CyberShake Study 24.8 - Revision history

Scottcal: /* Data Products */

2025-10-01T16:47:03Z

Data Products

Scottcal: /* Lessons Learned */

2025-08-13T18:42:41Z

Lessons Learned

Scottcal: /* Background model */

2025-08-05T17:11:52Z

Background model

Scottcal: /* Background model */

2025-08-05T05:30:07Z

Background model

Scottcal: /* Lessons Learned */

2025-06-04T20:43:02Z

Lessons Learned

Scottcal: /* Validation */

2025-05-13T05:03:13Z

Validation

Scottcal at 05:02, 13 May 2025

2025-05-13T05:02:50Z

Scottcal: /* Performance Metrics */

2025-04-25T19:47:17Z

Performance Metrics

Scottcal: /* Performance Metrics */

2025-04-21T19:17:52Z

Performance Metrics

Scottcal: /* Performance Metrics */

2025-02-26T19:54:23Z

Performance Metrics

← Older revision		Revision as of 16:47, 1 October 2025
Line 18:		Line 18:

	Broadband GMM comparisons are available here: [https://opensha.usc.edu/ftp/kmilner/markdown/cybershake-analysis/study_24_8_bb/gmpe_comparisons_NGAWest_2014_NoIdr_Vs30Thompson2020/ Broadband GMM Comparisons].		Broadband GMM comparisons are available here: [https://opensha.usc.edu/ftp/kmilner/markdown/cybershake-analysis/study_24_8_bb/gmpe_comparisons_NGAWest_2014_NoIdr_Vs30Thompson2020/ Broadband GMM Comparisons].
		+
		+	Updated statewide maps using Study 24.8 and Study 22.12 results are available here: [https://opensha.usc.edu/ftp/kmilner/markdown/cybershake-analysis/study_24_8_lf/multi_study_hazard_maps/resources/ 2025 Statewide Maps].

	== Science Goals ==		== Science Goals ==

@@ Line 1,366: / Line 1,366: @@
 == Lessons Learned ==
-*The SRFs used in this study weren't quite what was intended; they had subfault relocation on and used a different risetime.  Consider running an SRF check before starting the study.
+*The SRFs used in this study weren't quite what was intended; they had subfault relocation on and used a different risetime.  Consider running an SRF check before starting the study.  More details about the subfault perturbation issue are given on the [[CyberShake_Study_22.12#Subfault_Perturbation | Study 22.12 wiki page]].
 *The runtime requested for post-processing jobs was higher than what was needed, increasing queue times for awhile.
 *The scottcal account went over quota on /home at Frontera, an issue for writing rvGAHP files.

@@ Line 219: / Line 219: @@
 === Background model ===
-<b>For Study 24.8, we decided to go with the [[NC1D]] model, which extends the Sierra region of SFCVM.</b>
+<b>For Study 24.8, we decided to go with the [[Sierra Foothills 1D]] model, which extends the Sierra region of SFCVM.</b>
 There are several candidates to use as a background model for the regions outside of the 3D model region.

← Older revision		Revision as of 05:30, 5 August 2025
Line 218:		Line 218:

	=== Background model ===		=== Background model ===
		+
		+	<b>For Study 24.8, we decided to go with the [[NC1D]] model, which extends the Sierra region of SFCVM.</b>

	There are several candidates to use as a background model for the regions outside of the 3D model region.		There are several candidates to use as a background model for the regions outside of the 3D model region.

