ML19260E177

Review of Volcanologic Discussions in PSAR & Related Documents

ML19260E177
Person / Time
Site:	05000514, 05000574
Issue date:	12/22/1979
From:	Newhall C DARTMOUTH COLLEGE, HANOVER, NH
To:
Shared Package
ML19260E172	List: ML19260E170 ML19260E173 ML19260E177
References
TETR-791222, NUDOCS 8002130602
Download: ML19260E177 (96)

Text

k hEVIEh CF VCLCAHOLCGIC DISCUSSIGUS Ih THE PSAf: ALE hELATED CCCUMEUTS, PHILIPPINE UUCLEAR POWER PLANT //1 Christopher Newhall 22 Cecember 1979 Cocuments Reviewed:

1. Ebasco, PSAR Section 2.5.1 casic Geologic and Seismic Information Appendix 2.5.C Microearthquake Studies Appencix 2.5.H.1 Geochronological Investigations Appendix 2.5.H.2 Geochemistry of rocks in western Luzon Appendix 2.5.H.3 P aleon.a gn e tism Appendix 2.5.J Stochastic time serics analysis of geochronolo61 cal data

2. Ebasco, Responses to PAEC Cuestions of 25 August 1977 3, Ebasco, Geology of Unit 1 Excavation February 1978 4

Report of the IAEA Safety Mission 7 July 1978

5. Ebasco, Additional Safety related data, 30 June 1978, response to the IAEA Safety Mission (pp. 56-8f, entitled Volcanism and Volcanic lia: arcs)

6. PAEC evaluation of the responses on the issues raised by the 1978 IAEA Safety Mission 22 March 1979

7. Ebasco, Response to PAEC Question #3 re: the PSAR, 8002130 4 6Q2

s entitled "Evicence substantiating the increcibility of volcanism on the west flank cf Mt. !!atib, and the assessn.ent of volcanic hazards at

t.

!;atib.

8. A.YC topographic maps: 1: 50,000 and 1:250,000 scale topographic maps of the Bataan Peninsula 2

a 6

Summary of Principal Points

1. Long " repose" pericas, thermal springs, youthful morphology and young pyroclastic ceposits all suggest that Natib may b? dormant rather than extinct, and therefore poses finite risks of renewed eruptive activity.

This is, I think, a point of General agreement.

2. Ebasco has correctly raised most questions necessary to evaluate volcanic hazards at the Napot Pt. site.

In so doing, they have made a very useful contribution to what I hope will be further discussion of evaluating volcanic hazards for nuclear plant sites in general.

In attempting to answer the questions raised, they have used many standara geclogic techniques and a few new ones, but they have failed to use several other wiaely-accepted techniques for evaluating volcanic hazaras.

3. Detailed Seologic maps, stratigraphic sections and volcanic hazard maps for Mt. Natib are not Given.

We are given instead reconnaissance geologic maps and a generalized stratigraphic column.

Detailed maps and columns are within the state-of-the-art and would provide much additional information needed to understand the eruptive history and hazards of Mt.

Natib.

With the detailed geologic base one can then say exactly where, when and what types of eruptions occurred; without the detailed geology one can only discuss generalities of eruptive style, age ranges and minimum recurrence frequencies for various eruptive phenomena.

With detailed volcanic hazara maps one can 3

=

+

see the. areal distritution of each type of volcanic deposit and the relative frequency with which each type of activity affects each point on the map.

4. The age of the youngest deposit- /and eruptive activity) of Natib has not been adequately determined and may be, I suspect, more recent than the 69,000 years indicated by Ebasco.

Similarly, the age of the Napot Pt. deposits has not been adequately determined.

5. Ebasco uses reasonable methodologies to estitaate eruption probabilities, but because they use incomplete historical and geochronological data, their estimates appear to range from approximately correct to approximately 100 times too low.

6. The interpretation of a segmented subcuction zone seems reasonable; the interpretation of two periods of subduction, first along the ' Jest Luzon Trough and now along the Manila Trench, is not supported by their data.

7. Ebasco's early differentiation between probabilities of eruption at the summit and eastern vents is not reasonable; later Ebasco documents recognize the implausibility of such differentiation.

An eruption from the western flank of Natib is less likely than one from the summit or eastern vents, but it is not " incredible" in the sense that that word is normally used.

8. Specific volcanic hazards are discussed in some cetail.

Most of these hazards are adequately discussed, but the ciscussion of pyroclastic flows and surges places too much stress on topographic protection of the site while ignoring numerous 4

I 1

exceptions to this topographic control.

An apparent tendency toward major landslices on Natib and a probable past history of highly explosive volcanism leading to calcera collapse are inadequately discussed.

9. State-of-the-art volcano surveillance could detect premonitcry volcanic activity at Natib.

Given premonitory activity, though, the present state-of-the-art does not tell us whether an eruption is imminent (eg. tomorrow), will occur next month, or will not occur for many years in the future.

Careful monitoring and establishment of criteria for shutdown and fuel evacuation will be essential.

10. Many questions about nuclear power plants and volcanic hazards, eg. about what data needs to be collected, acceptable levels of risk, possible engineering measures against volcanic hazards, and procedures in the event of a volcanic crisis could be better answered if we had a carefully prepared set of volcanic hazard guidelines analogous to the existing seismic hazara guidelines.

5

a 6

Table of Contents I. Introduction II. Overall coverage of the volcano-related safety documents:

Has Ebasco asked the right questions?

III. Approaches and answers: i:as Ebas :o used the appropriate approaches to these questions, and are their answers complete and reasonable?

A. General volcano-tectonic setting E. Credibility of renewed eruptive activity on Mt Natib

1. Geologic map and stratigraphic control, for sampling purposes and for determination of temporal trencs in eruptive products, eruption style or eruptive frequency.

2. Geochronological procedures and statistical analysis of geochronological data.

3. Annual probability of an explosive eruption of Mt. Natib 4

Thermal springs

5. Tectonic interpretations based on geochemical data including the division of Mt. Natib into two subduction regimes.

6. Distinction between the probabilities of eruption from the summit and from the E and W flanks of Mt. Natib.

C. Volcanic hazards in the event of an eruption of Mt. Natib.

IV. General comments on monitoring anc prediction 6

t s

I.

Introduction The proposal to build a nuclear power plant on the south flank of Mt. Natit, Eataan, Philippines requires a careful examination of the volcanic hazards at that site.

Both qualitative and quantitative estimates of volcanic hazards are needed, so that planners may weigh the severity and probabilities of each hazard and their net costs (in terms of safety risk and aoded design specifications) against the economic and social benefits of the plant.

Volcanic crises in the Caribbean over the past several years, and the ensuing debate, (cf. editorials and letters to the editor, Journal of Volcanology and Geothermal Research, 1976-present) have emphasized a need for clear, complete and realistic evaluation of volcanic hazards, coupled with a constructive dialogue between volcanologists and government officials.

The two quotations which follow illustrate a pragmatic approach.

Referring to the Caribbean crises and to debate over the Philippine site, William:.

nd McEirney (1979, pp. 359-360) write:

... Surveillance in itself does not alleviate volcanic hazards.

It must be coupled with land use management, protective measures to reduce the impact of an eruption, and, above all a

realistic appraisal of potential effects.

Tazieff (1977) has drawn attention to some of these problems, pointing out that geologists who are only vaguely familiar with the behavior of active volcanoes tend to take an unnecessarily conservative view of potential events and often advise precautionary measures out of all proportion to risks that experienced volcanologis.s would recognize as unrealistic.

When faced with conflicting advice from two or more geologists, political authorities prepare for the worst eventualities, even though their precautions may entail wholesale evacuation and economic consequences that are worse than the eruption itself.

7

4 i

' Examples of this type of exagger ted response are seen in the costly over-design of certain

.,strial installations, such as nuclear power plants that have been proposed in volcanic regions.

The conservatism of plant.ers who wish to guard against every possible hazard can easily be exploited by persons who wish to obstruct the project by advising prohibitively expensive precautions to guard against risks that are exceecingly unlikely to occur but are very difficult to cisprcyc.

It is possible to take prudent precautions without exaggerctint hazarcs anc recorrending actions that cre cut of proportion to the econcmic and social consequences they entail."

Making a similar point, Crandell and Mullineaux (1975) write:

"...what assurance have we that the next eruption of Mount Rainier, for example, will be the same kinc as has occurred repeatedly during the last 10,000 years, rather than a catastrophic Mcunt Mazama-type eruption which will destroy the volcano and devastate the entire adjacent region?

It cannot be said, of course, that Mount Rainier will never be another Crater Lake.

Eut the possibility of such a violent eruption occurring within the next few centuries is so remote that it cannot be planned for either economically or pragmatically."

An important point, implicit here, is that there is a whole spectrum of severities and probabilities in volcanic hazards, and an even wider spectrum when one subjects a nuclear power plant to these volcanic hazards.

A nuclear power plant in a volcanic area can be designed to withstand minor volcanic hazards, thereby making minimal risks even smaller, but at the same time a volcano-nuclear accident could be far worse than a simple volcanic " accident".

The questions, then, are two:

1. How does one go about making a quantitative or

" realistic" evaluation of the volcanic hazards in an area, ano

2. !!ow do these hazards apply to a nuclear power plant?

The first question requires detailed information about Mt.

8

4 l'atib combined with a broader Lase of information about similar volcanoes elsewhere in the wcrld (i.e. combined with what one m.ight call "volcanclogic common sense").

The most general and immediate " common sense" is one which all parties have recognized

-- that !!atib is r.orphologically young, has (geologically) young pyroclastic deposits, has therc.al activity and is part of a chain of active volcanoes along an active subduction zone, and therefore poses finite volcanic risks which must be evaluated.

