Difference between revisions of "UCVM ucvm with sw4 using cvmsi"

From SCECpedia
Jump to navigationJump to search
 
(14 intermediate revisions by 2 users not shown)
Line 1: Line 1:
−
== Followup Questions 11/2025 ==
+
== Questions 11/2025 ==
  
−
Region of interest, -118.4232,33.58, -117.3459,34.3519
 
  
 
LBNL TARGET:  
 
LBNL TARGET:  
Line 21: Line 20:
  
  
−
Solution: Using taper with -z 0,700 -L 200,700,1500, to mitigate the feature  
+
LBNL: Using taper with -z 0,700 -L 200,700,1500, to mitigate the feature  
  
 
{|
 
{|
Line 54: Line 53:
  
  
−
=== Adding tapering introduces discontinuities ===
+
Difference in GTL bounds for CVM-S4.26.M01 matches with CVM-S4
 +
 
 +
about density: https://strike.scec.org/scecpedia/CVM-S4.26#Density_Based_on_Vs
 +
 
 +
=== Adding elygtl:taper introduces discontinuities ===
  
 
just cvmsi
 
just cvmsi
Line 72: Line 75:
 
|}
 
|}
  
 +
extending z range with -z0,1000 and -L 200,700,1500
 +
 +
{|
 +
| [[FILE:cvmsi_vp_taper_z1000-200-700-1500.png|thumb|300px|cvmsi taper vp 700z]]
 +
| [[FILE:cvmsi_vs_taper_z1000-200-700-1500.png|thumb|300px|cvmsi taper vs 700z]]
 +
| [[FILE:cvmsi_density_taper_z1000-200-700-1500.png|thumb|300px|cvmsi taper density 700z]]
 +
|}
 +
 +
=== Difference in Vp and Vs GTL depth traced to CVM-S4 ===
 +
 +
We selected (33.95575, -117.50369), in the middle of the cross-sections above, and ran this point at a variety of depths through the CVM-S3 and CVM-S4 implementations from https://scedc.caltech.edu/data/3d-velocity.html .
 +
 +
In CVM-S version 3, we can see that there is a transition between 720 and 760m:
 +
<pre>
 +
  Lat      Lon        Depth(m)  Vp      Vs      Rho
 +
33.95575 -117.50369    120.00  1790.2    730.8  1795.2
 +
33.95575 -117.50369    160.00  1999.7    816.4  1888.2
 +
33.95575 -117.50369    200.00  2322.3    948.1  2031.5
 +
33.95575 -117.50369    240.00  2418.5  1000.0  2074.2
 +
33.95575 -117.50369    280.00  2505.4  1000.0  2129.9
 +
33.95575 -117.50369    320.00  2585.1  1169.1  2143.7
 +
33.95575 -117.50369    360.00  2659.2  1218.6  2156.6
 +
33.95575 -117.50369    400.00  2728.5  1265.4  2168.6
 +
33.95575 -117.50369    440.00  2794.0  1310.0  2179.9
 +
33.95575 -117.50369    480.00  2856.0  1352.5  2190.7
 +
33.95575 -117.50369    520.00  2915.1  1393.3  2200.9
 +
33.95575 -117.50369    560.00  2971.6  1432.5  2210.7
 +
33.95575 -117.50369    600.00  3025.8  1470.4  2220.1
 +
33.95575 -117.50369    640.00  3077.9  1507.0  2229.1
 +
33.95575 -117.50369    680.00  3128.2  1542.5  2237.9
 +
33.95575 -117.50369    720.00  3176.7  1577.0  2246.3
 +
33.95575 -117.50369    760.00  5903.3  3424.9  2718.9
 +
33.95575 -117.50369    800.00  5903.3  3424.9  2718.9
 +
</pre>
 +
 +
In CVM-S version 4, we see that the transition is now between 160 and 200m for Vp and rho, but between 280 and 320m for Vs:
  
−
about density: https://strike.scec.org/scecpedia/CVM-S4.26#Density_Based_on_Vs
+
<pre>
 +
  Lat      Lon        Depth(m)  Vp      Vs      Rho
 +
33.95575 -117.50369    120.00  2770.8    980.0  2302.5
 +
33.95575 -117.50369    160.00  4188.8  1000.0  2526.4
 +
33.95575 -117.50369    200.00  5903.2  1000.0  2797.1
 +
33.95575 -117.50369    240.00  5903.3  1000.0  2797.1
 +
33.95575 -117.50369    280.00  5903.3  1000.0  2797.1
 +
33.95575 -117.50369    320.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    360.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    400.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    440.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    480.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    520.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    560.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    600.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    640.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    680.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    720.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    760.00  5903.3  3424.9  2797.1
 +
33.95575 -117.50369    800.00  5903.3  3424.9  2797.1
 +
</pre>
  
−
=== Test adding tapering for cvmsi ===
+
The readme for CVM-S version 4 states that changes were made to the San Bernardino Valley, so perhaps that is what led to this change.
  
