ML24305A187

OEDO-24-00083 10 CFR 2.206 - Diablo Canyon Units 1 and 2 Seismic CDF - Petition Supplement 10-30-2024

ML24305A187
Person / Time
Site:	Diablo Canyon
Issue date:	10/31/2024
From:	Bird P Environmental Working Group, Friends of the Earth, San Luis Obispo Mothers for Peace (SLOMFP), Univ of California - Los Angeles
To:	Gerfen P Plant Licensing Branch IV, Pacific Gas & Electric Co
Lee S, 301-415-3158
References
Download: ML24305A187 (1)

Text

1 High Seismic Hazard and Risk at Diablo Canyon Power Plant Due to Thrust Faulting Under the Irish Hills (a report prepared at the request of the Diablo Canyon Independent Safety Committee of the California Public Utilities Commission, summarizing arguments, data, and citations previously submitted to the U.S. Nuclear Regulatory Commission in 3 Declarations and 1 oral presentation, 2023-2024) by Peter Bird, Professor Emeritus Department of Earth, Planetary, and Space Sciences University of California Los Angeles, consulting to:

San Luis Obispo Mothers for Peace, Friends of the Earth, and Environmental Working Group Los Angeles 30 October 2024

2 ABSTRACT The crustal basement under Diablo Canyon Power Plant (DCPP) is composed of exotic Franciscan Complex and arc-derived Cretaceous turbidite sandstones, assembled into the accretionary prism of the Farallon\\North America subduction zone by systemic and continuous thrust faulting in Cretaceous through Paleogene times.

The Irish Hills are a new isostatically-supported fold/thrust belt created by horizontal crustal shortening since 6~5 Ma. Evidence that the same stress field continues today includes small thrust earthquakes under the Irish Hills, larger thrust earthquakes in the region, stress direction data, and GPS geodesy.

However, the Seismic Source Characterizations (SSCs) by Pacific Gas & Electric Company

[PG&E, 2015; 2024] concluded that hazard at DCPP from 2 modeled thrust faults and the Local Area Source is less than hazard from the offshore dextral Hosgri fault. Potential hazard from blind thrust faults is essentially dismissed. This biased conclusion is due to their reliance on 4 false assumptions, which I challenge and then correct. Then, I propose 3 independent methods of estimating the total slip-rate of all shallow-dipping thrust faults under the Irish Hills; these methods agree on a range of 2.0~2.8 mm/a.

The simplest and most expedient way to estimate the seismic hazard from such fast and widespread thrust faulting is to choose a model characteristic thrust earthquake and then to compute its recurrence rate. Based on tectonic and geometric similarities, I choose the 1 January 2024 thrust earthquake under the Noto Peninsula in Japan. This event produced Peak Ground Acceleration (PGA) of 100% to 230% of gravity at 5 digital strong-motion seismometers up to 42 km from the rupture.

According to the Seismic Probabilistic Risk Assessment (SPRA) by PG&E [2018], such PGA values (and the associated higher spectral accelerations) would likely cause a Seismic Core Damage (SCD) accident (i.e., meltdown) at DCPP. Therefore, SCD Frequency (SCDF) at DCPP is roughly the same as the recurrence rate of the model characteristic earthquake, with is from 1.0x10-3 to 1.4x10-3 per year of plant operation.

Appendix A details other procedural errors in the SSCs of PG&E [2015; 2024], specifically the omission of modern deformation modeling, and failure to use globally-calibrated strainrate-to-seismicity conversions, that would have prevented their serious underestimates.

Appendix B responds to an unfounded assertion in PG&Es recent letter to NRC regarding the use of the Noto Peninsula earthquake as a model characteristic thrust earthquake for the Irish Hills.

3 I. INTRODUCTION During Cretaceous through Paleogene times, the coast of California was the site of a subduction zone which consumed thousands of kilometers of oceanic Farallon plate lithosphere and created the North American volcanic arc, of which the Sierra Nevada plutons are the remaining roots

[e.g., Atwater & Stock, 1998]. Just east of the former Farallon Trench, there was a wide accretionary prism in constant overturning motion [Cloos, 1982] which was fed from both sides.

Exotic oceanic rocks (limestones, bedded cherts, basalts, serpentines) were scraped off the subducting Farallon plate and added to North America. At the same time, voluminous graywacke turbidite sands, derived from the North American volcanic arc, were deposited in the trench, and then quickly scraped off the subducting Farallon plate and re-incorporated into North American crust [e.g., Wakabayashi, 1999]. This process of systemic and continuous thrust-faulting formed the crustal basement on which DCPP now sits.

The present Irish Hills and the San Luis Range are a younger dextral-transpressional orogen that has formed since ~3.5 million years (or mega annus, Ma) [Page et al., 1998], or more likely since 6 Ma [Austermann et al., 2011; Bird & Ingersoll, 2022] when the motion of the Pacific plate changed its direction to become more compressional relative to North America. This means that the region can be expected to be cut by a number of both strike-slip and thrust (horizontally compressional) faults.

The persistence of a horizontally compressive thrust-faulting stress regime up to the present is shown by:

(1) The 2003 San Simeon m6.6 and 1983 Coalinga m6.2 earthquake both had thrust mechanisms

[Global Centroid Moment Tensor Catalog, Ekstrm et al., 2012]. This is evidence of highly-compressive horizontal stresses in the Coast Ranges region, suggesting a likelihood of seismic thrust-faulting in other locations as well.

(2) Closer to DCPP, two recent small earthquakes had thrust-faulting mechanisms with the expected SSW-NNE direction of maximum horizontal compression: 2023.12.27 m3.1 at 6.2 km depth under the Irish Hills, and 2024.01.01 m5.4 slightly offshore from the NW end of the Irish Hills (D. J. Weisman, pers. comm., 2024.01.02). This shows that the regional stress regime and orientation documented above also apply in the immediate vicinity of DCPP.