The second question is tougher, anc there are no formal guicelines as yet to guide discussion.

How does one quantify the severity of each possible volcanic event with respect to a nuclear power plant?

And how does one decide exactly how much risk is acceptable?

'n' hat constitutes an " active" volcano, and are all " active" volcanoes unsuitable sites?

By analogy with scismic guicelines, what is a " capable" volcano?

'nhat is a

" credible" volcanic event?

How does one hancle those risks which defy quantification as to severity and/or probability?

Are there any volcanic risks which only become risks in the presence of a nuclear plant?

What is the desi6n life for the purposes of volcanic hazard evaluation, i.e. how many years will the plant operate and how many years after that will there still be concentrations of radioactive material on the site subject to volcanic dispersal?

ahat engineering measures can be taken against volcanic hazarcs and what volcanic hazards are essentially beyond the engineer's art?

What procedures should be established for monitoring volcanic activity and shutting down 9

t the plant anc evacuating fuel in a volcanic crisis?

These are tough questions which properly need to be acoressed in general Suidelines before evalusting the volcanic hazards for any specific site.

The science of volcanology can mcke a good cent in the first problem -- making " realistic" evaluations of volcanic risks at a given point on the land.

We are still far from a consensus (in fact, debate hasn't even begun in earnest) on the second problem, namely, how do general volcanic hazards translate into specific volcanic hazards for nuclear reactors?

The above-mentioned safety documents and this review focus on the general assessment of volcanic risks, with only occasional remarks on the specific risks to a reactor.

This review is emphatically not a complete, fully-quantitative evaluation of the volcanic hazards at the Napot Pt.

site, much less an evaluation of the hazards to a nuclear plant at that site.

It is insteac a review of Ebasco's volcanologic methodology and interpretations, measured against established volcanologic methods and this poorly-defined but important "volcanologic common sense".

The volcanological common sense is at the same time a calm " realistic appraisal of potential effects" and a cautious questioning: "Is this reasonable in the light of what we know about similar volcances?" All information specifically about Natib is from Ebasco; I am providing sonie additional general information -- pieces of the body of observations behind "volcanologic common sense" -- in the hope that it will contribute to the ultimate ciscussion of guidelines 10

s for evaluation of volcanic hazards at nuclear plant sites and, in the meantime, to further discussion of the volcanic hazarcs of the f;apot Pt. site.

11

n II. Cverall Ccverage of the Volcano-Felated Cafety Locuc.ents:

has Etasco Asked the Hight '.uestions?

There is no simple cookbcok in volcanclogy which outlines exactly what one neec to know in order tt ;ssess volcanic hazards.

There are several useful sunraries of volcanic hazarcs and their assessment, includin6 tnose by ' Walker (1974), Crandell anc Fullineaux (1975) anc Macdonald (in Eolt et al., 1977).

I have prepared a set of questions, drawing literally fron those papers, and have indicatec in each case whether the Etasco documents have accressed that question.

The questions are as follow:

1.

hhat is the general volcanic and tectonic setting of the propcsed site?

Eces the volcano in question stand alone or is it part of a chain or set of other vcleanoes?

Etasco has addressed these questiens.

2. hhat has been the past eruptive history of the volcano?

hhat have been the types and sequence of events?

what have been the relative frequencies of each type cf volecnic event?

These questionn ask qualitatively what the volcanic ;czarcs cre anc in a cruGe scri-quantitative way which cnes are present more frequently than others.

Ebasco has asked about the types of events and their relative frequencies.

They have inquirec in general wav atout the succession of activity from one vent to ancther, but they have not inquirec abcut sequences or patterns of successive eruptive esents.

12

3. hhat is the chronology of these events, i.e.

exactly when and how often have they occurred?

When did the volcano last erupt?

How old is the volcano?

Is the record p r in.a r il y historical or is it primarily based on geochronological information like radiometric aEcs of rocks, denarochronology, etc.?

What is the annual probability of an eruption, and the annual probability that an eruption will exhibit specific forms of activity (eg. lava flows, pyroclastic flows, etc.)?

Do the annual probabilities of eruption change with time (since the previous eruption),

i.e.

does the volcano have a memory of its previous eruption?

These questions help quantify the answers to question 2.

Ebasco has addressed all of these questions, though touching only lir,htly on the last question.

4 What is the probability that specific areas will be affected by each type of volcano hazard?

Which areas have been affected by specific volcanic events and how often?

What are the areal distributions of various types of deposits, for specific eruptions and for the eruptive history as a whole?

Ebasco has addressed the first question in some detail, but has addressed the second two questions only in general terms on a reconnaissance geologic map.

5. Have there been any changes in the composition of the magmas or, more directly, any changes in the eruptive style of the volcano which might incicate a trend of activity?

For example, have the magmas been increasing in silica content, and 13

a the eruptions been trending toward more infrequent but explosive types?

Ebasco has addressed both the chemical and eruptive style questions, though a lack of detailed cata has restrictec discussion to a general, long-term view.

With the few exceptions noted above, Ebasco and their consultants have asked a good, complete set of questions for evaluating volcanic risk.

The next section (and the bulk of this review) consicers the techniques used to answer the questions, and the conclusions proper.

14

i III. Approaches aic Answers Evaluation of volcanic hazards is a relatively simple matter where there are long historical records of activity and short intervals between activity; one can then say as a first approximation that future volcanic activity will likely be much the same as past activity.

Where historical records are short relative to the intervals between activity, one must turn to a variety of other pieces of information, including comparisons with similar volcanic areas elsewhere, and determination of the volcanic history of the cone from a study of its deposits.

Ebasco has used both approaches, including some specific techniques not previously applied to volcanic hazard studies (eg.

refinement of tectonic models to differentiate eruption probabilities at different vents; statistical analysis of Eeochrononological data), and they have also avoided other specific techniques which are generally employed in volcanic hazard studies (eg. detailed geologic mapping and the use of widespread, distinctive airfall layers in establishing a detailed stratigraphy and volcanic history of a cone ("tephrostratigraphy" anc "tephrochronology"); extensive use of C-14 for dating young volcanic deposits; preparation of maps showing rnes of relative risk for various volcanic phenomena).

(Tb-.nP.iques may be traced back to classic volcano studies cf Jhc: c.q.nsson (1967),

Nakamura (1962) and Aramaki (1965).

Steen-McIntyre (1977) describes current tephrochronological techniques in detail, and 15

n'alker (1974) anc crandell anc Mullineaux (1975) cite volcanic hazard studies which employ these techniques.) Most of the Ebasco approach consists of generally accepted procedures in volcano stucies (eg. identifying the different types of deposits present, determining the general areal distribution of different types of deposits, and determining radiometric ages).

The question of standard vs. non-standard techniques is not by itself important -

- the important questions become: If one uses non-standard techniques, does the data collected allow for the same detail and c;nfidence of interpretation, are the new methods of interpretation scientifically souno, and are the results consistent with volcanologic " common sense" (determined in part by the use of standard techniques)?

16

A. Ceneral Volcanic and Tectonic Setting Section 2.5.1.1.6.1 of the PSAR is a summary of the tectonic setting of the site region.

Except for a reference to u n su bs t a n'. ia t ed subduction along the West Luzon Trough and a distinction of volcanic domains which will be questioned below, this section is a commendable synthesis of thought for a tectonically complex area.

I think special note should be given to the discussion and diagram of seEmentation of the present subducted clab and overlying plate (p. 2.5.1-19 through 21, Fig.

2.5.1-11 and FiE. 2.5.H.2-80); this is an excellent example of segmentation of an island arc as the separate " fingers" of a subducted slab descend at slightly cifferent an51es and rates.

An understanding of this segmentation leads to a much improved understanding of seismicity and explosive volcanism in island arcs; the reader may consult Stoiber and Carr (1973) ano Ranneft (1976) for a general discussion of segnentation anc Carr anc Stoiber (1977) for a ciscussion of cestructive earthquakes related to this segmentction.

The Iba, San Antonic, Manila Bay and pcssibly Taal Fracture Zones are segment bouncaries.

Segmentation and associated stri<e slip movements parallel to the volcanic chains may also play a role in the development of graben fault valleys teneath or just inland from the volcanic chain (eg.

the Central Valley boundary fault system running dcwn the east side of the upliftec Zambales ultramafic (ophiolite) complex and the Bataan Peninsula)

Segmentation also seems to create offsets in volcanic chains and unusually explosive volcanoes may lie 17

preferentially on segment bouncaries (eg. Taal?).

The Napot Point site lies on the southwest flank cf Mt.

Natib.

Mts. Natib, Pinatubo and Mariveles form the volcanic chain of Ebasco's Seismic Zones III and IV.

The chain of Zones III and IV is but one part of the larger, segmented chain of volcanoes associated with eastwarc dipping subduction along the Manila Trench.

To the north there are several active volcanoes in the Sabuyan and Eatanes Islands; to the south is Taal, also recently and frequently active.

Differences in the frequency of volcanic activity from one segment to another may reflect differences in the dips and relative rates of subduction of each segment.

Frequent volcanic activity in the Batanes and Babuyan Islands suggests that that segment is more active than the Zone III anc IV segments, but I see no reason to suppose that volcanic activity en Eataan is "now waning" or "will probably cease in the geologically near future, if it is not already over" (p. 8, "Evicence substantiatin2...").

To the contrary, the documents seem to show that the Pinatubo-Eataan Peninsula volcanoes simply have infrequent eruptions in human terms (between 1 and 10 per thousand sears, based on the historical record and C-14 dates),

and since seismic activity shows that subduction is still continuing we should expect the volcanism to continue as well.

