Linda Al A)k Sigma WG. Jack Baker, Ken Campbell, Brian Chiou, Ellen Rathje, Adrian Rodriguez- Marek, Melanie Walling. Outline

Transcription

1 /5/4 SSHAC Level 3 NGA- East Project WS- 2C/3A, October 29-3, 24 Berkeley, CA Preliminary Tau,, and Models for CENA Linda Al A)k Sigma WG Jack Baker, Ken Campbell, Brian Chiou, Ellen Rathje, Adrian Rodriguez- Marek, Melanie Walling Outline General TI Team approach for sigma logic tree Background; ObjecIves; General Approach Preliminary results for CENA data analysis Data distribuion, GMPE funcional form, and residual plots Analysis results for Tau,, and and comparisons Candidate sigma models New preliminary CENA models for Tau,, and ExisIng models 2

2 /5/4 ObjecIves for NGA- East Develop standard deviaion models for ground moion (PSA) applicable for CENA for: Between- event standard deviaion, Tau Ergodic within- event standard deviaion, Single- staion within- event standard deviaion, Models applicable for M 4. to 8.2; distance up to, km; hardrock sites; PGA, PGV and T. to sec 3 General Approach Analysis of residuals of CENA data with respect to developed GMPE to evaluate trends:, distance, V S3 dependence, closeness index Compare CENA,, and Tau to exising models at common magnitudes (<5.8) to inform CENA models at larger M ExisIng models: NGA- West2 models, European models, other Develop new models and evaluate exising models 4 2

3 /5/4 Residual Components δ = δb + δw δ es e es B e : Between- event (inter- event) residual for earthquake e δ W : Within- event (intra- event) es residual at staion s for earthquake e Ala)k et al. (2) 2 2 σ = τ + φ τ : Between- event standard deviaion φ : Within- event standard deviaion 5 Overview of Single- StaIon Given muliple recordings of GM at an individual site, S, allows esimaing the systemaic site effects, δ S2S S, and removing them from the GM variability: Single- Sta-on δ S2S S represents the systemaic deviaion of the observed amplificaion at this site from the median amplificaion predicted by the model δs2s s = Nbofeqks s Nbofeqk s e= δw es Nbofsta s= φ S2S = ( δ S2S ) 2 s Nbofeqk s Nbofsta Nbofeqk s ( δws ) 2 es s= e= φ SS Nbofsta = Nbofeqks s= φ = φ + φ 2 2 SS S 2S 6 3

4 /5/4 Summary of Terminology Residual Between- event residual, δ B e Within- event residual, δ W es Standard Devia)on Between- event standard deviaion, τ=sd(δ B e ) Within- event standard deviaion, φ=sd(δ W es ) Site- to- site residual (site term), δs2 S s Site- to- site variability, φ S2S =SD(δS2 S s ) Single- staion within- event residual, δw S es Single- staion within- event standard deviaion, φ ss =SD(δW S es ) Total (ergodic) standard deviafon: σ= φ 2 + τ 2 Single- stafon standard deviafon (parfally non- ergodic): σ SS = φ ss 2 + τ 2 Al AIk et al. (2) 7 PRELIMINARY RESULTS FOR CENA DATA ANALYSIS: - Data distribu-on, GMPE func-onal form, and residual plots - Analysis results for Tau,, and and comparisons 8 4

5 /5/4 CENA Data DistribuIon at T.sec (Min of 5 recs/eqk) Total: 3,933 recs 65 eqks 746 staions PIE:,849 recs eqks 424 staions 9 CENA Data DistribuIon at T.sec (cont d) 27 staions with < 3 recs 5

6 /5/4 CENA Data DistribuIon (cont d) 2: Paths contained within CUS, Canada (3,44 recs) 3: Paths contained within Appalachian (7 recs) 4: Paths contained within AtlanIc Coast (9 recs) 5: Paths crossing 2, 3, 4 (69 recs) Preliminary CENA Model (PSA) R rup < 5km R rup 5km ln ( ) Y c c M c M 2 = ( rup + ) + rup + + ln ( S ) c ln R c c R c c V ( ) a ( ) ln Y = c c ln 5 + c + c M + c M ( c + c ) ln ( R + c ) + c R + c + c ln ( V ) 4 4a rup 6 7 rup 6 8 S 3 Regression done for distances up to, km. Does not include Gulf region. 2 6