(3) SSW-NNE directions of most-compressive stress shown by data in the World Stress Map

[Mueller et al., 1997; Heidbach et al., 2008, 2016], and by interpolation of stress directions using the method of Bird & Li [1996], are almost perpendicular to the traces of the regional thrust fault trend (Inferred Coastline, San Luis Bay, and Los Osos fault traces). This strongly suggests that currently these faults are either purely or dominantly thrusts, and that strike-slip is occurring only on the offshore Hosgri and perhaps Shoreline faults.

Given this geologic history and structure, one might expect to read that thrust faulting under the Irish Hills is the major source of seismic hazard to DCPP. However, in PG&Es [2015] SSC, the Los Osos and San Luis Bay thrust faults have seismic hazard contributions (specifically, recurrence rates of PGA over 1 g and spectral accelerations over 2 g which would cause core

4 damage) adding up to less than the hazard from the offshore strike-slip Hosgri fault, and consequently less than half of the total hazard. This imbalance remains uncorrected in the SSC of PG&E [2024].

II. CORRECTING 4 FALSE ASSUMPTIONS in the SSCs of PG&E [2015; 2024]

One of the reasons that PG&Es [2015; 2024] SSCs seriously underestimated seismic hazard due to thrust faulting is that they were guided by 4 assumptions that are actually false.

1. 1. The Irish Hills are uplifting as a rigid block, with no internal deformation.

Therefore, PG&E concluded that thrust faulting occurs only at the margins (Los Osos thrust, San Luis Bay thrust) with fault throw (vertical offset) rates equal to marine terrace uplift rates of ~0.2 mm/year, and that thrust faulting occurs nowhere else.

HOWEVER:

The geologic map (Figure 1) shows tight folding of Late Miocene sedimentary rocks has occurred since 6~5 Ma. Therefore, the Irish Hills are not rigid, and additional blind thrust faults are active in the interior.

The Pismo syncline is the primary structural feature exposed in the Irish Hills [PG&E, 2014].

Here beds have been rotated ~45, which angle is supported by both mapped surface dips in outcrops (geologic map, ibid), and by the overall dip of unit Tmo Tertiary Miocene Obispo Formation in the borehole-controlled cross-section of Figure 13-17 of the PG&E [2015] SSC.

This folding began after deposition of the youngest strata in the core of the fold (Tmpm), and prior to deposition of the Squire Member of the (Pliocene) Pismo Formation (Tpps), probably ~5 Ma. This folding implies upper-crustal strains of ~0.8, and mean strain-rates of ~0.8 / 5 Ma =

5x10-15 per second (/s). This is ~10x faster than rates of off-modeled-fault (or continuum) deformation that are typical in the long-term neotectonics of the western US [5x10-16 /s per Bird, 2009]. This high rate of permanent straining implies a high rate of faulting and of earthquakes, even if the relevant thrust fault traces are not always exposed.

Also, rigid-body uplift would not produce crustal thickening. Therefore, if the Irish Hills were a rigid block, they would have a positive isostatic gravity anomaly. However, gravity data (Figure

2) shows a negative isostatic gravity anomaly, indicating more than simple Airy compensation by crustal roots (more than the typical Airy ratio of 6:1).

THEREFORE:

The Irish Hills are being deformed by numerous unmapped and/or blind thrust faults in addition to the 2 thrust faults modeled by PG&E. One prominent possibility (Figure 3) is the Inferred Coastline thrust fault (my term) passing offshore DCPP along the southwest coast of the Irish Hills, and dipping under the plant. The inferred location of this fault trace is based on: (1) the southwestern front of the Irish Hills is a topographic scarp with a smooth arcuate shape, mirroring the slightly-lower scarp on the northeast which has been formed by slip on the Los Osos thrust fault; (2) the documented uplift of marine terraces on the northeast side, relative to the continental shelf; (3) intense deformation of Miocene rocks in the plant area, which could be

5 due to a forced fold over a blind thrust fault tip; and (4) the tectonic implausibility of the San Luis Bay thrust fault simply terminating (just south of DCPP) without a connection to the Hosgri fault. (Actually, PG&E did suggest that the San Luis Bay thrust fault may terminate by merging with the Shoreline fault, but the slip rate of the Shoreline is so low that this is not plausible.)

The crustal basement under the folded sedimentary rocks of the Irish Hills is mainly Franciscan Complex, which contains numerous Cretaceous-Paleogene thrust faults available for reactivation. Slip on those thrust faults would not reach the surface (allowing for mapping) because such slip would encounter and fold the layered Neogene sedimentary rocks of the Pismo syncline. Thus, there are an unknown number of blind thrust faults active, such as those that produce devastating earthquakes under the Zagros Mountains of Iran, or in Nepal.

1. 2. Active thrust faults may dip at any angle.

PG&E assigned alternative model dips of 30, 50, and 80 for the Los Osos thrust fault, with a combined weight of 70% to the dips of 50 to 80 in their logic-tree. They also assigned alternative dips of 45 to 75 for the San Luis Bay thrust fault.

HOWEVER:

125-year-old Mohr/Coulomb friction theory shows that thrusts never form at dips steeper than 45, and most commonly dip at ~25 for rock friction coefficient of 0.85 [Byerlee, 1978; Figure 4]. This result comes from the simple formula:

(

)

1 1

(

)

arctan 2

thrust dip f

=

where f is the coefficient of friction. This equation is derived from classic Mohrs-circle analysis of shear stress and effective normal stress acting on all possible planes within a uniform rock material, which predicts the orientation of the plane(s) that should break first.

THEREFORE:

Dips of 50 or 80 are mechanically impossible; such faults would not slip under the present horizontal compressive stress regime. Instead, some new thrust fault would form with dip ~25.

There are also important implications for the metric seismic potency rate (per m of fault trace) which is defined as = (slip rate) x (down-dip width of the seismogenic portion of the fault).

This important measure of earthquake generation varies as 1/sin2(dip) when throw-rate is held constant (as in these 2 SSC studies, where it is fixed at the marine terrace uplift rate).

Compared to reasonable estimates (obtained with dip of 25), an assignment of 50 dip reduces seismic potency rate by a factor of 3.3x. An assignment of 80 dip reduces seismic potency rate by factor of 5.4x.