18

E.

Credibility of Renewed Eruptive Activity on Mt. tiatib

1. Detailed geologic maps and stratigraphic colur.:ns In order to understand the Eeologic history of a volcano, and hence its volcanic hazards, one needs a detailed, flow by flow, bed by bed geologic rap and stratigraphic column.

These must cover both the cone proper and surrcunding areas of lesser topographic relief.

Detailed geologic maps anc stratigraphic columns are the fundamental tools upon which all samplinE must be based (geochemical, geochronological, paleomagnetic, sedimentological, etc.), to which all analytical data must be referenceu, and from which all inferences about the behavicr of that volcano draw their basic strength.

Healy (1963) anc Crandell and Mullineaux (1967, 1975) stress detailed mapping and stratigraphic studies, and the same techniques have been wicely used in Indonesia, Japan, anc other areas cf active volcanism and high population density.

Eooth (1977) provides a popular description of the same approach, emphasizing studies of deposits from single eruptions.

Ebasco has moved in the same direction but with reconnaissance mapping (PSAR Figs. 2.5.1-42,A,2 ano 2.5.1-43) anc a generalized stratigraphic column (PSAR Fig. 2.5.1-44), rather than detailed maps and columns.

If the sparse sample locations shcwn in Fig. 2.5.1-29 are at all representative of the traverses mace by the Ebasco geologists, it is clear why they have not provided detailed maps for Nt. Natib.

They note quite correctly that stratigraphic correlation within volcanic centers is 19

difficult, because " individual units are laterally discontinuous and few contain distinguishing features" (f3AR p. E.5.1-56), but they conclude incorrectly that such correlation is " impossible".

It has been this writer's experience in the Philippines, and I believe the experience of all others who make a concerted effort to do so, that one can in fact make detailec, flow by flow, bec by bed maps of volcanoes and establish nearly complete stratigraphic control for all units.

When stratigraphic columns include both the cone proper anc surrcuncing areas of lesser topographic relief (where erosion of airfall ano other pyroclastic deposits is less severe), one has the best possible chance of deciphering the eruptive history of a volcano.

Detailed mapping ana stratigraphic study is clearly within the state of the art and within the capability of the Ebasco consultants, and their failure to do such work severely limits the usefulness and reliability of all analytical data and the confidence we can place in their evaluations of volcanic hazards.

Specifically, without careful mapping and stratigraphic control, Ebasco can only identify

a. the youngest sample collected (not necessarily the youngest volcanic products of Mt. Natib),

b.

intervals between the ages of their samples (not necessarily, inceed not even probably the actual repose periods (actual periods between successive eruptions )), and c.

the range of ages of their samples (not necessarily the actual range of ades of exposed rocks on Natib, and almost 20

certainly not including the oldest, now buried rocks of Mt.

Natib).

On the basis of their reconnaissance mapping they have no way of cetecting, with any degree of confidence, a.

actual repose periocs b,

any eruptive hiatus

c. periods of increased or decreased frequency of activity, d.

short-term variations in eruptive style (eg. over 10's, 100's or 1000's of years),

e.

any discrepancy between the age detern inations and actual stratigraphic position,

f. which dated samples, if any, are from the same eruption,

g. which samples are from " major" pyroclastic eruptions" or "significant volcanic events".

Except for the youngest pyroclastic layer, if it is indeed one layer, no evicence is given for thicknesses, areal extents and hence volumes of individual units and/or products of single eruptions, nor are grain size (and shape) analyses presented for the various pyroclastic units. These are standarc procedures for interpreting the character of past explosive eruptions.

h. which samples came from which vent.

Geographical proximity to a vent does not necessarily r.ean that a sample came from that vent, especially in the case of pyroclastic rocks.

I note the inclusion of many questien marks in Table 2.5.H.2-14 7 out of 17 samples from.Natib could not be unequivocally assigned to one vent.

Samples from Center 3 are reportedly recognizable on the basis of geographical proximity to that vent 21

and the distinctive inclusion of lake seciments in one pyroclastic unit.

i. which sar.ples represent the calcera-ferning eruption (s),

what the nature of those eruption (s) were, and when said eruption (s) occurred.

Summit collapse in andesitic volcanoes to form calderas is almost invariably associatec with major explosive eruptions of the same order of magnitude or larger than the eruption of Katmai.

j. what is the relationship between the,etivity of vent 3 and the inferred magma or hot rock teneath the summit caldera?

Lid an eruption from vent 3 drain some of that ragma and cause the summit caldera collapse?

These questions are all critical to the question of volcanic hazards of Mt. Natib, and they remain unanswered.

The approach which Ebasco has taken (and which all tco n.any other workers in volcanic geology take as well) is to put the cart before the herse --the sampling, analyses and interpretation before the detailed mapping and stratigraphic correlation.

Even the brute force ap--

ah with an unusually large number of samples (several hundred for Natib alone) would not provide all the information which proper mapping and stratigraphic study would provide.

One must indeec recognize that geologic mapping on volcanic cones is difficult, anc even more cifficult given heavy tropical vegetation, box canyons and repcrted guerrilla activity, but one must also admit serious limitations in the cata 22

without said mapping.

If uncertainties remain after such mapping., then one must also indicate how these uncertainties in stratigraphy carry over into incertainties in interpretation.

In neither case should one pretend that one Knows any trore about the volcanic history of a cone than one knows atout the detailea areal distribution and stratigraphy of its products.

23

4

2. Geochronological procedures The fundamental failure to make detailed geologic maps anc stratigraphic columns has been discussed above.

This means that radicmetric dates have been made on essentially rancom samples without stratigraphic context or control, rather than on the youngest, oldest and selected Intermeciate units in the stratigraphic column.

The single exception to this random selection is the young pyroclastic blanket on the east side of Natib, discussed below.

Eased on random sampling, Ebasco provices 27 radiometric ages and one fission track age for Natib rocks, plus approximately 70 radicmetric ages for the reniainder of the site region.

Relevant paleomagnetic data is also presented.

Carefully handled, this woulc be a windfall of absolute age information, and it coes provide a valuable general time framework for the activity of Natib and the surrounding volcanoes.

These dates inherently uncerestimate the actual age of the cones, because the samples are from the surfaces rath than frcm the interiors of the cones, but that fact does detract from their general usefulness in discussing the vs

.nic history of the region.

In detail, however, I note the following:

a.

In the date of Figs. 2.5.1-29 E-I, no C-14 ages have been given for Mt. Natib.

(I have not seen Table 2.5.1-3 referred tc on p. 2.5.H.1-14 of the PSAR.) C-14 is the method of choice for dating volcanic deposits less than approxiaately 6C,000 years 24

old, and Ebasco has used the method to demonstrate 635, 2350 and 8CSC year old deposits on Mt. Pinatubo.

Carbonized wood is reported in the hapot Pt. tuff (PSAR p.

2.5.2-65) and in silts deposited before the " youngest" eruption of Natib (p. 2.5.1-53);

I co not find reference to dates on these samples.

Even if these should prove to be older than the working range of C-14, this would be valuable information; a C-14 age cetermination of these samples is the obvious starting point for an age determination of the Napot Pt. tuff, and yet it does not appear to have been done.

Additional C-14 dates st.ould be obtained for the young pyroclastic blanket on the east and for as many other " young" deiosits as possible.

Charcoal and wcoc are not well preservec ci, alcanoes in tropical climates, but they are occasionally preserved -- one simply needs to look carefully for small pieces as well as large pieces.

b. 2380 -230Th and the rels,ed fission track method are potentially useful for dating volcanic rocks between a few thousand and approximately 250,000 years of age.

Only one unit from Natib, a young pyroclastic blanket on the east flank, was cated by the 2380 -23CTh method.

Three samples were taken from one location and two from a nearby location.

These and two additional samples were also datec by the fission track method.

Ebasco discounts all of their 23EU -230Th anc fission track ages (PSAR,

p. 2.5.H.1-13) and singles out sample 776-5CW as a sample which has not remained a chemically closed system (Fig.

2.5.H.1-2).

However, in their later response to the IAEA Safety Mission, 25

entitled "Volcanisn and Volcanic Hazards", Ebasco cites this same sample, with a new correction factcr, as a rellcble sample (reportec ace =69,0C0 years) which has remainec a closec systcr (p. IE, 'n.:c ccterrinaticn ot the youngest r.aterial on

't.

Natib').

The latter reference to sample 776-5E'.s states that it is "unlike the samples discussed in Appendix 2.5.H.1" and yet it has the exact same sample number.

Was the new sample recollected and if so, where are the data to show that it has remained a closed system?

If the sar.ple was not recollected, there appears to be a serious contradiction here.

(Note: The youngest pyroclastic blanket is describec as " gray, poorly consolidated, pumice-rich pyroclastic deposits" (cf.

Volcanism and volcanic hazards,

p. 82).

Chemical analyses show this to be andesitic (Sio2:59%).

I wonder whether unconsolidated andesitic pumice-rich pyroclastics would remain gray in this climate for 69,000 years.

In my experience, such units in the Philippines weather rapidly anc turr yellowish brown within a few tens or hundreds of years.)

Cne additional sample was dated by the fission track method (376-1G, Mt. Pinatubo), and its age was originally given as 9000 3 9000 yrs.

The same eruption was dated using C-14 with a result of 2330 years.

Has the fission track age of 376-1C also been revised, allowing comparison with the C-14 age?

c.

The principal dating technique used in this stucy was E-Ar.

When dating andesites, the whcle rock is generally ased rather than mineral separates anc this appears to have been the 26

procedure here.