7 /5/4 Between- Event Residual Plots 3 Between- Event Residual Plots (cont d) 4 7

8 /5/4 Within- Event Residual Plots vs. Distance T. sec T. sec 5 Within- Event Residual Plots vs. V S3 T. sec T. sec 6 8

9 /5/4 Within- Event Residual Plots by Region 2: Paths contained within CUS, Canada 3: Paths contained within Appalachian 4: Paths contained within AtlanIc Coast 5: Paths crossing 2, 3, 4 7 PRELIMINARY RESULTS FOR TAU 8 9

10 /5/4 Constant Tau Nmin Nmin 3 Nmin 5 Nmin Mmin 2. Mmin 3. Mmin 4. Mmin Tau.5 Tau Nmin 3 5 Nb Eqks Mmin 2.: 65 eqks Mmin 3.: 56 eqks Mmin 4.: 26 eqks Mmin 5.: 6 eqks 9 Constant Tau: PIE/Tectonic.9.8 Nmin 5 All Eqks Tectonic Only PIE Only.9.8 Nmin 5, Mmin 3. All Eqks Tectonic Only PIE Only Tau.5 Tau At T. sec, All Eqks: 65 Tectonic Eqks: 54 PIE: At T. sec, All Eqks: 56 Tectonic Eqks: 45 PIE: 2

11 /5/4 Constant Tau: Comparisons Tau Mmin 3. CENA - Tectonic Only Zhao6 BSSA4 - M 4. CB4 - M 4. CY4 - M 4. ASK4 - M M- Dependent Tau and Comparisons T.sec CENA - Tectonic Only Zhao6 BSSA4 CB4 CY4 ASK Tsec CENA - Tectonic Only Zhao6 BSSA4 CB4 CY4 ASK4 Tau.5 Tau

12 /5/4 Depth- Dependence of Tau.9 T.sec All Eqks Tectonic Only.9 Tsec All Eqks Tectonic Only Tau.5 Tau Hypo Depth (km) Hypo Depth (km) 23 PRELIMINARY RESULTS FOR PHI 24 2

13 /5/4 Constant.9.8 Nmin 5, Mmin 3. All Eqks Tectonic Only PIE Only.9.8 Mmin Region 2, All Eqks.3 Region 2, Tectonic Only Region 2, PIE Only.2 Region 5. Region 3 Region 4.. At T. sec: All Eqks: 3,749 recs Tectonic:,9 recs PIE:,849 recs 25 Constant Comparisons Mmin 3. Mmin3 - Mmax6 - Rmax CENA - Tectonic Only CENA - PIE Only.3 Zhao6 BSSA4 - M4. - R3 - Vs76.2 CB4 - M4. - Vs76. CY4 - M4. - Vs76 meas ASK4 - M4. - Vs76 meas CENA - Tectonic Only.3 CENA - PIE Only ASK4.2 CB4. CY4 BSSA4.. Comparisons with proposed models Comparisons with NGA- W2 computed using within- event residuals 26 3

14 /5/4 M- Dependent and Comparisons T. sec T.sec T.sec - Rmax35km CENA - Tectonic CENA - PIE ASK4.2 CB4 CY4 BSSA4 27 M- Dependent and Comparisons T. sec Tsec Tsec - Rmax35km CENA - Tectonic CENA - PIE ASK4.2 CB4 CY4 BSSA4 28 4