Thus, PG&E underestimated seismic potency of these 2 thrusts (which were the only ones they modeled) by large factors.

6

1. 3. Geologic structures older than ~0.33 Ma are irrelevant to seismic hazard estimation.

PG&E based the throw-rates of the San Luis Bay thrust fault and the Los Osos thrust fault on vertical offsets of marine & fluvial terraces with Upper Pleistocene ages, typically ~0.12 Ma.

PG&E never attempted to model the uplift and folding of sedimentary rocks in the Irish Hills which occurred since 5 Ma.

HOWEVER:

A detailed statistical analysis of geologic constraints on fault offset rates in the western United States by Bird [2007] found that the probability of inapplicability of a dated offset feature to neotectonics (defined in that paper, and shown in Figure 5) is equally low for all offset features up to 3 Ma (late Pliocene) in age, and almost as low for features of 5-6 Ma (Miocene/Pliocene boundary, or the time at which the Irish Hills began to form).

Furthermore, that study concluded that a single offset feature is very rarely enough to make the fault offset rate well-constrained; instead, 4 offset features are needed to achieve a 50%-chance that the rate is well-constrained, and 7 offset features are needed to guarantee it (Figure 6).

Thus, PG&E was negligent and unprofessional in failing to consider additional geologic constraints from older offset features, such as the once-planar Obispo Formation beds. PG&E should have created one or more structure models showing how this formation (and overlying sedimentary rocks) came to be bent into the present Pismo syncline and other folds in the center of the Irish Hills (Figure 7).

THEREFORE:

All the structures in the Irish Hills, which formed since 5 Ma, should have been studied and modeled to provide geologic constraints on the rates of thrust-faulting. It is strikingly negligent that they never considered or attempted this.

1. 4. GPS geodetic velocities are not useful for site-specific seismic hazard estimation.

PG&E operated a GPS receiver at DCPP, and PG&E [2015] reported the shortening direction across the Irish Hills as ~N15E (Figure 8), but did not report the shortening rate. The PG&E

[2024] update added no new geodetic information, even though an additional 9 years of data should have greatly reduced all uncertainties!

When questioned on this point at recent DCISC meeting, the PG&E representative explained that GPS velocity profiles cannot be used to measure fault heave rates unless the profile of GPS stations extends far from the fault on both sides, and that the coastline near DCPP makes this impossible.

HOWEVER:

Seismicity has been successfully forecast using only GPS data (onshore) and plate-tectonic models (offshore), both in southern California [Shen et al., 2007] and globally [Bird et al., 2010;

7 Bird & Kreemer, 2015; Figure 9]. Therefore, GPS data are very useful. Any deformation model used in SSC should fit GPS strain-rate constraints (within their uncertainties).

Furthermore, the problem of unavailable GPS velocities offshore is less serious in the case of thrust faults that dip away from the coast; theoretical models of dislocation patches in elastic half-spaces show that most of the interseismic strain occurs above the hanging wall, which in this case means on-land.

And, even if a GPS velocity profile across the Irish Hills does not record all of the interseismic heave rate, it still provides a useful lower limit on the rate of crustal shortening.

THEREFORE:

Models of neotectonic deformation, informed and guided by GPS velocity data, should be used in the estimation of seismic hazard. Specifically, Shen & Bird [2022] computed a suite of kinematic finite-element (F-E) models of neotectonics across the western US based on geodetic, geologic, & stress data with program NeoKinema. Their preferred model, which has been incorporated into the 2024 update of the USGS National Seismic Hazard Model, shows long-term-average convergence of crustal blocks on both sides of the Irish Hills/San Luis Range region at velocities of ~1 mm/a, for a total of ~2 mm/a of local horizontal convergence rate.

III. ESTIMATING the TOTAL RATE of THRUST FAULTING UNDER the IRISH HILLS The neotectonic uplift rate of the Irish Hills has been determined by PG&E (or possibly by its contracted consultants) to be approximately 0.2 mm/year, based on topography and ages of uplifted marine terraces compared to a global sea-level history. This is basic geologic data, and we are willing to stipulate that this uplift rate is approximately correct.

However, these throw rates (vertical offset rates) for the bounding Los Osos and San Luis Bay thrust faults are compatible with much higher slip-rates if the dips of these faults are shallow.

Also, thrust faulting on additional unmapped blind thrusts (e.g., Inferred Coastline thrust, and other unknown thrust faults within the Franciscan Complex basement) makes additional contributions to hazard. Fortunately, there are 3 ways to estimate the total rate of thrust fault slip in the Irish Hills without necessarily knowing the exact positions of each active fault plane.

Method #1: Isostatic:

The neotectonic uplift rate of the whole Irish Hills region is uniform at 0.2 mm/a. However, active thrust faulting must explain not only this increase in topography, but also an increase in crustal thickness that is much greater.

Isostasy means equal standing or, more clearly, standing still when vertical loads are equal. Because the asthenosphere acts tectonically like a viscous fluid, vertical columns of lithosphere will rise or fall until their gravitational loads (per unit area) on horizontal surfaces in the asthenosphere are equal. If a hilly region (like the Irish Hills) has extra mass above sea level (relative to adjacent terranes), it must have a compensating mass-deficiency at depth. The most common finding is that elevated regions have crustal roots meaning unusually thick crust beneath them. This satisfies isostasy because all crustal rocks are less

8 dense than all mantle rocks. While there are alternative models (e.g., Pratt isostasy in which variations in mantle lithosphere are important), simple Airy isostasy balances all vertical columns at the depth of the (deepest) Moho. The quality of this rule of thumb can be checked by measuring surface gravity on a grid of points, and then computing the isostatic gravity anomaly which will be near zero if Airy isostasy applies. In fact, isostatic gravity anomalies in the western US can be loosely described as 0+/-50 mGal. A positive value indicates undercompensation (deficient crustal roots) and a negative value indicates overcompensation (crustal roots thicker than expected). Seismic refraction studies by George P. Woolard and others have shown, since the early 20th century, that global plots of crustal thickness vs. surface elevation show a trend line with a slope of ~6.