 
== Question 9/2025 ==
 
== Question 9/2025 ==
Line 97: Line 156:
 
Would you know of any literature or documentation that describe these characteristics?
 
Would you know of any literature or documentation that describe these characteristics?
  
−
== Background of the posted question ==
+
=== Background of the posted question ===
  
 
Accessing UCVM velocity model via SW4's UCVM reader branch and then run the SW4 simulation.
 
Accessing UCVM velocity model via SW4's UCVM reader branch and then run the SW4 simulation.
Line 109: Line 168:
 
Is this version of velocity model, recommended for ground motion simulation ? (another question)
 
Is this version of velocity model, recommended for ground motion simulation ? (another question)
  
−
 
+
=== Response from Scott ===
−
== Response from Scott ==
 
  
 
It looks like the blocky structure appears in the original CVM-S4 (https://strike.scec.org/scecpedia/UCVM_ucvm_with_sw4_using_cvmsi#Base_cvm, top row), and therefore also appears in CVM-S4.26 and CVM-S4.26.M01, both of which were derived from it.  I believe this is the northern edge of the LA basin as defined in the original CVM-S4.  When using this model for CyberShake Study 22.12 with the Ely taper, we didn't see this sharp east-west boundary.  You can see our cross sections at https://strike.scec.org/scecpedia/CyberShake_Study_22.12#Cross-sections, in the 3rd column.  However, we ran with a minimum Vs of 500 m/s and 100m grid spacing.
 
It looks like the blocky structure appears in the original CVM-S4 (https://strike.scec.org/scecpedia/UCVM_ucvm_with_sw4_using_cvmsi#Base_cvm, top row), and therefore also appears in CVM-S4.26 and CVM-S4.26.M01, both of which were derived from it.  I believe this is the northern edge of the LA basin as defined in the original CVM-S4.  When using this model for CyberShake Study 22.12 with the Ely taper, we didn't see this sharp east-west boundary.  You can see our cross sections at https://strike.scec.org/scecpedia/CyberShake_Study_22.12#Cross-sections, in the 3rd column.  However, we ran with a minimum Vs of 500 m/s and 100m grid spacing.
Line 122: Line 180:
 
Please let me know if you have follow-up questions.  We can also discuss modifications to the model if you'd like to avoid some of these features; for example, I know Hu et al. (2022) [https://doi.org/10.1093/gji/ggac175] used an alternative method for deciding where to apply the taper.
 
Please let me know if you have follow-up questions.  We can also discuss modifications to the model if you'd like to avoid some of these features; for example, I know Hu et al. (2022) [https://doi.org/10.1093/gji/ggac175] used an alternative method for deciding where to apply the taper.
  
−
== Plots from explorer with the posted parameters ==
+
=== Plots from explorer with the posted parameters ===
  
 
{|
 
{|
Line 158: Line 216:
 
|}
 
|}
  
−
== Base cvm ==
+
=== Base cvm ===
  
 
{|
 
{|
Line 186: Line 244:
 
|}
 
|}
  
−
== using elygtl:ely ==
+
=== using elygtl:ely ===
  
 
{|
 
{|
 
| [[FILE:sw4_cvmsi_ely0_0m.png|thumb|300px|cvmsi ely0 0m]]
 
| [[FILE:sw4_cvmsi_ely0_0m.png|thumb|300px|cvmsi ely0 0m]]
 
|}
 
|}

Latest revision as of 22:39, 6 November 2025

Questions 11/2025

LBNL TARGET: Performing 5 Hz simulations based on CVM.S4.26.M01 and apply additional GTL with the Ely tapering method. For high frequency simulation requirements, the velocity model needs some additional modifications.

What causes the Ripple-like pattern (bullseyes) in the cvmsi?

 Region of interest: -118.4232,33.58, -117.3459,34.3519

Due to local borehole information that are in the CVMS(rule-based) with its GTL processing

cvms bullseyes vs
cvms5 bullseyes vs
cvmsi bullseyes vs


LBNL: Using taper with -z 0,700 -L 200,700,1500, to mitigate the feature

cvmsi taper vs

Transition depth to high velocity differs between Vp(~400m) and Vs(~250)

LBNL has a variable grid size/spacing based on depth: 50m grid for 0-650m depth, then 100m grid for 650-2500m, 200m grid for 2500-6000m, 400m grid for 6000m and below. The horizontal and vertical spacing is the same within the same depth range. 