15 /5/4 R- Dependent - T. sec T.sec T.sec - Mmin3 - Mmax6 CENA - Tectonic Only CENA - PIE Only Rrup (km).6.4 CENA - Tectonic Only CENA - PIE Only ASK4.2 CB4 CY4 BSSA Rrup (km) 29 R- Dependent - T. sec Tsec Tsec - Mmin3 - Mmax6 CENA - Tectonic Only CENA - PIE Only Rrup (km).6.4 CENA - Tectonic Only CENA - PIE Only ASK4.2 CB4 CY4 BSSA Rrup (km) 3 5

16 /5/4 MR- Dependent T. sec T.sec - M3 to4 T.sec - M4 to CENA - Tectonic Only.4 CENA - Tectonic Only.3 CENA - PIE Only.3 CENA - PIE Only.2 ASK4 CB4.2 ASK4 CB4. CY4 BSSA4. CY4 BSSA Rrup (km) Rrup (km) T.sec - M5 to CENA - Tectonic Only CENA - PIE Only ASK4 CB4 CY4 BSSA Rrup (km) 3 MR- Dependent T. sec Tsec - M3 to4 Tsec - M4 to CENA - Tectonic Only.4 CENA - Tectonic Only.3 CENA - PIE Only.3 CENA - PIE Only.2 ASK4 CB4.2 ASK4 CB4. CY4 BSSA4. CY4 BSSA Rrup (km) Rrup (km) Tsec - M5 to CENA - Tectonic Only CENA - PIE Only ASK4 CB4 CY4 BSSA Rrup (km) 32 6

17 /5/4 V S3 - Dependence of T.sec Tsec CENA - Tectonic Only CENA - PIE Only CENA - Tectonic Only CENA - PIE Only Vs3 (m/sec) Vs3 (m/sec) 33 PRELIMINARY RESULTS FOR PHISS 34 7

18 /5/4 Data DistribuIon at T. sec Min of 5 recs/eqk and 5 recs/sta: 3,33 recs (,632 Tect -,4 PIE) 65 eqks ( PIE) 337 staions 35 and Nmin = 5 recs S2S

19 /5/4 Constant Nmin 5 and Mmin 3. All Eqks Tectonic Only PIE Only Nmin = 5 and Mmin = 3. Region 2: All Eqks Region 2: Tectonic Only Region 2: PIE Only Region 5 Region At T. sec: All Eqks: 2,869 recs Tectonic:,468 recs PIE:,4 recs 37 Constant Comparisons Mmin3 - Mmax6 - Rmax35km CENA - Tectonic CENA - PIE ASK

20 /5/4 M- Dependent Comparisons T. sec Per. sec - Rmax km CENA - Tectonic Only CENA - PIE Only Per.- Rmax35km CENA - Tectonic CENA - PIE ASK M- Dependent Comparisons T. sec Per. sec - Rmax km CENA - Tectonic Only CENA - PIE Only Per.- Rmax35km CENA - Tectonic CENA - PIE ASK

21 /5/4 R- Dependent and Comparisons T. sec.2.8 Per.- M3 to6 CENA - Tectonic CENA - PIE ASK4.2.8 Per.- M3 to5 CENA - Tectonic CENA - PIE ASK Distance (km) Distance (km) 4 R- Dependent and Comparisons T. sec.2.8 Per.- M3 to6 CENA - Tectonic CENA - PIE ASK4.2.8 Per.- M3 to5 CENA - Tectonic CENA - PIE ASK Distance (km) Distance (km) 42 2

22 /5/4 MR- Dependent Comparisons Per.- M3 to4 CENA - Tectonic CENA - PIE ASK4 Euro T. sec Per.- M4 to5 CENA - Tectonic CENA - PIE ASK4 Taiw Euro Distance (km) Per.- M5 to Distance (km) CENA - Tectonic CENA - PIE ASK4 Taiw Euro Distance (km) 43 MR- Dependent Comparisons Per.- M3 to4 CENA - Tectonic CENA - PIE ASK4 Euro T. sec Per.- M4 to5 CENA - Tectonic CENA - PIE ASK4 Taiw Euro Distance (km) Per.- M5 to Distance (km) CENA - Tectonic CENA - PIE ASK4 Taiw Euro Distance (km) 44 22