As we have seen, simple Airy isostasy implies that the ratio of crustal thickening to topographic rise is about 6. Therefore, a simple isostatic model for the total rate of thrust-fault slip under the Irish Hills is at least:

(0.2 mm/year uplift) x 6 / sin(25 dips) = 2.8 mm/year If this crustal thickening is occurring on a single thrust fault of dip 25°, then its rate of slip should be 2.8 mm/a. Or, if the crustal thickening is driven by two oppositely-vergent and overlapping thrust faults then each should have a slip-rate of ~1.4 mm/a. Obviously, more complex models with more thrust faults can be devised, but the implication for total strain and seismicity due to thrust-faulting will remain unchanged.

Method #2: Thrust fault heave rate I provide one example in Figure 7: Throw of the Obispo Formation at the San Luis Bay-Inferred Coastline thrust fault is 1.6~2.2 km since 5 Ma, implying throw-rate of 0.32~0.44 mm/year, and fault slip rate of 0.76~1.04 mm/year.

If thrusting in the Irish Hills has been symmetrical(?), then a minimum total thrust slip-rate by this method would be 1.52~2.08 mm/year. (However, this estimate neglects any internal blind thrusts, so it is only a lower-limit estimate.)

Method #3: GPS geodetic horizontal shortening rate Our national-scale GPS-based deformation models (cited above) produced estimates of ~2 mm/a of long-term-average horizontal crustal shortening across the Irish Hills. Therefore, total thrust fault slip rate under the Irish Hills would be (~2 mm/year) / cos(25) = ~2.2 mm/year.

Since NeoKinema models are complex (although well-accepted), another approach is to look directly at the GPS site velocities. This means looking at interseismic relative velocities, not the long-term-average velocities which are provided by modeling.

Interseismic velocities are always spatially smoother, so there is a possibility that they may not capture the full rate of tectonic deformation when the dataset is spatially limited by an adjacent shoreline. I apply this approach to the geodetic dataset that was used as input by

9 Shen & Bird [2022] and all other USGS-sponsored modelers who participated in the 2022 Update to the National Seismic Hazard Model.

Using these raw GPS velocities, we see that station DCAN (Diablo Canyon) is converging at 0.4 mm/a with respect to DAPK (Prefumo Cyn., mid-Irish Hills) and converging at 1.1 mm/a with respect to CHOR (on Hwy 1, NNE of DCAN). Alternatively, site 2110 (slightly NNE of Point San Luis) is converging at 1.7 mm/a with respect to DAPK (half-way across the Irish Hills), and converging at 2.4 mm/a with respect to CHOR (across the full width of the Irish Hills). One way to summarize these results is to note that the average of the DCAN-CHOR interseismic shortening rate with the 2110-CHOR interseismic shortening rate is 1.8 mm/a. Allowing for the bias that interseismic rates may underestimate near a coastline, these interseismic rates tend to support the long-term-average shortening rate of 2 mm/a across the Irish Hills that the Shen & Bird [2022] NeoKinema model predicts.

IV. A MODEL CHARACTERISTIC THRUST EARTHQUAKE Each time a false assumption in the PG&E SSCs was removed, thrust-faulting activity (seismic potency rate) in the Irish Hills went up by a large factor. It is important to estimate how these factors combine, and how high seismic hazard (and seismic core damage frequency, SCDF) may be at DCPP. This could be done with a new SSC study and a new SPRA study, except that we cannot afford years of time and millions of dollars.

Instead, we will use a much simpler method to show that the lower limit on seismic hazard (and SCDF risk) due to thrust-faulting alone is much higher than the total hazard claimed by PG&E.

We will do this by adopting a characteristic great thrust earthquake for this tectonic setting, and then estimating its frequency in the Irish Hills.

In SSC and PSHA studies that include fault seismic sources with very incomplete information, it is traditional to assume a periodic characteristic earthquake model. While this is only an approximation of the chaotic earthquake dynamics in the real Earth, it has the advantage of allowing simple arithmetical conversions between the triad of basic parameters: slip per earthquake, long-term geologic slip-rate, and earthquake recurrence interval. For example, to compute the recurrence interval for large characteristic thrust-faulting earthquakes under the Irish Hills, it is sufficient to divide the mean coseismic slip in the characteristic earthquake by the long-term tectonic slip-rate of thrusts under the Irish Hills.

The Noto Peninsula on the northwest coast of Japan is tectonically analogous to the Irish Hills:

Both are elliptical blocks of crust now being uplifted from beneath shallow seas between two conjugate intraplate thrust faults [Toda & Stein, 2024].

Both occur within crust that formed in a subduction setting, with significant input of arc-derived graywacke clastic sediments.

Both experienced an extensional phase prior to the present compressional phase. The opening of the Sea of Japan was Miocene, preceding present AM-OK plate convergence

[Bird, 2003]. The future Irish Hills region experienced a brief episode of crustal melting, volcanism, and uplift due to formation of a slab window after the 28 Ma disappearance of

10 the Farallon plate at this latitude [Nicholson et al., 1994; Wilson et al., 2005]. The Edna normal fault (and others) were active during this Miocene episode, but not after. In Late Miocene the region cooled and subsided, and remained tectonically quiet until 6-5 Ma.)

Both include active thrust faults that were accidentally omitted from their respective national seismic hazard models because they are either blind or under water.

The primary difference is that horizontal convergence in the Noto Peninsula is faster (~10 mm/a vs. ~2 mm/a), allowing us a better opportunity to observe its seismicity in our brief historical datasets. Further discussion of these similarities and differences in in Appendix B.

On 1 January 2024, at 07:10 UTC, a very large earthquake occurred beneath the Noto Peninsula on the northwest coast of Ishikawa Prefecture, Japan. Its magnitude was 7.6 on the moment-magnitude scale used by the Japan Meteorological Agency, and 7.5 on the moment-magnitude scale used by USGS. This thrust-faulting shock achieved a maximum JMA seismic intensity of Shindo 7 and Modified Mercalli intensity of IX (Violent) [Wikipedia, 2024]. These intensities are very high.