The lower limit of this method under ideal circumstances is approximately 50.0n0 years; the lower limit selected by Ebasco for reliable K-Ar sges on Natib rocks is approxic,ately 180,000-200,000 year.', below which contamination with atmospheric argon introduces sericus error.

Therefore, no reliable K-Ar ages below approximately 20G,000 years are to be expected, even if the samples are in fact younger; the K-Ar method can only say, as in the case of sample 476-3W (Pinatubo) and prcbably samples 975-9W (Pinatubo) and 376-1G (Pinatubo, see discussion aLove), that they are simply less than 200,000 years old.

Two lesser points atout K-Ar cates should also be made:

a. The discussion about sample 476-3h and a hot lahar (p.

2.5.H.1-5) is misleading.

There are no hot lchars outside of eruptions, and so even if that particular boulcer has been reheated, an eruption at that aate is i n.p li e d.

b. The deposits of samples 376-1c ana 975-9h (both Pinatubo) were also datea by C-14 (Fig. 2.5.1-29C) anc found to be 2330 and 8050 years old respectively.

Why was there any question (p.

2.5.H.1-5) about "very recent argon loss"?

There is no detectable radiogenic argon because there hasn't been enough time for it to form, not because of "very recent argon loss",

d. Correlation between paleomagnetic vectors and K-Ar ages of lavas is a useful exercise and seems to show a counterclockwise rotation of ceclination from the Gauss normal epoch (3 3-2.4 my) through to the present.

If one then had 27

paleomagnetic vectors on undisturbec samples without radiometric dates, one could guess the paleomagnetic epoch (and event) during which the samples were formea.

de Eoer ana Ebasco have attempted such an inference for the Napot Pt. tuffs (p. 2. 5. 11. 3 - 1 1 ) but I cannot accept their results because they indicate cn the preceding page that both inclinations and declinations in the tuffs are disturbed, possibly by rotation of magnetic vectors during compaction.

It is circular to correct the Napot Pt. tuff inclinations to Gilsa normal event inclinations and then decide that the corrected declinations indicate a Gilsa age.

There is no documentation of a polarity sequence at Napot Pt.,

eg. an overlying reversely polarized unit, to suggest that the normal polarity of the Napot Pt. tuffs could not be of the present day Erunhes normal epoch.

A minimum age for the Napot Pt. tuff basec on offshore corals is only 6200 years.

Once again the need for proper, careful stratigraphic work is emphasizea.

Three important summary points can be made:

a.

There is no reliable age for the deposits of Napot Pt.

b.

There are no dates for any samples from the western flanks of Natib (incl. the Napot Pt. tuff) with the C-14, 238U -

230Th or fission track methods, i.e. with techniques appropriate for rocks less than 200,000 years old.

In the absence of proper stratigraphic control to show that ;he youngest rocks have in fact been dated, ano especially in the presence of undated, paleomagnetically normal samples, one has no way to evaluate whether or not there are rocks on the western or southwestern 28

slopes youncer than 200,000 years old.

c. If one has no C-14 dates for Natit anc no reliable dating method for rocks between 60,000 and 200,000 years old, one cannct expect to cocument much volcanism on !;atib in the last 200,000 years.

Indeed, how can we even know whether the reported waning of volcanism during the past 0.5 my is real or merely an artifact of the datir methods used?

Furthermore, how can we discuss the youngest volcanisn of Mt. Natib -- it might have been 50,000, 5000 or 500 years ago.

29

3. Annual probability of an explosive eruption of Mt. Natib Pt. 1 Historical recoras of volcanic eruptions in similar tectonic settings The frequency of volcanic activity along convergent plate boundaries is in the broadest sense proportional to the rate of convergence.

Where convergence rates are high, volcanic activity is high (eg. Japan) and where convergence rates are low, volcanic activity is low (eg. the Cascades).

This broad correspondence is complica*.

by the thickness of the crust, the compositions of the magmas, local and regional geologic structures, the degree to which sediment is subducted or scraped off a subductec plate, anc a myriad of other factors.

Convergence across the Manila Trench boundary is estimated at approximately 5 cm/yr, compcrable to that in Central America and much of Indonesia, and so we might reasonably expect the frequency of volcanic eruptions in these regions to also be comparable.

The rate we need to calculate is no. of eruptions /my/ volcano, for the set of all potentially active volcanoes in a region.

In

" Evidence substantiating the incredibility of volcanism on the west flank of Mt. Natib, and the assessment of volcanic hazards at Napot Ft.",

Ebasco cites reasonable anc impcrtant criteria for selecting volcanoes with a potential for renewec eruption:

a.

situation in a volcanically active islanc arc or continental margin environment (geologically similar to the Bataan Peninsula),

30

h.

structure of a larEe composite cone or calcera with well preservec volcanic morphology, or

c. volcanic products eruptec more recently than about 5 million years ago.

Following essentially these criteria, the Cata Sheets for post-

ticcene Volcanoes of the World show 399 potentially active volcanoes in Indonesia and the Philippines, which is the figure used by Ebasco.

To determine the rates of eruptions worldwide or in a given region we can refer to the historical record of volcanic eruptions.

There are severci excellent ccn4pilations of historical information; the most complete of these for the present purposes is a new compilation by Simkin et al (in press) of the Smithsonian Institution.

In aadition to listing dates of eruptions, Simkin's compilation gives standarc information about the eruptions and includes an estimate, termed the VEI (Volcanic Explosivity Index), of the relative explosive magnituce of each eruption (Newhall and Self, in preparation).

The latter is important in evaluating potential volcanic risks and in keeping different " magnitude" eruptions separate in statistical calculations.

VEI=0 eruptions are non-explosive, VEI:1 eruptions weakly explosive, VEI:2 eruptions weakly to moderately explosive (many eruptions for which we have little historical information are classed as VEI:2 eruptions by default), VEI: 3 eruptions moderately to highly explosive, VEI: 4 eruptions highly explosive 31

and VEI:5 errotions cataclysmically explosive.

The 1968 Vulcanian eruption of Fayon Volcano was a VEI: 3 eruption, the 1963 eruption of Agung Volcano and the 1965 eruption of Taal Volcano were VEI:4 eruptions, and the 1912 eruption of Katmai was a middle-of-the-range VEI:5 eruption.

Exact criteria for estimating VEI's are given in Table 2.

I have prepared a summary of historical information about eruptions in the Philippines and Indonesia since 1500 A.D.

(Table 1,

data from Simkin et al.,

in press).

This figure shows the numbers of reported eruptions through time, broken down by VEI class; it also shows the numbers of eruptions with reportea lava flows and nuees ardentes (pyroclastic flows).

Note a general increase in the number of reported eruptions through time from 1500 toward the present, with an apparent decrease in the number of VEI:2 eruptions in the 1940's to 1960's.

Eoth the general increase and the recent decrease in VEI:2 eruptions are probably artifacts of historical reporting rather than real increases or decreases.

In general, the historical record of volcanic eruptions deteriorates backwarcs in time, and although the recorc is worst for small eruptions and best for large, more memorable eruptions, it ultimately deteriorates for all scales of eruptiens.

The recent decrease in VEI:2 eruptions is largely if not wholly an artif act of borld '4ar II and then of better reporting (allowing us to have fewer " default" VEI:2's anc more eruptions assigned to their proper VEI category).

32

A statistical method for detecting or confirming incomplete reporting is given in Stepp (1973).

As long as reporting is complete and eruption frequer.cies are more or less constant, eruption reports should form a Poisson distribution, and the variance of the mean eruption frequency should be inversely proportional to the number of observations (years, decaces, etc.)

in the sample.

If one assumes a Foisson process, a log-log plot of the square root of the variance (standard deviation) vs.

cumulative time before the present will form a straight line with a -0.5 slope, for the time span over which reporting is complete.

When the reporting becomes incomplete, though, the cumulative n.ean eruption rate (x) decreases and thus the standarc deviation

( f A [V n

) drops below what it would be for complete reporting.

The date at which the points on this plot drop below the straight line is taken to be the earliest date of complete reporting.

Fig. 1 of this review is such a plot, analogous to fig. 4 of Stepp (1973), for historical volcuM sm in the Philippines and Indonesia.

In this figure I have distinguishec different "explosite magnitudes" of volcanic eruptions just as Stepp has used earthquake intensities.

If the apparent increase in the frequency of volcanic eruptions thrcagh time has been a real increase in frequency, we should expect all VEI classes to show simultanecus changes in slope in this plot (Stepp, 1973).

They do not, which suggests that alternative explanaticn -- that the apparent increase is an artifact of improved reporting.

For a perfect confirmation of incomplete reporting, however, one woulo 33

expect small VEI:2 eruptions to have the shortest period of complete reporting, anc the larEe VEI=5 eruptions to have the longest.

The patterns for VEI=3, VEI:5 and "sericus" eruptions (see definition of " serious" below) show this effect, but the patterns for VEI=2 and VEI=4 eruptions do not.

The former curves indicate incomplete reporting; I believe the latter two curves are a problem in the VEI classification, with the default assumption and recently improvec quality cf rescrting (see note alcve) n.a s k i n g incomplete reportirg for VEI:2's ano exaggerating incomplete reporting for VEI:4's.

Additional support for the idea of incomplete reporting comes from the 1960's, when the Volcanological Society of Japan began systematic reporting of eruptions.

The reportec frequency pyroclastic flows and VEI=3,4 eruptions increased with improved, more detailed reporting, and if we postulate that the 1960's were not an unusually " active" decade (see Table 1), then we can say that reporting of pyroclastic flows and VEI=3,4 eruptions in previous decaces was incomplete.

This can be tested for the 1970's shortly; preliminary counts show about the same overall level of activity as in the 1960's -- a few more pyroclastic flows and VEI: 3 eruptions but fewer VEI:4 eruptions.