23 /5/4 EvaluaIon of Path Effects for PIE Closeness index (CI) is used as an indicator of path similariies CI i, j, k = ( R + R ) 2 i, k ΔH i, j j, k H i,j = distance between the hypocenters of eqk i and j recorded at site s R i,k and R j,k = hypocentral distances from site k to eqk i and j. CI ranges from for co- located eqks to 2 for eqks located in opposite epicentral direcfons from the site 45 Path Effects for PIE: Closeness Index Calculate expected CI for randomly located earthquakes Calculate average CI for PIE Compare average CI for PIE to expected CI for randomly located earthquakes 46 23

24 /5/4 Expected CI for Randomly Located Eqks. Generate a number of randomly located earthquakes (Nhyp) in a circular area of radius R around a staion 2. Calculate CI for each pair of earthquakes at the staion 3. Calculate mean CI 4. Vary Nhypo and R Expected CI for randomly located earthquakes Closeness Index Region 2 δws Δ = i, j, k ik δws 2ϕ ss jk Region2 - T. Tectonic - Avg CI :.4 PIE - Avg CI :.55 Δ Closeness index, CI CI i, j, k = ( R + R ) 2 i, k ΔH i, j j, k Expected CI for randomly located earthquakes

25 /5/4 Closeness Index Region 3 and Region3 - T. Avg CI Region5 - T. Avg CI Δ Δ Closeness index, CI Closeness index, CI δws Δ = i, j, k ik δws 2ϕ ss jk CI i, j, k = ( R + R ) 2 i, k ΔH i, j j, k 49 Closeness Index; All Data Binned.4 T..4 T σ Δ σ Δ Lin et al. (2) CI for randomly located earthquakes:.38.2 Mean CI from data:.4 Data Closeness index, CI.4 Lin et al. (2) CI for randomly located earthquakes:.38.2 Mean CI from data:.98 Data Closeness index, CI σ Δ = mean 2 ( Δ ) At CI = à à At CI = 2 à à ss 2 ( ) ( ( 2 2 )) δ = Var δws δws E δws δws φ ρ δp P δp P φ 2 = = ijk ik jk ik jk P2P ik jk 5 25

26 /5/4 CANDIDATE MODELS FOR CENA: - New preliminary candidate models for Tau,, and - Exis-ng models for Tau,,, and ergodic Sigma 5 Proposed Tau Models Preliminary new models derived using tectonic CENA data Constant (M- independent) CENA model Period- dependence M- dependent CENA model: ExtrapolaIon beyond M4.5 ExisIng Tau models: EPRI 23: M- dependent based on avg of preliminary 4 NGA- W2 Tau values; Increased by

27 /5/4 Constant CENA Tau Model.9.8 CENA - Mmin 3. CENA - Mmin Tau Used data with Mmin 3. to derive Tau model (45 versus 2 earthquakes) 53 Constant CENA Tau Model (cont d) CENA Zhao6 BSSA4 - M 4. CB4 - M 4. CY4 - M 4. ASK4 - M 4. Tau

28 /5/4 M- Dependent CENA Tau Model T.sec CENA BSSA4 CB4 CY4 ASK T.sec CENA BSSA4 CB4 CY4 ASK4.6.6 Tau.5 Tau For M < 4.5; Tau is mean of values for M 3. to 4.5 bins 2 opions for M > 4.5 extrapolaion: Same slope as CB4 Tau from M 4.5 to 5.5 Tau reaches the same CB4 Tau for M >= M- Dependent CENA Tau Model (cont d) T.5sec CENA BSSA4 CB4 CY4 ASK Tsec CENA BSSA4 CB4 CY4 ASK4.6.6 Tau.5 Tau For M < 4.5; Tau is mean of values for M 3. to 4.5 bins 2 opions for M > 4.5 extrapolaion: Same slope as CB4 Tau from M 4.5 to 5.5 Tau reaches the same CB4 Tau for M >=