We learned 2 essential facts from this 2024.01.01 m7.5 earthquake [Toda & Stein, 2024]:

(1) We have the advantage of the finite-fault solution (USGS, 2024; Figure 10), which maps the amount of coseismic slip onto the active fault plane. This study showed maximum slip of 3.7 m under the center of the Noto Peninsula, with a mean slip that I visually estimate as 2.0 m (or 2000 mm) within the seismogenic depth range, under the part of the fault trace that parallels the Noto Peninsula.

(2) Peak ground accelerations (PGA; Figure 11) at 5 strong-motion seismometers were 1.0~2.3 g as far as 42 km from the rupture.

V. CONCLUSION: SEISMIC CORE DAMAGE FREQUENCY at DCPP:

The two SSC studies by PG&E [2015; 2024] seriously underestimated the seismic hazard from thrust-faulting under the Irish Hills because they relied on 4 demonstrably false assumptions.

Three independent analytic methods give values for the total slip-rate on all shallow-dipping thrust faults under the Irish Hills: 2.8 mm/year, ~2.0 mm/year, or 2.2 mm/year.

Using the 2024.01.01 Noto Peninsula earthquake as a characteristic thrust earthquake (with its 2 m of mean slip) yields recurrence times for great thrust earthquakes under the Irish Hills of 715 years, 1000 years, or 910 years, respectively.

This raises the question of whether PGA of 1.0~2.3 g will cause seismic core damage (SCD) at Diablo Canyon Units 1 & 2? Answering this question quantitatively becomes technical and difficult, given that spectral accelerations critical to individual component failures are typically twice as large as PGA; that is, perhaps 2.0~4.6 g at vibration frequencies of 5~10 Hz for the Noto Peninsula earthquake analog.

The 2018 SPRA [PG&E, 2018] is the most recent available to me. Within this document, Table 5.4-4 (page 65) shows how PG&Es overall SCDF of 2.8x10-5 /yr was obtained. In principle, it

11 should be possible to use this information to estimate the probability of SCD at each level of shaking. My interpretation of the table is that the probability of SCD is ~6% at 2 g, rising to

~73% at 3 g and to >98% at 4 g. The problem is that the acceleration levels quoted in this table are not clearly identified; are they PGAs or (more likely) spectral accelerations? The context in this SPRA report suggests that they are spectral accelerations: the introductory section 3.1.3 Seismic Hazard Analysis Results and Insights only discusses 5 Hz spectral accelerations, and the primary graphs that it refers to (Figure 3 Reference Rock Hazard by Source for 5 Hz Spectral Acceleration and Figure 3 5 Hz Control Point Mean and Fractiles Horizontal Hazard) are plots of 5 Hz spectral acceleration.

Therefore, my interpretation of this report is that a PGA event of 1.0 g would produce 5 Hz spectral accelerations of ~2 g, and incur ~6% chance of SCD. However, a PGA event of 1.5 g would produce 5 Hz spectral accelerations of ~3 g, and incur a ~73% chance of SCD. And the peak Noto-earthquake observation of PGA of 2.3 g would produce spectral accelerations of ~4.6 g, and incur >98% chance of SCD at DCPP.

It will probably be controversial exactly which of the Noto Peninsula seismograms give the median and worst-case forecasts of shaking at DCPP. The paragraph above shows that this is a critical point. Clearly these questions need to be resolved by independent experts, preferably in a revised SSC study followed by a revised SPRA study. In the meantime, for purposes of evaluating the acceptability of PG&Es two SSCs, it is sufficient to assume that the levels of shaking seen in the Noto Peninsula earthquake will cause seismic core damage at DCPP if and when they occur in the Irish Hills of California.

Assuming that such a great thrusting earthquake would cause seismic core damage at DCPP, its seismic core damage frequency (SCDF) is at least 1.4x10-3 /year, or 1.0x10-3 /year, or 1.1x10-3 /year, respectively.

[This is before the hazard contributions from strike-slip faults like the Hosgri and Shoreline are added.]

12 APPENDIX A. MISSING PROCEDURAL STEPS in the SSCs by PG&E [2015; 2024]

Another fundamental problem with PG&Es [2015; 2024] seismic risk analyses (besides logic guided by false assumptions) is the subjective (i.e., committee-based, not algorithm-based) creation of fault geometry models to support the 2015 and 2024 SSCs. These kinematically incomplete models also biased PG&Es [2018] SPRA. Their fault geometry models do not meet basic scientific standards for objectivity and reliability because are not geometrically self-consistent, nor are they consistent with GPS and regional stress directions. They did not take account of, or make use of, then-published and available scientific developments in:

-measurement of crustal motion by permanent and campaign Global Positioning System (GPS) receivers [e.g., Shen et al., 2003; Kreemer et al., 2003, 2014; Kreemer, 2016]; or

-computer modeling (including kinematic finite-element models) of such data, in combination with geologic and stress data, to compute long-term crustal strain rates and fault slip rates [e.g.,

use of program NeoKinema: Bird, 2009; Field et al., 2013, 2014; Parsons et al., 2013]; or

-recent initiatives in seismic hazard estimation which do not assume that a complete inventory of active faults is available, but instead compute the expected seismicity across the map area from crustal long-term (permanent, not elastic) strain rates and fault slip rates (if and where available) using a calibration of global shallow seismicity categorized by plate-tectonics [Bird & Kagan, 2004; Bird & Liu, 2007; Bird et al., 2009; Bird et al., 2010]. Two motivations for the development of such models were that: (a) a number of recent large earthquakes in the California region have occurred in places where no seismogenic fault, or only short disconnected faults had been recognized (Landers 1972 m7.3, Hector Mine 1999 m7.1, El Mayor-Cucupah 2010 m7.2, Ridgecrest 2019 m6.5 + m5.4 + m7.1); and (b) the discovery that the global distribution of shallow earthquakes shows that they spread in bands of half-width 257 km [Bird & Kagan, 2004]

around plate boundary faults of the Continental Transform Fault (CTF) type.