Region of interest:   33.599659 -117.505759, 34.311847 -117.501624
cvms transition vp
cvms transition vs
cvms transition density
cvms5 transition vp
cvms5 transition vs
cvms5 transition density
cvmsi transition vp
cvmsi transition vs
cvmsi transition density


Difference in GTL bounds for CVM-S4.26.M01 matches with CVM-S4

about density: https://strike.scec.org/scecpedia/CVM-S4.26#Density_Based_on_Vs

Adding elygtl:taper introduces discontinuities

just cvmsi

cvmsi vp
cvmsi vs
cvmsi density

cvmsi with -z0,700 and -L 200,700,1500

cvmsi taper vp 700z
cvmsi taper vs 700z
cvmsi taper density 700z

extending z range with -z0,1000 and -L 200,700,1500

cvmsi taper vp 700z
cvmsi taper vs 700z
cvmsi taper density 700z

Difference in Vp and Vs GTL depth traced to CVM-S4

We selected (33.95575, -117.50369), in the middle of the cross-sections above, and ran this point at a variety of depths through the CVM-S3 and CVM-S4 implementations from https://scedc.caltech.edu/data/3d-velocity.html .

In CVM-S version 3, we can see that there is a transition between 720 and 760m: 

  Lat      Lon         Depth(m)   Vp       Vs      Rho
33.95575 -117.50369    120.00   1790.2    730.8   1795.2
33.95575 -117.50369    160.00   1999.7    816.4   1888.2
33.95575 -117.50369    200.00   2322.3    948.1   2031.5
33.95575 -117.50369    240.00   2418.5   1000.0   2074.2
33.95575 -117.50369    280.00   2505.4   1000.0   2129.9
33.95575 -117.50369    320.00   2585.1   1169.1   2143.7
33.95575 -117.50369    360.00   2659.2   1218.6   2156.6
33.95575 -117.50369    400.00   2728.5   1265.4   2168.6
33.95575 -117.50369    440.00   2794.0   1310.0   2179.9
33.95575 -117.50369    480.00   2856.0   1352.5   2190.7
33.95575 -117.50369    520.00   2915.1   1393.3   2200.9
33.95575 -117.50369    560.00   2971.6   1432.5   2210.7
33.95575 -117.50369    600.00   3025.8   1470.4   2220.1
33.95575 -117.50369    640.00   3077.9   1507.0   2229.1
33.95575 -117.50369    680.00   3128.2   1542.5   2237.9
33.95575 -117.50369    720.00   3176.7   1577.0   2246.3
33.95575 -117.50369    760.00   5903.3   3424.9   2718.9
33.95575 -117.50369    800.00   5903.3   3424.9   2718.9

In CVM-S version 4, we see that the transition is now between 160 and 200m for Vp and rho, but between 280 and 320m for Vs: 

  Lat      Lon         Depth(m)   Vp       Vs      Rho
33.95575 -117.50369    120.00   2770.8    980.0   2302.5
33.95575 -117.50369    160.00   4188.8   1000.0   2526.4
33.95575 -117.50369    200.00   5903.2   1000.0   2797.1
33.95575 -117.50369    240.00   5903.3   1000.0   2797.1
33.95575 -117.50369    280.00   5903.3   1000.0   2797.1
33.95575 -117.50369    320.00   5903.3   3424.9   2797.1
33.95575 -117.50369    360.00   5903.3   3424.9   2797.1
33.95575 -117.50369    400.00   5903.3   3424.9   2797.1
33.95575 -117.50369    440.00   5903.3   3424.9   2797.1
33.95575 -117.50369    480.00   5903.3   3424.9   2797.1
33.95575 -117.50369    520.00   5903.3   3424.9   2797.1
33.95575 -117.50369    560.00   5903.3   3424.9   2797.1
33.95575 -117.50369    600.00   5903.3   3424.9   2797.1
33.95575 -117.50369    640.00   5903.3   3424.9   2797.1
33.95575 -117.50369    680.00   5903.3   3424.9   2797.1
33.95575 -117.50369    720.00   5903.3   3424.9   2797.1
33.95575 -117.50369    760.00   5903.3   3424.9   2797.1
33.95575 -117.50369    800.00   5903.3   3424.9   2797.1

The readme for CVM-S version 4 states that changes were made to the San Bernardino Valley, so perhaps that is what led to this change.

Question 9/2025

Question from Rie to Scott,

..conducting SW4 simulations within the LA Basin for both small and large magnitude earthquake events (please see the attached slides for reference where we are interested in). While validating the ground motions, we have come across several features in the velocity model—built from the SCEC model—that we were hoping to better understand with your insight.