29 /5/4 EPRI 23 Tau Model.9.8 EPRI 23 - M5 EPRI 23 - M6 EPRI 23 - M7.7.6 Tau Comparison of Candidate Tau Models Tau

30 /5/4 Proposed Models Preliminary models derived using tectonic data with distances less than 4 km values affected by site classificaion issues Two new derived models: Constant M- dependent and issue of extrapolaion to large M ExisIng models: EPRI 23: M- dependent based on avg of preliminary 4 NGA- W2 values 59 Constant CENA Model.8 Mmin 3. Rmax :,9 recs Mmin 3. Rmax 4: 65 recs Mmin 4. Rmax : 727 recs Mmin 4. Rmax 4: 25 recs.6.4 Mmin 3. - Rmax.2 Mmin 3. - Rmax 4 Mmin 4. - Rmax Mmin 4. - Rmax 4.. Used data with Mmin 3. and Rmax 4 km to derive model 6 3

31 /5/4 Constant CENA Model (cont d) CENA BSSA4 - M4. - R25 - Vs76.2 CB4 - M4. - Vs76. CY4 - M4. - Vs76 meas ASK4 - M4. - Vs76 meas.. 6 M- Dependent CENA Model T.sec T.sec CENA - Rmax 4 BSSA4 - R25 - Vs76.2 CB4 - Vs76 CY4 - Vs76 meas ASK4 - Vs76 meas.4 CENA - Rmax 4 BSSA4 - R25 - Vs76.2 CB4 - Vs76 CY4 - Vs76 meas ASK4 - Vs76 meas For M < 4.5; is mean of values for M 3. to 4.5 bins 2 opions for M > 4.5 extrapolaion: Same slope as BSSA4 from M 4.5 to 5.5 reaches the same BSSA4 for M >=

32 /5/4 M- Dependent CENA Model (cont d) T.5sec Tsec CENA - Rmax 4 BSSA4 - R25 - Vs76.2 CB4 - Vs76 CY4 - Vs76 meas ASK4 - Vs76 meas.4 CENA - Rmax 4 BSSA4 - R25 - Vs76.2 CB4 - Vs76 CY4 - Vs76 meas ASK4 - Vs76 meas For M < 4.5; is mean of values for M 3. to 4.5 bins 2 opions for M > 4.5 extrapolaion: Same slope as BSSA4 from M 4.5 to 5.5 reaches the same BSSA4 for M >= EPRI 23 Model EPRI 23 - M 5. EPRI 23 - M 6 EPRI 23 - M

33 /5/4 Comparison of Candidate Models EPRI 23 - M 5.4 EPRI 23 - M 6 EPRI 23 - M 7.3 New - Constant New - MDep - M5 - Option.2 New - MDep - M5 - Option 2. New - MDep - M > Option New - MDep - M> Option ExisIng Ergodic Sigma Models EPRI 23: and Tau are based on NGA- West2 standard deviaions Atkinson and Adams 23: NaIonal seismic hazard maps of Canada Atkinson and Boore 26 Toro et al

34 /5/4 ExisIng Ergodic Sigma Models 67 Proposed Models Preliminary models derived using tectonic data with distances less than 4 km Two new derived models: Constant; M- dependent ExisIng models: Constant: PRP SWUS Project Italian Model (Luzi et al. 24) M- dependent: Thyspunt Project South Africa Hanford Project SWUS Project R- dependent: PRP MR- dependent: PRP 68 34

35 /5/4 Overview of ExisIng Models PRP: Global dataset (CA, Switzerland, Taiwan, Turkey, Japan), M >= 4.5, Rrup <= 2 km No regional dependence of Models: constant, R- dependent, MR- dependent Thyspunt Project and Hanford Project: Global dataset; PRP dataset M- dependent model 69 Overview of ExisIng Models (cont d) SWUS Project: DCPP: NGA- W2 data with M >= 5. and Rrup < 5 km 2 constant models (CA only and global) M- dependent model (CA only) PVNGS: Distant sources: constant model for M >= 5.5 and R = 2 to 4 km; NGA- W2 dataset Local sources: 2 constant models (NGA- W2 and Euro data of ASK3) with M >= 5. and Rrup < 5 km Luzi et al. (24): Italian data (3 sets) with M 4. to 7. and Rjb < 2 km Constant Models Others: BCHydro and Hanford Project for subducion earthquakes - Constant 7 35