In particular, PG&E [2015; 2024] should have used the globally-calibrated strainrate-to-seismicity conversion (see paragraph above) to obtain the seismic moment rate for their Local Area Source component. By using a seismic moment based on microseismicity in a few quiet decades between seismic crises, they seriously underestimated it.

In 2012, I participated in a Senior Seismic Hazards Analysis Committee (SSHAC) Level-III workshop sponsored by Pacific Gas & Electric Co. (PG&E) and run by Lettis Consultants International, regarding seismic hazard at the Diablo Canyon Power Plant. I presented results on both strike-slip and compressional deformation rates affecting the region, which were derived from my latest NeoKinema computer models of neotectonics. (These models were prepared for the Southern California Earthquake Centers project Unified California Earthquake Rupture Forecast version 3, and also for the US Geological Surveys 2013 Update to the National Seismic Hazard Model.) At that time, I offered to share my modeling codes and methods with the TI team, at no cost to them, but that offer was declined.

13 APPENDIX B. Dismissal of PG&Es Objection to the Noto Peninsula Earthquake Model On 24 October 2024, PG&E submitted PG&E Letter DCL-24-103 to the Petition Review Board of the Nuclear Regulatory Commission (nrc.gov ADAMS Accession # ML24298A234) which contained Enclosure 2, a technical report by Lettis Consultants International, Inc., entitled Phase 1 Review of the Tectonic and Geomorphic Setting of the January 1, 2024, M7.5 Noto Earthquake, Noto Peninsula, Japan (56 pages). Herein I will refer to this as the LCI report.

The principal subject of the LCI Report was the 2024.01.01 Noto Peninsula earthquake, and whether this could be accepted as a model characteristic thrust earthquake for the setting of the Irish Hills in California (as I have suggested). Here I will quote some of their concerns in italic type, followed by my responses. Points of agreement or common knowledge will not be included. For example, LCI agrees that the Noto Peninsula is an inverted fault-bounded basin that has been uplifted between oppositely-vergent thrust faults, and that the start of its compressional phase was 4-3 Ma, similar to the 5 Ma initiation of the Irish Hills.

LCI Report: The San Luis-Pismo Block [in the Irish Hills] is interpreted to have been uplifted as a rigid block during the late Quaternary by reverse slip on both the Los Osos fault and Southwestern Boundary zone faults.

Response: LCI acknowledges that this idea of a rigid block is copied from PG&E [2015]. I have refuted it in previous sections of this report.

LCI Report: The Q1 terrace [in the Irish Hills] is almost flat, with a relatively gentle shoreline slope, which along the southwestern coastline of the Irish Hills is locally steeper, possibly reflecting increased colluvial and/or alluvial cover (AMEC, 2011a).

Response: This increased slope could also be interpreted as due to continuing forced folding of the hanging-wall of the blind Inferred Coastline thrust (my term).

LCI Report: Compared to the Noto Peninsula faults that are readily identifiable from their acute geomorphic expression on the seafloor, the relative lack of expression of the Irish Hills bounding faults indicates that despite exhaustive investigation from detailed topographic, bathymetric, geologic, and geophysical investigations onshore and offshore, these faults are not directly analogous to those generating uplift along the Noto Peninsula. [Section 4.2.1]

Response: This difference in fault-scarp expression is expected based on the fact that horizontal shortening in the Noto Peninsula is about 5x faster than in the Irish Hills (~10 mm/a vs. ~2 mm/a), whereas the initiation ages are similar (4-3 Ma vs. 5 Ma). The Noto Peninsula thrusts have slipped further, and are currently slipping faster, so they have more prominent scarps. This is not an important distinction.

OVERVIEW: PG&Es cover letter to NRC first describes Enclosure 2 (the LCI Report) and then states that, This study shows significant differences between the two regions and refutes the Petitioners claims that the Noto earthquake is a direct corollary for potential earthquakes and hazard models for DCPP. However, there is no basis for this statement in the LCI Report, which instead details dozens of similarities between the Noto Peninsula and the Irish Hills.

14 FIGURES Figure 1. Geologic map of the Irish Hills region [PG&E, 2014]. The large Pismo syncline fold in the southwestern half of the region can be identified by a single strip of beige color, surrounded by matched strips of green color, matched strips of light blue color, and matched strips of purple color. Dips in both limbs of this fold average ~45.

15 Figure 2. Isostatic gravity anomaly map of the Irish Hills region [PG&E, 2024]. Note that negative values (about -22 mGal) are typical in the Irish Hills. These negative values indicate that the hills are over-compensated meaning that their supporting crustal roots are thicker than would be predicted by simple Airy isostasy.

16 Figure 3. Detail of the fault-trace map of PG&E [2015], edited to show the dip-directions of thrust faults (with triangles), and the slip-directions of dextral faults (with paired arrows) and to add my proposed Inferred Coastline thrust fault trace, continuing the fault system which PG&E recognized as the San Luis Bay fault zone. Colors of different fault traces have no importance, except to separate them visually.

17 Figure 4. Reproduction of Figure 5 from Byerlee [1978], showing results of decades of laboratory tests on the friction of diverse rock types. Note that a friction coefficient (slope) of 0.85 is typical. This result is loosely referred to as Byerlees Law. There was no pore pressure in these experiments, so the horizontal axis label NORMAL STRESS could also be interpreted as EFFECTIVE NORMAL STRESS.

18 Figure 5. Reproduction of Figure 8 of Bird [2007], showing how the chance of inapplicability to neotectonics (vertical axis) varies with the difference between the ages of two sets of piercing points along the same fault. In California, inapplicability to neotectonics is only a problem for offset features older than 5 Ma. All younger features are equally relevant.

19 Figure 6. Reproduction of Figure 9 of Bird [2007], showing results from a study of dated fault offsets across the Gorda-California-Nevada orogen (GCN region). The resulting fault offset rate is only well-constrained (as defined by the vertical axis label) when more than 4 pairs of offset features are used in a combined statistical analysis. Therefore, a single dated offset feature (as used by PG&E [2015; 2024]) is not enough.