To correct for the deterioration of the historical recora through time one can either assume that current reporting is more cr less complete anc extrapolate back from it, or one can run a multiple regression analysis with time as an indepencent variable, correcting each reported frequency with an accition 34

frca the regression equatior;.

I have used both approaches and obtain similar corrected values.

The fellowing list shows reported values for the 160 year period 181G-1969, compared with values frcm a recent 20 year perica (1950-1969) n.u l t i p li e d by 8.

reported reportec calculated 1810-1969 195C-1969 1810-1969 VEI=5 eruptions 3

0 0

VEI:4 eruptions 6

4 32 VEI=3 eruptions 32 8

64 VEI=2 eruptions 6C5 7C 560 VEI:1 eruptions 34 5

40 VEI:0 eruptions 13 6

48 Total eruptions 693 93 747 VEI: 3,4,5 eruptions 23 9

72 w/ pyro, flows PEI:1,2 eruptions 38 7

56 w/ pyro. flcws Eruptions with 108 17 136 lava flows In Appendix 5 of " Evidence substantiating...", "cEirney classes an eruption as "large-scale" or "sericus" if in said eruption 35

a.

ash or pumice was eruptec in such quantities that at least 2 m would have accumulated at a distance of 5 km from the

source, b.

explosive energy released in the eruption exceedec a value of 10 ergs, or c.

violent directed blasts or glowing avalanches had a destructive effect at the base of the volcano.

The first of these criteria will only be satisfied in exceptionally big, explosive VEI:4 and 5 eruptions.

There are very few eruptions for which ash thicknesses have been reported, so there will be even fewer reported cases satisfying criterion "a" than the already small number of eruptions likely to satisfy that criterion.

The second criterion is much more conservative, but again suffers from the problem of reporting.

Explosive enerEy release is almost never reported in older records, nor even the volutes of tephra on which the ener5y calculation is based.

If we compare the explosive energy release of recent eruptions against the VEI, thcugh, it seems that all VEI 3,4 and 5 eruptions ano some VEI:2 eruptions will be incluced.

As shown by Yokoyama (1956, 1957), a " total" or " explosive" energy release of 10 ergs corresponds to a mocerate-size explosive eruption, of "Tsuya magnitude" III-IV (Tsuya, 1955) ano an ejecta volume of 6

7 7 (VEI:2 range).

If we use criterion "b"

literally, I 10

-10 m

believe we would include many eruptions which few volcanologists would consider "sericus".

The third criterien, violent cirected 36

blasts or glowing avalanches which had a cestructive effect at the base of the volcano, is perhaps the closest to the lower limit of " serious" eruptions, if one rephrases it to incluce all eruptions with reportec pyroclastic t~1cws or " glowing avalanches", irregardless of whether destruction was reportec at the base.

Destruction is only reportea where there are settlements and cultivated lands.

What we really need is data on the maximum distances travelec be each reported pyroclastic flow, but that informaticn is virtually never incluced in historical records; consequently I think it would be much safer to assume that any reported pyroclastic flow coula have reached the base of its respective volcano.

McBirney's three criteria are, therefore, acceptable with the last-mentioned modification; indeed, the second criterion is very conservative.

The problem now becomes: How can we apply these criteria to the historical record, where ash thicknesses, eruptive energies and tephra volumes are rarely reported?

Translating the three criteria into terms of VEI values and reported pyroclastic flows (both reported in Simkin et al.,

in press), the criteria might now include eruptions

a. with an estimated VEI cf 3, 4 or 5, corresponcing to Vulcanian or more explosive eruptions with repcrted or probable column heights in excess of 5 km above the crater and/or tephra 7 3 volumes in excess of 10 m Most eruptions with a VEI of Ereater than or equal to 3 are capable of generating pyroclastic flows; indeec, about half of the VEI:3 eruptions had reported 37

pyroclastic flows and the latter weren't even widely recognized until after 19C2. or b.

with reported pyroclastic flows but with a VEI less than 3

For the san.e 160 year period considerec above, McEirney counts 12 " serious" events in Indonesia and the Philippines.

I count at least 41 using translated criterion "a" anc another 38 using translcted criterion "b",

for a total of 79 " serious" events.

There is always a question of how to decide when one eruption stops and another begins, and this might account for some difference in our counts.

I have required a lull of 1 year before callinE renewed activity a new eruption; some workers use a period of 2-3 months, others might use a period of several years.

Eoth McEirney anc I have ccunted eruptions, not individual events within eruptions -- otherwise the nur.ber of pyroclastic flows would increase dramatically, as there are often several pyroclastic flows per eruption.

There does not appear to be any substantial difference in what we are calling " serious" --

several eruptions in the shorter, Ebasco/McEirney list (eE. Awu 1896, Merapi 1920-21 and 1930-31 and !!itok-Hibok 1951) satisfy translated criterion "b"

only, i.e. are smaller and less explosive than VEI: 3,4 or 5 eruptions, but do have pyroclastic flows.

Five more on the short list are only VEI: 3 eruptions.

Extrapolating back from the data for the period 1950-1969, i.e.

multiplying the numbers of events in that 20 year pericd by 8, I calculate 96 eruptions with VEI: 3,4 or 5, and an additional 56 38

smaller eruptions uith pyrcelastic flows, for a total of 152

" serious" events.

This correcticn for inccmplete historical reporting increases the dif f erence in our estir.ates f rom 7-folc to 13 fold.

This cifference between cur estimates appears not be a cifference in the types cf eruptions we are considerinc

" serious", but rather in the completeness and consistency cf the lists.

Still another approacn to the question of eruption probability is to lock at the total number of potentially active subducticn zone volcanoes (post-Fiocene but withcut historically reported eruptions), anc the number of these which have awakened curing the past SC, 100 and 2CC years.

Ecr each volcanc one needs te estimate the tin.e elapsed from the beginning of recorded observations to its reawakening.

The estin. ate would then be in units of number of " reawakening" eruptions per year cf pcssible recording per pctentially but not reportedly active volcano.

As an example of the data, consider the Japanese volcanc Cn-Take, which erupted this past October for the first t i n.e in more than 1CCC years of recorced histcry.

For this eruption, one woulc adc cne " reawakening" anc 1CCC years to the respective cua.ulative surs for either the Japanese region er for a worlcwice sanple.

For potentially active volcanoes which have not yet erupted in historic time, one wcule adc no eruptions but cnly the number cf 39

years for which recoros of that volcano have been kept.

This snculd be done for both the Philippines-Indonesia region anc for subcuction zone volcanoes worldwice.

Cne could also ask what prcportion of the " reawakened" volcanoes awake with VEI 5, 4,

3, etc. eruptions.

A quick check of the 80 most explosive eruptions since 1700 shows thst 19 of these were from volcanoes thought to be extinct; the most recent of these were from Lamington (1951)

(which was not even known to be a volcano) and Bezymianny (1955-56).

A final historical approach was suggested by Walker (1974),

who considered 180 volcanoes (worldwide) in areas havint reasonably complete historical records of 5CO years or longer.

He found that the median volcano erupted once every 220 years, 20% of the volcances erupted less than once every 1000 years, only 2% of the volcanoes erupted less than once every 10,000 years, anc only 1% of the volcanoes erupted less than once every 50-60,000 years.

This is, of course, based on statistical extrapolation, and it shoula also be noted that halker has included a wice range of volcanoes in his sample, has counted every eruption (rather than every "sericus" eruption), and has used the IAVCEI Catalogue of Active Volcanoes anc Solfatara Areas (r ather than the larger sample taken by Ebasco from the Cata 2heets for post-Piocene Volcances of the Wcric).

Walker's results cannot be directly compared with the estimates by Ebasco and this writer of serious events / volcano / year, but his calculation does indicate that many, many volcances have 40

eruptions on the order of 1 per 10 to 1000 years, while very few erupt less frequently than once per 10,000 years.

It is this kind of general information that is addec to our " volcano 1cgic ccmtr.on sense."

41

Pt. 2: Statistical analyses of geochronological data for the site regicn

a. Stepp's Method As shown above, Stepp's method should have been usea in the first instance to analyze historical eruption records for the Philippines and Indonesia.

Had Ebasco cone so, they would have seen the deterioration in the record of historical volcanism, and the discrepancy between their rates and those of recent volcanic activity.

There is a loss of information ever in the best of geological records, through erosion, burial, mapping errors, etc., and there is certain to have been a major loss of information from reconnaissance sampling of a geoloEical record spanning several million years.

It is remarkable and incredible, then, that Ebasco interprets Fig. 2.5.J-3 of the PSAR to show that

...the data points for the Eataan Peninsula...suggest a fairly complete data set from the preser.t tin.e to a bout T=0.5 n.y,

with an eruption rate of somewhere between 5 aad 10 per my."

I fail to see any basis for this conclusion, anc it certainly goes against volcanologic common sense.

hecall that a comparison of recent recurrence rates for erut. cions with lava flows (i.e. the types of erupticr; represented in the raciometric age dates) shows 17 such eruptions for Inconesia and the Philippines in the past 20 years, at a total of 399 potentially active vents, cr apprcximately 4200 u30C flows /my for the 2 volcanoes cf the Bataan Peninsula, vs. the 5-10 flows /my cited by Etasco.

If a) the combined volume of "ts.

"atit and Mariveles is in the orcer 42

of 10 " r 3 b) the volume of a typical andesitic lava flew is in the order of 10 m 3, c) the volume ratio of pyroclastics tc lava flows on tts. Natib and 'tariveles proper is in the order of 3: 1, and d) the ages of liatib anc Mariveles are in the orcer of 3 my, then together they might have erupted on average 500-1C00 lava flows /my.