@@ Line 1,368: / Line 1,368: @@
 *The scottcal account went over quota on /home at Frontera, an issue for writing rvGAHP files.
 *Due to the lack of support for outgoing connections, there was not an easy way to bundle jobs on Frontier and we ran the SGT jobs individually.
 == Stress Test ==

@@ Line 707: / Line 707: @@
 == Validation ==
-We performed [[validation runs for the Loma Prieta BBP event | Loma Prieta Validation]]
+We performed [[Loma Prieta Validation | validation runs for the Loma Prieta BBP event]].
 == Hazard Curve Tests ==

← Older revision		Revision as of 05:02, 13 May 2025
Line 704:		Line 704:

	However, since we are using a denser mesh (80m) and a lower minimum Vs (400 m/s), we will not apply site correction to the low-frequency seismograms before combining. Rob states, <i>You may recall that there is a klugy process we have used to estimate the Vref value based on the Vsmin and grid spacing of the model. But, it has only been applied for the case of h=100 m and Vsmin=500 m/s. What I found here is that estimating Vref using the 80m & 400m/s model is that Vref is almost always less than Vs30 (i.e., Vsite). This means that when the site adjustment is applied, the motions are deamplified. This is why we see such a poor fit for the 3D case when using the estimated Vref values. My conclusion at this point is that if we run the 3D calculation with 80 m grid spacing and Vsmin of 400 m/s (or lower), then we probably do not need to apply any site adjustments.</i>		However, since we are using a denser mesh (80m) and a lower minimum Vs (400 m/s), we will not apply site correction to the low-frequency seismograms before combining. Rob states, <i>You may recall that there is a klugy process we have used to estimate the Vref value based on the Vsmin and grid spacing of the model. But, it has only been applied for the case of h=100 m and Vsmin=500 m/s. What I found here is that estimating Vref using the 80m & 400m/s model is that Vref is almost always less than Vs30 (i.e., Vsite). This means that when the site adjustment is applied, the motions are deamplified. This is why we see such a poor fit for the 3D case when using the estimated Vref values. My conclusion at this point is that if we run the 3D calculation with 80 m grid spacing and Vsmin of 400 m/s (or lower), then we probably do not need to apply any site adjustments.</i>
		+
		+	== Validation ==
		+
		+	We performed [[validation runs for the Loma Prieta BBP event \| Loma Prieta Validation]]

	== Hazard Curve Tests ==		== Hazard Curve Tests ==

@@ Line 1,710: / Line 1,710: @@
 At the end of the main study, project geo156 had used 671,656 node-hours.  User callag has used 88881.2 node-hours.  project EAR20006 had 511480 node-hours remaining (out of 676,000).  User scottcal used 125411.21 node-hours.
-In total, the study used 55483 node-hours on Frontier and 124907 node-hours on Frontera.
+In total, the study used 55483 node-hours on Frontier and 124907 node-hours on Frontera.  This is difficult to translate into core-hours.  Each Frontier node has 64 cores plus 8 GPUs.  Each GPU has 110 compute units, so we will assume each node has 64 + 110*8 = 944 'cores'.  Each Frontera node has 56 cores.  We translate this to 52.4M core-hours on Frontier and 7.0M core-hours on Frontera.
 The study ran from 9/24/24 11:49:59 PDT to 11/8/24 6:15:15 PST, a duration of 1075.4 hrs (44.8 days).

← Older revision		Revision as of 19:17, 21 April 2025
Line 1,727:		Line 1,727:

	126,818,608 three-component seismograms (half LF, half BB), 11.5 billion low-frequency IMs, and 22.8 billion broadband IMs were calculated.		126,818,608 three-component seismograms (half LF, half BB), 11.5 billion low-frequency IMs, and 22.8 billion broadband IMs were calculated.
		+
		+	Total number of rupture variations is 63,409,304.

	== Production Checklist ==		== Production Checklist ==

← Older revision		Revision as of 19:54, 26 February 2025
Line 1,712:		Line 1,712:
	In total, the study used 55483 node-hours on Frontier and 124907 node-hours on Frontera.		In total, the study used 55483 node-hours on Frontier and 124907 node-hours on Frontera.

−	The study ran from 11/8/24 at 6:15:15 PST.	+	The study ran from 9/24/24 11:49:59 PDT to 11/8/24 6:15:15 PST, a duration of 1075.4 hrs (44.8 days).