20 Figure 7. Mark-up (in red) of geologic cross-section Figure 13-17 from PG&E [2015]. The basis for the steep dip shown (in black) for the Los Osos thrust fault is weak, and I suggest an alternative dip of 25. I also add the Inferred Coastline thrust fault (continuing along the line of the San Luis Bay thrust fault) on the left side with dip of 25. Dashed red lines suggest how these two thrust may have offset each other during 5 m.y. of finite strain, to create a suite of shallow-dipping blind thrust faults. This model explains the deep microseismicity better than PG&Es model. At top left, throw (vertical offset) of the Inferred Coastline thrust is labelled, based on apparent offset of geologic unit Tmo (Tertiary Miocene Obispo Formation).

21 Figure 8. Reproduction of Figure 5-10 from PG&E [2015], showing how two alternative networks of GPS geodetic stations were used to estimate a plausible azimuth of ~N15E for the most-compressive strain-rate axis. However, their figure is misleading because a generic strike-slip strain-rate tensor symbol was used to show strain-rate orientations, whereas in fact the GPS data indicate almost pure thrust faulting. That is, the WNW-ESE-trending pair of arrows in each symbol should be ignored.

22 Figure 9. Global forecast of shallow seismicity by Bird & Kreemer [2015], shown with colors on a logarithmic scale to represent variation over 5 orders of magnitude. This forecast was based entirely on GPS geodetic velocities and plate boundary locations, and did not use the catalog of historic earthquakes. However, the similarity to actual recorded earthquakes in 1918-1976 (shown with black dots) is very strong. This result demonstrates the value of GPS geodetic data for the forecasting of seismicity.

23 Figure 10. Reproduction of a figure from Toda & Stein [2024], showing the slip pattern of the 2024 Noto Peninsula earthquake (based on the USGS finite-fault solution), superposed on topography. Note that this earthquake ruptured bilaterally from a hypocenter with low slip; if it had only propagated to the SW, the magnitude and rupture length would have been less.

Also note that the mean slip (within the seismogenic depth range) in the Noto Peninsula portion of the rupture is about 2 m.

24 Figure 11. Reproduction of a figure of Toda & Stein [2024], showing iso-intensity contours (inset map) and the PGA-versus-distance graph for the 1 January 2024 Noto Peninsula earthquake. Note that PGA of 100% to 230% of gravity were recorded at 5 digital strong-motion stations on the Noto Peninsula, up to 42 km from the rupture. It is also apparent that the standard USGS ShakeMap model of the attenuation of PGA with distance is not applicable to this thrust earthquake.

25 REFERENCES CITED Atwater, T., and J. Stock [1998] Pacific-North America plate tectonics of the Neogene southwestern United States: An update, Int. Geol. Rev., 40, 375-402.

Austermann, J., Z. Ben-Avraham, P. Bird, O. Heidbach, G. Schubert, and J. M. Stock [2011]

Quantifying the forces needed for the rapid change of Pacific plate motion at 6 Ma, Earth Planet.

Sci. Lett., 307, 289-297, doi:10.1016/j.epsl.2011.04.043.

Bird, P. [2003] An updated digital model of plate boundaries, Geochemistry Geophysics Geosystems, 4(3), 1027, doi:10.1029/2001GC000252.

Bird, P. [2007] Uncertainties in long-term geologic offset rates of faults: General principles illustrated with data from California and other western states, Geosphere, 3(6), 577-595; doi:10.1130/GES00127.1, + 9 digital file appendices.

Bird, P. [2009] Long-term fault slip rates, distributed deformation rates, and forecast of seismicity in the western United States from fitting of community geologic, geodetic, and stress direction datasets, J. Geophys. Res., 114, B11403, doi: 10.1029/2009JB006317.

Bird, P., and Y. Li [1996] Interpolation of principal stress directions by nonparametric statistics:

Global maps with confidence limits, J. Geophys. Res., 101(B3), 5435-5443.

Bird, P. and Y. Y. Kagan [2004] Plate-tectonic analysis of shallow seismicity: Apparent boundary width, beta, corner magnitude, coupled lithosphere thickness, and coupling in seven tectonic settings, Bull. Seismol. Soc. Am., 94(6), 2380-2399.

Bird, P., and Z. Liu [2007] Seismic hazard inferred from tectonics: California, in: S. E. Hough and K. B. Olsen (ed.), Special Issue on: Regional Earthquake Likelihood Models, Seismol. Res.

Lett., 78(1), 37-48.

Bird, P., and C. Kreemer [2015] Revised tectonic forecast of global shallow seismicity based on version 2.1 of the Global Strain Rate Map, Bull. Seismol. Soc. Am., 105(1), 152-166, doi:

10.1785/0120140129.

Bird, P., and R. V. Ingersoll [2022] Kinematics and paleogeology of the western United States and northern Mexico computed from geologic and paleomagnetic data: 0 to 48 Ma, Geosphere, 18(5), 1563-1599, https://doi.org/10.1130/GES02474.1.

Bird, P., Y. Y. Kagan, D. D. Jackson, F. P. Schoenberg, and M. J. Werner [2009] Linear and nonlinear relations between relative plate velocity and seismicity, Bull. Seismol. Soc. Am., 99(6),

3097-3113, doi: 10.1785/0120090082.

Bird, P., C. Kreemer, and W. E. Holt [2010] A long-term forecast of shallow seismicity based on the Global Strain Rate Map, Seismol. Res. Lett., 81(2), 184-194, plus electronic appendices.

Byerlee, J. [1978] Friction in rocks, Pure Appl. Geophys., 116, 615-626.

Cloos, M. [1982] Flow melanges: numerical modeling and geologic constraints on their origin in the Franciscan subduction complex, California, Geol. Soc. Am. Bull., 93(4), 330-345.

26 Ekstrm, G., M. Nettles, and A. M. Dziewonski [2012] The Global CMT project 2004-2010:

Centroid moment tensors for 13,017 earthquakes, Phys. Earth Planet. Int., 200/201, 19.

Field, E. H., G. P. Biasi, P. Bird, T. E. Dawson, K. R. Felzer, D. D. Jackson, K. M. Johnson, T. H.