This is a bit far from the 5-10/my estimatec by Ebasco.

Until r: ore convincing geochronological data is presented, _in careful stratigraphic context, to cemonstrate that these volcanoes really have been essentially inactive for the past 0.5 my, I think it wculd be wiser to assur..e that they have behavec much like other volcanoes in the Philippinc-Indonesia regicn.

Stepp's method works just fine where it is properly applied to historical data, but its application to the Ebasco gecchronological cata gives a result which is inconsistent with historical experience and volcano 1cgical common sense.

b. hickman's Method:

A differ (nr statistical technique was usec by hickman (1966) to evaluate the time dependence of eruption probabilities for a given volcano, and to infer differences in volcano " plumbing",

triggering mechanisms, etc.

Wickman considered single volcances for which complete historical reccrcs are available, calculated repose periods anc the time elapsec since the previous eruption, and then calculated the probability of another erupticn within the next month, two months, etc.

Contrary to the Ebasco assertion that this methoc anc Ctepp's can " provide meaningful results from incomplete age data" (PSAh, p.

2.5.J-1), Wickman 43

cpecifically avoids incomplete historical data (Wickman, 1966, p.

335).

The extrapolation of e c methcd to a whole recicn from single volcanoes is not unreasonable, but the incor..pleteness of the data sinply precludes any meaningful results about actual eruption rates or repose periocs.

Ebasco calculates a rate of 20 eruptions per my (1 per 50000 yrs) for the Cataan Peninsula anc 30 per my (1 per 33333 yrs) for the site region, and calls these eruption rates (p. 2.5.J-3), but in fact these are little more than a statistical summary of their sampling program!

Although Wickman's method cannot be applieo to the present data, it does raise another important point, namely that probabilities of eruptions are age-dependent.

If a vc1cano has a

" memory" of its previous eruption, the probability of its next erupticn will vary with time.

Some volcanoes becoroe more likely to erupt as their repose period drags on, some become less likely tc erupt, others show con. plex cecreases and increases in their age-dependent eruption rate, and still others show no ' memory' at all.

One would need very complete geochronological cata for

!.'atib to evaluate its age-dependent eruption rate --probably more complete than the preserved, exposed deposits themselves.

About all one can do in this case of prehistoric volcanism is to realize that eruption rates and probabilities are usually age-cependent, and that estimates based on an assumec Foisson distributicn of erupticns from a snall number of vents are likely to be in error.

If the age-dependence is short-term or long-term relative to the recurrence rate of the specific volcanic event in U4

question (eg. rajor explosive eruptions), then the error may be tinor; if the age-dependence is of the same crcer as the recurrence frequency, the effect may be significant.

In sumnary, the Ebasco reports estin: ate the probability of renewed major explosive activity at Mt. Natib in two ways: from McEirney's analysis of historical data and from application of two statistical techniques to their geochronological cata.

The

~Y first estimates are about 1.88 x 10 serious events / volcano / year

-3 (or 1 per 5319 years per volcano), vs. about 2.38 x 10 serious events /vcleano/ year (1 per 42C years per volcano) estin.atec by this writer.

The differences lie not in a cerinition of

" serious" but rather in the completeness ano consistency of interpretation of the histcrical records.

The secono estimates

~

are approximately 3 x 10 serious events /yr (or 1 per 33333 years) at each of the two Bataan volcanoes, basec indirectly on dates of lava flows, and again are in sharp contrast to this

-2 writer's estimates of 2 38 x 10 serious events / volcano / year.

The principal problem with the statistical estimates seen.s to be in failure to ackncwledge the e;treme incompleteness of the geochronological data set.

t does little good to acknowled6e possible or partial incompleteness of the data (eg.

p.

2.5.J-1,2) anc then proceec as if the incompleteness can somehcw be igncred.

The Ebasco repcrts discuss " eruption recurrence frequencies" and

" repose periods", when in fact they are discussin6 pericas between their dated lava flows anc an approximation of their temporal sampling censity.

Certainly the histcrical reccrcs are 45

rore helpful than the geochronological data, but even they neea more careful analysis than has been made.

With such a discrepar.cy in estin.ates, we might re-consicer the simple calculation made by Walker (1974).

Recall that he found that the r.edian volcano in his sar.ple erupted on average once every 220 years, and only 20 erupted less than once every 10,000 years.

We are being asked to believe that Natib is one of the select few %

cf the world's volcanoes with such long repose periods.

I would be more prepared to accept that if I had n. ore confidence in the geochronological and statistical procedures used.

46

4 Thermal springs Thermal springs in volcanic terrain, perhaps withcut exception, occur where there is near-surface hot rock anc/or a fault which serves as a passageway in a grounawater ccnvection system connecting the surface waters with hot rock at depth.

The thermal springs in the Natib calcera are concentrated along arcuate faults of the caldera.

Calderas in andesitic cones typically form by collapse into vacated space after violent eruption of large volumes of magma, and/or by foundering of the upper part of the cone into shallow magma bcdies.

There is every reason to believe that the Natib caldera thermal springs are along faults reaching to hot rock at mocerately shallow depths (in the order of 1 km or less).

Thermal springs are commonly used to distinguish tetween extinct volcances (without thermal sprinEs) and those which are merely dormant (with thertal springs).

(In this classification the term " active" is reservec for volcanoes with historically recorded eruptions

-- cf. Macdonald in Eolt et al.,

1977, pp.

64-65).

One must note that there are thermal springs in many rerions which are quite independent of volcanoes per se, and also that thermal sprinEs are more common in the later life stages of a volcano than in its early stEdes; on the other hand, one n.ust note that many volcances with thermal springs arc known to be historically active, and it is this inferred heat source at cepth which leads one to suspect the possioility of renewea eruptive activity.

The Ebasco report acknowlecges "a scurce of heat in cr 47

below Mt. Natib Volcanic Complex at the present t ir.e " (FSAh, p.

2.5.1-35); r. ore details of the springs are given in Section 2.5.1.1.7 anc Fic. 2.5.1-30 The springs are not exceptionally hot thermal springs, but the actual tenperatures are a complicated function of temperatures at depth, grounawater circulation rates, Icwer limits of the circulation system relative to the hot rock, dilution from cool, near-surface groundwater (especially durinE the rainy season), etc.

The discussion of water chemistry and the implied temperatures ( PS A fi p.

2.5.1 46) is useful and does not indicate unculy high temperatures at depth (90 degrees C), but it is hard to know whether these relatively cool ten.peratures reflect relatively cool rock at cepth or n erely reflect circulation patterns or dilution that mask high temperatures at depth.

Fig. 2.5.1-39 shows the Cabayo Thertal ' Wells abcut 7-8 km E of the site.

Are these associated with F.t.

Natib?

Is there a fault in this area?

Is there a fault connecting TS-2 and the Spa on Mt. Mariveles, and if so, how far does it extend in the northwesterly projection of that line (directly towarc the Napot Pt. site)?

These are questions that need to be asked.

46

5. Geochemical and tectonic mocels Ebasco and Lr. Raglanc have made a useful contributior. to the body cf chen.ical data on Philippine volcanic rocks.

Furthermore, their documentation of three gecchen,ical provinces across the arc is reasonable and interesting in this regicn of complex tectonic patterns.

In detail, though, I have strong reservations about their mocel for two sulcuction zones, separate in time and space, and am convincec that their data for F.t.

Natib does not even suggest that this one volcanc has been built by subduction on twc proposea subcuction zones.

Ebasco presents a model in which n.agmas feedir% Centers 1 and 2 of "t.

hatit are generatec by east-cipping subduction associated with the West Luzon Trcuth.

The 1ccus cf subduction then shifts apprcxinately SC-CC km westwarc, to forr.. ar.c plunge downward frce the Manila Trench.

They base this hypothesis on two facts:

a.

that there might Le 2 trenches west of this part of Luzon (the inner of these, the K.

Luzon Trcuch, is a sediment-filled trouEh which is hypothesizec to be a former trench), and b.

there is poor spatial resolution between the western (tholeiitic) volcanic group and the central (calc-alkaline) volcanic group, inclucing "less than 10 km" separation on Mt.

Natib proper.

They state that the chemical gracient in silica-normalized K2C values frcm Center 1 to Center 3 implies a magma source for tholeiitic centers 1 and 2 which was apprcximately SC km shallower than the present subcuction zone (FSAR p.

2.5.1-59),

which frcn Fig. 2.5.H.2-E0 woulc te a scurce at approximately 25 49

km depth.

It is unusual for calc-alkaline magmas to be benerated at depths as shallow as 75 km and it is unheard of for tholeiitic cr any other magmas to be generated at 25 km depth alcng subduction zones.

Is the seismic contcuring incorrect, and/or is the conclusion of two separate subducticn zones incorrect?

The cata in Fig. 2.5.H.2-E 3 do not suppcrt the Ebasco interpretation.

There is too much age overlap between vents 1 and 2 and vent 3 to say that volcanism stcpped at one before the other began; indeed, clearly it aic not.

It is not meaningful to note the minimal overlap in ages between Centers 1 anc 3 but 16nore the fact that the Lees of Center 2 (sane domain as Center

1) overlap substantially with those of Center 3 Furthermore, where is the steep chemical gradient between Centers 1 and 3, 4-5 km apart, which Ebusco says requires their n.odel?

(ESAR, pp.

2.5.H.2 46 through 2.5.H.2-49)

In Fig. 2.5.H.2-82 there certainly is no simple progression of normalized K20 values frcm west to east acrcss the ::atib volcanic complex; instead, there is a cruce increase in normalized K2O from west to east with many individual reversals in trend.

