN7 1 2 St corrective coefficients EW component at NRN7
Evaluation of NTC 2008 - Shape Normalized (to a g ) design response spectrum for Narni A-T3 (in black) Weak motions from Narni experiment (NTC for A T3, in grey) 16/12/2000 Mw 4.2, R 5.5
Evaluation of NTC 2008 - Amplitude (OPCM 3519, 28/04/2006) 10% of probability (T 475 yr) to exceede in the next 50 yr PGA (on rock) between 0.150 and 0.175 g (http://esse1-gis.mi.ingv.it/) (Barani et al., 2009), Italian predictive model (Bindi et al., 2010, in black), for an event of Mw 5 at 5 km, as inferred from the deaggregation analysis (SA at 0.2 s). Grey line is NTC for A-T3 Grey squares are SA (for A T1) corrected for the St topography coefficients
Example of other Italian sites
Aulla, A - T3 PGA (%) = 159
Sellano, B - T2 PGA (%) = 32
Lauria, A - T2 2σ 1σ 2σ 1σ H V PGA (%) = 69
Castelvecchio Subequo, A - T3 CA03 (crest), CA02 (base) PGA (%) = 126
Evaluation of NTC in Amplitude for CA03
Final comments Considering the experimental evidences (also reported in bibliography), a no negligible amplification of ground motion in presence of a particular topography exists. All investigated site are characterized by: 1) strong polarization effects for directions perpendicular to the main elongation of the topography; 2) overestimation with respect to the ground motion predictions for normal rock site; 3) amplification involving only a narrow range of frequency. NTC (D.M. 14/01/2008) The St corrective coefficient (ranging from 1 to 1.4) proposed by NTC08 is period independent and seems not to be useful to well predict the real amplifications. A simple shift (in amplitude) of the design response spectrum for A-T1 site leads to underestimate the frequencies of interest. SEISMIC INPUT Positive residuals between recorded data at the top of topographies and predicted ones, highlight high frequency amplifications (> 1 Hz) also for hard-rock site (A category): these recordings don t have to be used as seismic input for other analyses.
da NTC 2008, cap 3.2.2 (Azione sismica) per condizioni topografiche complesse è necessario predisporre specifiche analisi di risposta sismica locale;... per configurazioni superficiali semplici si può adottare la seguente classificazione...... ma esistono nella realtà configurazioni semplici?
thanks for your attention