Jordan, C. Madden, A. J. Michael, K. R. Milner, M. T. Page, T. Parsons, P. M. Powers, B. E.

Shaw, W. R. Thatcher, R. J. Weldon, II, and Y. Zeng [2013] Unified California Earthquake Rupture Forecast, version 3 (UCERF3)-The time-independent model, U.S. Geol. Surv. Open-File Rep., 2013-1165 (and Cal. Geol. Surv. Spec. Rep. 228, and Southern California Earthquake Center Pub. 1792), 97 pages (main report) + 20 Appendices; http://pubs.usgs.gov/of/2013/1165/.

Heidbach, O., M. Tingay, A. Barth, J. Reinecker, D. Kurfe, and B. Müller [2008] The World Stress Map database release 2008, doi:10.1594/GFZ.WSM.Rel2008.

Heidbach, O., M. Rajabi, K. Reiter, M.O. Ziegler, and the WSM Team [2016] World Stress Map Database Release 2016, doi:10.5880/WSM.2016.001.

Kreemer. C. [2016] GEM Strain Rate Model v.2.2, GEM Final Reports on Scientific Projects, Global Earthquake Model Project, Pavia, Italy, 11 pages.

Kreemer, C., W. E. Holt, and A. J. Haines [2003] An integrated global model of present-day plate motions and plate boundary deformation, Geophys. J. Int., 154, 8-34.

Kreemer, C., G. E. Klein, Z.-K. Shen, M. Wang, L. Estey, S. Wier, and F. Boler [2014] Global Geodetic Strain Rate Model, GEM Technical Report, 2014-07(V1.0.0), 129 pages; doi:

10.13117/GEM.GEGD.TR2014.07.

Mueller, B., V. Wehrle, and K. Fuchs [1997] The 1997 release of the World Stress Map, http://www-wsm.physik.uni-karlsruhe.de/pub/Rel97/wsm97.html.

Nicholson, C., C. C. Sorlien, T. Atwater, J. C. Crowell, and B. P. Luyendyk [1994] Microplate capture, rotation of the western Transverse Ranges, and initiation of the San Andreas as a low-angle fault system, Geology, 22(6), 491-495.

Pacific Gas and Electric Company [2014] Geologic Map of the Irish Hills and Adjacent Area, 1:32,000, DCPP Geologic Mapping Project, Ch9.GEO.DCPP.TR.14.01 R0, https://www.pge.com/includes/docs/pdfs/safety/systemworks/dcpp/report/Ch9.GEO.DCPP.TR.14

.01_R0_Plates.pdf, NRC ADAMS Accession No. ML14260A068.

Pacific Gas and Electric Company (PG&E) [2015] Seismic Source Characterization for the Diablo Canyon Power Plant, San Luis Obispo County, California; report on the results of SSHAC level 3 study, Rev. A, March; 652 pages plus Appendices. Available online at http://www.pge.com/dcpp-ltsp; downloaded 2023.05.11.

Pacific Gas and Electric Company (PG&E), 2018. Letter DCL-18-027 re: Seismic Probabilistic Risk Assessment for the Diablo Canyon Power Plant, Units 1 and 2 - Response to NRC Request for Information Pursuant to 10 CFR 50.54(f) Regarding the Seismic Aspects of Recommendation 2.1: Seismic of the (sic) Near-Term Task force Review of Insights from the Fukushima Dai-Ichi Accident (Apr. 24, 2018) (NRC Accession No. ML18120A201).

27 Pacific Gas and Electric Company (PG&E) [2024] Diablo Canyon Updated Seismic Assessment:

Response to Senate Bill 846, 1 February 2024, 392 pages.

Page, B. M., G. A. Thompson, and R. G. Coleman [1998] Late Cenozoic tectonics of the central and southern Coast Ranges of California, Geol. Soc. Am. Bull., 110(7), 846-876.

Parsons, T., K. M. Johnson, P. Bird, J. Bormann, T. E. Dawson, E. H. Field, W. C. Hammond, T.

A. Herring, R. McCaffrey, Z.-K. Shen, W. R. Thatcher, R. J. Weldon, II, and Y. Zeng [2013]

Appendix C: Deformation Models for UCERF3, in: E. H. Field et al. (ed.), Uniform California Earthquake Rupture Forecast, Version 3 (UCERF3)-The time-independent model, U.S. Geol.

Surv. Open-File Rep., 2013-1165(Cal. Geol. Surv. Spec. Rep. 228, and Southern California Earthquake Center Pub. 1792), 97 pages; http://pubs.usgs.gov/of/2013/1165/pdf/ofr2013-1165_appendixC.pdf.

Shen, Z.-K., and P. Bird [2022] NeoKinema deformation model for the 2023 update to the U.S.

National Seismic Hazard Model, Seismol. Res. Lett., 93, 3037-33052, doi: 10.1785/0220220179.

Shen, Z.-K., D. C. Agnew, R. W. King, D. Dong, T. A. Herring, M. Wang, H. Johnson, G.

Anderson, R. Nikolaidis, M. van Domselaar, K. W. Hudnut, and D. D. Jackson [2003] The SCEC Crustal Motion Map, Version 3.0, http://epicenter.usc.edu/cmm3/.

Shen, Z.-K., D. D. Jackson, and Y. Y. Kagan [2007] Implications of geodetic strain rate for future earthquakes, with a 5-year forecast of M5 earthquakes in southern California, Seismol. Res.

Lett., 78(1), 116-120.

Toda, S., and Stein, R. S. [2024] Intense seismic swarm punctuated by a magnitude 7.5 Japan shock, Temblor, http://doi.org/10.32858/temblor.333 Wakabayashi, J. [1999] Distribution of displacement on and evolution of a young transform fault system: The northern San Andreas fault system, California, Tectonics, 18(6), 1245-1274.

Wilson, D. S., P. A. McCrory, and R. G. Stanley [2005] Implications of volcanism in coastal California for the Neogene deformation history of western North America, Tectonics, 24(3),

TC3008, doi:10.1029/2003TC001621.