Blocky structure north of LA (highlighted in red in the slides): 

 We observed a blocky structure characterized by elevated Vs compared to the surrounding region. This structure appears to disappear at 50 m depth and then re-emerges below. Is there a specific reason why this structure is not present at the 50 m depth level?

Vp–Vs relationship of the blocky structure (green color): 

 The structure shows higher Vs values, with Vp appearing higher at the surface (Z = 0 m) but becoming lower at 250 m depth. We also noted that the density is relatively low at the surface.

Vp–density relationship west of the blocky structure: 

 In this nearby region, the density is higher than that of the surrounding material, but Vp appears to be lower. (I expect Vp correlates with density)

Would you know of any literature or documentation that describe these characteristics?

Background of the posted question

Accessing UCVM velocity model via SW4's UCVM reader branch and then run the SW4 simulation. 

 CVM-S4.26.M01 (cvmsi), the velocity values are extracted using UCVM "withSCPBR" branch

internal query parameters were: 

 ucvm_query -f conf/ucvm.conf -m cvmsi,elygtl:taper -L 200,700,1500

Is this version of velocity model, recommended for ground motion simulation ? (another question)

Response from Scott

It looks like the blocky structure appears in the original CVM-S4 (https://strike.scec.org/scecpedia/UCVM_ucvm_with_sw4_using_cvmsi#Base_cvm, top row), and therefore also appears in CVM-S4.26 and CVM-S4.26.M01, both of which were derived from it. I believe this is the northern edge of the LA basin as defined in the original CVM-S4. When using this model for CyberShake Study 22.12 with the Ely taper, we didn't see this sharp east-west boundary. You can see our cross sections at https://strike.scec.org/scecpedia/CyberShake_Study_22.12#Cross-sections, in the 3rd column. However, we ran with a minimum Vs of 500 m/s and 100m grid spacing.

For question 1), I think this has to do with the way the merged taper is applied. The taper is constrained by the Thompson et al. (2022) Vs30 values, but every mesh point is evaluated when determining if to apply the taper or not. In other words, at (34.0, -118.0, 0m), the Vs values are calculated with and without the taper, and the method with the lowest Vs is selected. But then this process is repeated at 50m, 100m, etc., so at some depths the taper might be selected and at others not. I pulled Vs profiles of the top 500m for (-118, 34.1), which is in the dark blue region; (-118, 34.2), in the teal region; and (-118, 34.3), in the lime green region (attached as taper 34_1.png, taper 34_2.png, taper 34_3.png). From the smoothness of the profiles, it looks like the taper is always used for 34.1 and 34.3, but not always for 34.2. As a result, sometimes the points in the teal region are closer to the lime green region, meaning the blocky structure disappears, and sometimes they are closer to the dark blue region, meaning the blocky structure reappears.

For 2), it looks to me like at the surface Vp~1100 m/s and Vs~400 m/s, and at 250m depth Vp~4500 m/s and Vs~1800 m/s, since the color bars change dramatically from z=0 to z=250m.

For 3), I extracted velocity profiles with and without the merged taper. In the region you highlighted, even without the taper the Vp value hits the floor at 700 m/s, but density is about 2000 km/m3 (see attached "all props no taper.png"). This is also true in the original CVM-S4 model. So this isn't a result of the taper, but rather a native feature. Since I wasn't involved in the original model construction, I'm not sure what the logic is for these values, but they do seem to be what was intended in CVM-S4 and its derivatives.

Please let me know if you have follow-up questions. We can also discuss modifications to the model if you'd like to avoid some of these features; for example, I know Hu et al. (2022) [1] used an alternative method for deciding where to apply the taper.

Plots from explorer with the posted parameters

cvmsi taper200 map
cvmsi taper200 0m
cvmsi taper200 50m
cvmsi taper200 100m
cvmsi taper200 250m
cvmsi taper200 line1
cvmsi taper200 line1
cvmsi taper200 line2
cvmsi taper200 line2
cvmsi taper200 line3
cvmsi taper200 line3
cvmsi taper200 pt
cvmsi taper200 pt1
cvmsi taper200 pt2
cvmsi taper200 pt3
cvmsi taper200 pt4

Base cvm

cvms 0m
cvms5 0m
cvmsi 0m
cvmsi 50m
cvmsi 100m
cvmsi 250m
cvmsi all 0m
cvmsi all 50m
cvmsi all 100m
cvmsi all 250m

looking at basin,

cvmsi basin a
cvmsi basin b

using elygtl:ely

cvmsi ely0 0m
Retrieved from "https://strike.scec.org/scecwiki/index.php?title=UCVM_ucvm_with_sw4_using_cvmsi&oldid=30574"