The same fact is shown in my figure 2, which shows normalized K20 vs. distance across the are for all samples of the

!iatib complex (i.e. all ages), and for the perica 1.2 my to the present.

Note that ;he variability within Center I/3 is equivalent to the variability between Center 3 anc Centers 1 and 2.

I ara unable to draw any meaningful gecchemical gradient on my figure 2.

The tectonic model chosen by Etasco, anc usec to SC

cifferentiate volcanic risks at the summit and eastern vents of f:atib, appears to be based on a geochemical interpretation which is speculative at test.

FiE. 2.5.H.2-E0 shcws contcurs of silica-normalized K2C-across the arc; the suggesticn of a steep K2O gradient in the west, cecreasing in an eastward directicn (cf.

p.

2.5.H.2 43) is not apparent to this writer, in either plan view vs distance across arc or much less in vertical cross-secticn vs depth to the top of the subcucted s1cb.

An apparent east. arc cecrease in the cracient is seen in Fig. 2.> 5. l!. 2-E 1.

As I interpret the explanation or p.

2.5.H.2-44, the points represent averattc silica-ncrLalized K2C values vs fixec (center?) points of Erid squares projected to the A-A' line.

The site region was apparently divided into a grid of 5' x 5' quadrangles and the silica-normalized K2C of all samples within each square or quadrangle was averaged and projestec to A-A'.

If so, there are three important errors in Fig. 2.5.H.2-81, namely:

a. The figure uses sample locations rather than vent locations.

On a regional scale this is unimportant, but in detail it introduces error because samples, especially pyrcelastic samples or samples from distal ends of lava flows, may be several km or more in any direction from their vents.

Has Ebasco usec sample locations in defining their " steep dracient" across ':atib?

b.

The data points are from several segments, anc thus include samples frcm quite different " distance across arc - depth 51

to the top of the subductec slab" regimes.

The apparent change in the "K2O - distance across arc" gradient in Fig. 2.5.li.2-c1 may simply reflect different segments with different gradients, rather than a chanEe in the gracient for any single segr.;ent.

'l hat we need is a plot for each segment, and especially for the seisn.ic cne IV segment.

c.

The cata represents several cifferent time periods, including times of supposed subcucticn along the ' est Luzon Trough and subduction along the Manila Trench.

The case for subduction alcng the former, succeecec by subduction along the latter, would be better evaluatec by separate plcts fcr the two time periods.

If the " changing gracient frcm west to east" of Fig. 2.5.H.2-81 disappeared into two sc.octh r,radients in the two plots, there wculd be a case for separcte subcucticn; if not, the case for separate subduction would be thrcwn out.

It r.ay te difficult to r.ake such a separation into two time periocs, because of the considerable overlap in ages between the western and central groups.

The geochen.ical anc geochrcnological data r..ight also be interpreted to show a.

two in.bricate subduction zones, in which two parallel subducted slabs move more or less sinultaneously downward.

The convergence at this Philippine Sea-Asia Plate boundary could be taken up at any given tin.e along either or both of these zcnes, and magmas could be generatec along both.

Such imbricate subduction zones appear to exist beneath :lindanao, dipping 52

Westward and with greater separation than proposec here, but subduction teneath tne western side of Luzon :r.ay also be cor. plicated... or, better still, b.

a single subduction zone, in which the eastward and downward progress of the slab results in the progressive development of tholeiitic, western group magnas

(" start" 7.2 my),

through calc-alkaline, central group magmas

(" start" 4.1 my), to alkaline, eastern group raEmas

(" start" 1.7 my)(Fig. 2.5.1-35).

Within this r.odel there coulc be significant variability in ncrmalized K2C vs. depth points, for whatever reasons the variability has developed in Center 3 and in the arcs stucied by Miyashiro, !!iclson and Stoiber, and others.

The arrangement of cones in this portion of Luzon and the broad chemical groupings suggest a single subduction zone, modified only by segmentation of the subductec slab anc overlying crust.

Given the wide variability of silica-normalizec EEC values in single vents, continuing debate over the origin and significance of correlation between said values and depth to the tops of subducted slabs, and the fact that Mt. Natib is similar to many other cones along single subduction zones, I finc the Ebasco. interpretation of two subduction zones unnecessarily complicated and the consequent distinction cf volcanic risk between Centers 1+2 and 3 absclutely unjustified.

Cne additional ncte re: geochemistry.

Emphasis is placed on the tectonic implications of the geochemistry, and very little ciscussion is given of the more immediate volcanological 53

u i r.p li ca t i on s.

'a b a t does the gecchemistry of the rocks through time tell us of the evolution of the Mt. hatib magr.as?

Stucies of chemical changes through time on a volcano or volcanic con. plex can provice much information about the cifferentiation processes in the tragma, anc correlation of chemical changes with changes in eruptive style can lead to generalizations about the control of magma compositions on eruptive style.

Fig. 2.5.H.2-84 does show major element variations with age on Mt. ?!atib, and the apparent lack of long-term cyclicity or a trend toward divergent co n.p o s i t i o n s is ciscussed briefly on p.

2.5.H.2-47.

Unfortunately the sampling, mappin5 ano stratigraphic control was not in sufficient cetail to detect any shcrter-term cyclicity or trends, which would be of ri. ore immediate interest for the volcanic hazards question.

54

6. Listinction between probabilities of eruption from the summit anc from the east and west flanks of Mt. f.atib Ebasco distinguishes between a 1cw probability of renewed eruption from the summit and castern vents, anc increcibility of erupticn from the western flank (where no vent has formed to date).

The age data given in Table 2.5.H.2-14 of the PSAR suggests that the eastern vent has had the most recent activity; what is described as the most recent pyroclastic unit was also correlated with vent #3 However, the geochemical and tectonic r:odel assigning the eastern vent (#3) to a different and more recently active subduction zone than that of vents 1 and 2 is speculative at best and, as I have discussec above, apparently without any real basis.

In response to the IAEA anc PAEC concerns about this interpretation, Ebasco r.ocified their original distinction of eruption prc abilities (eg. PSAE pp

2. 5.!!. 2-49 to 2. 5.H. 2-SC ) to a more acceptable view that "the fact that (craters 1 and 3) are only 10 km apart (actually 4-5 km apart--ed) suggests that no conclusion can be drawn with respect to differences in their probabilities of eruption"(eg.

p.

12,

" Evidence substantiating..." and p.

59, "Volcanisa and volcanic hazards").

The likelihood of a new vent on the western flank of Mt.

Natib is probably lower than the likelihooc of renewec activity at the summit or eastern vents.

It is not, however,

" incredible", and I concur with the IAEA :nd PAEC rejecticn of that premise.

There are numercus volcance, in the world which 55

have shown activity on several flanks, and it is not uncommon for vents to te aligned alent faults passing directly through a volcano (eg. vents both north and south of Mt. Shasta)(see also Nakamura, 1c77).

If the eastern vent anc the nearby intrusive uplift are in any way fault-controllec, then activity tc the west would not be incredible.

The unusual breach in the calcera, along an E ',; lineament at the north enc of the caldera, raises the question of whether there is E-b faulting beneath (and within?) lia t ib.

Certainly there is marked E-b faulting along the transverse treaks cr segment boundaries (Fanila Eay, Iba anc San Antonio fault zones).

One might also ask about fault control of the 5?apalan slide block (scuth flank of Natib), and how the presence of a fault at the head of that landslice might bear on the discussion of possible eruptive vents.

56

E.

Volcanic Hazards in the Event cf an Erupticn of Mt. hatib Ebasco documents describe several kincs of volcanic hazarcs at the plant site.

Some hazarcs receive ccnsiderable attention, eg. ashfall, while other hazards receive virtually no mention at all (eg. " lateral blasts" or pyroclastic surEes).

All assessment of volcanic hazards has been premisec on an erupticn no closer than Natib volcanic center 2 (in the caldera).

This discussion will follow that premise even though I do not think that a western cr southern flank eruption is "increcible".

Ashfall The isopachs (lines of equal thickness) of Katnai ash, when superimposed on Mt. Natib with a prevailing wind toward the plant, show approximately 22' (6.7 m) of ash deposition at Napot Pt.

If the annual probability of a Katmai-scale eruption from Natib is 5.5 x 10 or higher then the plant should be analogy with the SSE (Safe Shutdown Earthquake) be cesigned to withstand 22' (6.7m) of ash (see Fig. 6 of "Evicence substantiating...",

Annual Probability of Ashfa11...).

Ebasco uses the data of Table 2 (same document) to estimate the points for the curve of Fig. 6, and concluces that the annual probability of a 22' ash fall at

~b

~

Napct Pt. is approximately 2 x 10 They also estimate 3x 10

/yr as the prcbability for large pyroclastic eruptions on the Eataan Peninsula, based upcn their statistical analyses in Appendix 2.5.J.

If the estimated probability of a hatmai-scale eruptien is correct to within an orcer of maEnitude, and I 57

suspect that it is, the plant should properly be cesigned for somewhere between inches cf ash anc 18 feet of ash; the exact thickness is very sensitive to the estimatec prcbability in Fig.

6.

The calcera in the summit suggests that Natib has actually had at least 1 K a tn.a i-s c a le eruption.

If so, when cid this occur, and which deposits correspond to this eruption?

It is disturbing that Etasco has not identified and carefully described such depcsits.

I wcnder if the likelibcod that o n e o f th e r.;o s t recent events on NatiL was a Katmai-scale eruption (of uncertain

~

age) would affect the estimate of 2 x 10 for the prcLability of a 22' ashfall at the site?