36 /5/4 Constant CENA Model CENA - Mmin 3. - Rmax CENA - Mmin 3. - Rmax 4 CENA - Mmin 4. - Rmax CENA - Mmin 4. - Rmax 4 Mmin 3. Rmax :,468 recs Mmin 3. Rmax 4: 494 recs Mmin 4. Rmax : 53 recs Mmin 4. Rmax 4: 7 recs Used data with Mmin 3. and Rmax 4 km to derive model 7 Constant CENA Model (cont d).9.8 CENA - Mmin 3. - Rmax 4 ASK4 - Mmin 3. - Mmax 6. - Rmax

37 /5/4 ExisIng Constant Models CENA - Mmin 3. - Rmax 4 New Model PRP - Central PRP - Low PRP - High CENA - Mmin 3. - Rmax 4 New Model Italy, Italy, 2 Italy, PRP: Global data M >= 4.5, Rrup <= 2 km Italian data: M 4. to 7., RJB < 2 km 73 ExisIng Constant Models (cont d) DCPP CA (Blue): M >= 5.5, Rrup < 5 km DCPP Global (red): same M- R range PVNGS Distant Sources (Blue): M >= 5.5, Rrup 2 to 4 km PVNGS Short Distance European Data (red) M >= 5., Rrup < 5 km PVNGS Short Distance Global Data (green) 74 37

38 /5/4 M- dependent CENA Model Per. CENA - Rmax 4 ASK4 Model CENA Model ASK Per. CENA - Rmax 4 ASK4 Model CENA Model ASK M- dependent CENA Model (cont d) Per.5 CENA - Rmax 4 ASK4 Model CENA Model ASK Per. CENA - Rmax 4 ASK4 Model CENA Model ASK

39 /5/4 ExisIng M- dependent Models T. sec T. sec Hanford Model Black Global data; Rrup <= 2 km DCPP M- dependent Model Red CA NGA- W2 data; Rrup < 5 km 77 ExisIng R- dependent Model: PRP.9 Per. CENA - Mmin 4..9 Per. CENA - Mmin Distance (km) Distance (km) 78 39

40 /5/4 ExisIng MR- dependent Model: PRP Per.- M3 to4 Per.- M4 to Distance (km) Distance (km) Per.- M5 to6 Per.- M6 to Distance (km) Distance (km) 79 ExisIng MR- dependent Model: PRP Per.- M3 to4 Per.- M4 to Distance (km) Distance (km).8 Per.- M5 to6 CENA ASK4.8 Per.- M6 to Distance (km) Distance (km) 8 4

41 /5/4 Summary New CENA models suffer from biased dataset (small M, large distance) and potenial site classificaion issues Constant models may not be reasonable to apply for large M M- dependent models may do bemer Issue of extrapolaion to larger M 8 To Do New CENA models: Using residuals from a revised regression. Issue of site terms for Lowest useable period for TA data Q- specific for PIE data Use Nmin 3 and increase Rmax to maximize dataset. Incorporate PIE? More thorough evaluaion of exising models 82 4

42 /5/4 THANK YOU! 83 and S2S for Different Nmin Nmin Nmin3 Nmin5 Nmin Nmin Nmin3 Nmin5 Nmin S2S Nb of Recs Nb of Sta Nb of Eqks Nmin 3, Nmin 3 3, Nmin 5 3, Nmin,

43 /5/4 V s3 - Dependent and Comparisons Per.- M3 to6- Rmax35km Per.- M3 to6- Rmax35km CENA - Tectonic CENA - PIE ASK4 CENA - Tectonic CENA - PIE ASK Vs3 (m/sec) Vs3 (m/sec) 85 43