Certainly if that caldera-forming eruption of f.atib was less than SCO,0C0 years ago (2 x 10 ' /yr

~

=

1 per 500,000 years) it would give one pause, as well as make one wonder whether there is an age depencence on the recurrence rate of Katmai-scale eruptions?

..ould having had one Eatmai-scale eruption make Natib more or less likely (than the regional average) to have another large-scale eruption?

The ashfall from smaller, more frequent eruptions will p.obably exceed the higher annual probability values on their Fig. 6 curve, unless the deterioration with time (before the present) in the recording of eruptions has been taken into account.

Small ashfalls are of relatively little concern, though

-- they are well within the art of the engineer, anc will be eroded away within a few years of each fall.

Finally, how is the procecure of keying the cesign for ashfall to the probabilities 58

~

of the SSE, r.axin.ur proLable storr. wave anc tsunami reconciled with the F AEC requiren.ent (p. 17 cf the FAEC evaluation cf Ebasco responses to the 1(;76 IAEA Safcty Mission) that the Applicant provide " assurances that the plant woulo...

remain safely shut down uncer a 22' ashfall conoition resulting freni a I:atib eruption..."?

Is the principle of keying the acceptable level of ashfall to the SSE or other risks acceptable?

Are the probabilities of the SSE and other risks better known than the probabilities of a Katmai-scale eruption?

59

Pyrocliatic flows ano surges ("pyroclastic": hot, f r a ce.e n t al )

Williams anc Pchirney (1979, Chpt. 7) present a gooc review of pyrcelastic flows; pyroclastic surges are discuscec on p.

94-96 of the same reference, as well as in Sparks anc talker (1973).

both pyroclastic flows and surges (collectively known as nuees arcentes, glowing cloucs or glowing avalanches) are mixtures of hot volcanic gases and entrained air, ash ano larger fragments which can speec down the slopes of a volcanc.

Both are of great concern because they are relatively cctr..on occurrences on anaesitic volcanoes and because they have been known to cause total, immediate destruction at consicerable cistances frca their vents.

The types of pyrcelastic flows which are a concern at 1:atib are the Ct. Vincer.t (or Scurriere), Krakatau c r.c A s crca types.

These f1cus f e rre by collapse of lar e erupticn e c l u :..n ar.c/or by c

a teneral cvcrficwinE of fragmental rsaterial froru the crater, anc becore fccused as discrete flows which are tuch like avalanches traveling on a cushion of their cWn hot air.

These types of flows can occur with very little warning.

Merapi and Pelean type pyroclastic flows result frcm explosion and collapse at the fronts of dotes or viscous lava f1cws, anc can for the present be discounted at Natib because there is no active don.e er viscous lava ficw atcve the calcera rim.

'iere such a cote er ficw tc cevelop above the site there would be a high risk to the site, but there sculd be a period of weeks and prcbably conths of unmistakeable cevelopment tefore risks would reach a serious 6C

level.

The risks associated with the dense, lower parts of pyrcelastic flcws are hurricane force blasts of cool ano then hot Cases. ash and blocks up to several meters in ciameter (biccks of cver 10 m diameter have been reported in recent pyroclastic flows on Hibob-Hibok anc Maycn).

Destruction is virtually total, and there is commonly depcsition of several meters of coarse, unsorted debris.

Deposits up to several tens of meters thick form in narrcw canyons anc/or frorc. excepticnally voluminous flows.

The risks associateo with the upper pcrtion of pyroclastic flows are similar except that the clcuas are predominantly hot cases and sand-size or finer ash.

Ash depcsition is usually in the crder of a meter or less, anc large blocks are rare.

Even without the coarse r.;aterial, these clouds can be devastating, as in the example of St. Pierre (1902).

Virtually all animal life cies of external and internal (respiratory system) burns.

Cut of roughly 29,0C0 people in St.

Pierre, only four survived; two were on the very edge of the flow, one was a prisoner in a cungeon cell of the local jail, anc one somehow survived while all others in the same builcing were killed.

Macdonald (1972) writes that the velocity of the blast was approximately 160 km/hr, and that " masonry walls 1 m thick were knocked over and torn apart, big trees were uprcotec, 13-cm cannon were torn f r cr.: their mounts, ano a 3-ton statue was carried 4m frcm its base."

C1

A still different kina of pyroclastic flow, the fissure or "kalley of Ten Thousand Snokes"-type, is a rare event but should be r:e n t i one c in connection with the 1912 Katmai eruption.

In addition to the Katmai airfcll discussec above, a fissure openec about 9 kn. northwest cf Katmai and poured out between 7 anc 11 km,3 of extremely hot ash ar.d purice, filling the adjacent valley (20 km lone by 5 km wice) to thicknesses of over 200 m.

Magma from Katmai drained te the northwest anc mixed with magma beneath the fissure, and the rapid erupticn and emptying of the Katn:ai magma chamoer resulted in collapse of the surmit of Katmai, to form a caldera not unlike that of Latib.

Eecausc the deposits remained hot for many years, the valley was re-narcec " Valley of Ten Thousand Smokes".

This is the only such flow in historical tire, but similar and even larger such flows are cor.rnon in the recent teologic record.

Pyroclastic surges have only recently attained wide recognition anc there is still debate about specific examples anc F.echanisms.

They are closely related to and might be considerec a kind of pyroclastic flow.

Some surges are associated with large eruption clouds and may occur immediately prior to St.

Vincent-type pyroclastic flows

(" ground surge" of Sparks anc Xalker, 1973); others are asscciated with explosive interactico with groundwater in phreatic or phreatemaEmatic eruptions

(" base surge" of Foore, 1967).

Surges constitute the majority of

" lateral blasts" discussec by killiams anc FcEirney (1979).

62

Ground surces are sir.ilar to the upper parts of previously discussed pyroclastic flows, in that they consist of het gases and ash which can travel laterally at speeds of over 15C kr./hr.

Areas affected by these surges show evirence of strong hot blasts (eg. trees uproctea or sandblasted anc charrea on cne side), but relatively light ash deposition.

Base surges were first recognized as the lateral blasts in atomic bomb tests, and were first recognized in volcanic eruptiens curing the 1965 eruption of Taal Volcano.

They are similar but somewhat ecoler and r.oister blasts of volcanic gases, air ano ash.

Surges show virtually no topographic control, and as such devastate entire sices of volcanoes (eg. Lamington, 1951; Taal, 1965; Sourriere of St. Vincent, 1979).

There is a third kind of lateral blast, the sc-called " directed blast", which results frcm explosions at the base of domes or at the fronts of viscous lava flows.

These result in Pelean type pyroclastic flows and may also throw lar6e blocks in ballistic trajectories.

This third kinc of lateral blast is not known to produces surges, although it is prer,ature to say that they cannot do sc.

Of the two types of surges, only the ground surge might be expected at Natib; base surges generally require wet vent environments (eg. lakes).

Lirectec blasts require active domes or viscous lava ficws grcwing over the crater or caldera rim.

FAEC questien #5 cf 25 August 1977 inquirec about lateral blasts, and Ebasco respenced that the plant is in no danger frca.

ballistic rock missles ana that.

= protectec by topography 63

frcm " glowing avalanches".

The dense lower parts cf " glowing avalanches" do shcw topographic control, but related " lateral blasts" (ground surges), like the upper clcucs of "Elowing avalanches", show little or no topographic centrol (see discussicn below).

Contrary to the Ebasco assertion that "any (pyroclastic flow) eruption in the summit caldera or on the east side would be intercepted by the main caldera or some other intervening topographic features," it is clear from the volcanoloEical literature that none of these three types of pyroclastic flows would have difficulty clearing the 500 :: high southwest caldera wall above the "apct Pt. site.

St. Vincent-type pyroclastic flows typically descend (collapse) out of the eruption column, and one in the 1902 eruption of Soufriere, St. Vincent was observed to pass over a somrca (old crater) wall 120C' higher than the lowest notch in the crater (llay, 1959).

Miller and Smith (1977) also described pyroclastic flows with " spectacular mobility".

The Asama ana Krakatau types foam out over crater riras (cf. IAEA question about " overflowing emulsions") and/or develop from colun.n collapse.

Flows will tend to pass thrcugh low points in the caldera rim, but in no case can the caldera wall be relied upon to protect the Napot Pt. site.

Cnce pyrcelastic flows commence their cownsicpe travel, their dense, lower parts tenc to be confined to drainages but their less cense upper parts are relatively unconstrained and can pass right over canyon walls and other topographic barriers in 64

their flow path.

Eidges parallel to the ficw cffer perhaps the Lest protection, but cnce again, the upper porticns of these ficws have teen known to sweep cown ridges and canycns clike (ec.

the destructive pyroclastic flcws of the 1902 Pelee eruption which destroyed the entire city of St. Pierre).

Small, relatively unenerEetic pyroclastic flows will tend to stcp at treaks in slope but large energetic ficws will be essentially unaffected; the volcanologic literature has many reported cases of pyroclastic flows descending fror. cones roughly the same size a s t. f.atit and reaching the sea at cistances comparable to that of !.apot Pt.

The Ebasco ciscussion of topographic control emphasizes general control and ignores a substantial number of exceptions.

I find it especially retarkable that the Ebasco reports stress the tcpographic control of pyroclastic ficws by citing the exarr.ples of Pelee (1929-30) and Lamington (1951).

The former were small pyroclastic flcws; Perret's conficence in topographic control was tempered in his recommendation that a volcanological observatory to be built on a clear topographic high point (F.orne Lenarc, Pelee) shoulc be "largely subterranean, fortified with a war-fortress turret..." (Perret, 1935, p.

105).

Perret The "orne Lenard with its plough-like foot situated continues: "

in the very center of their puth, splits the nuees arcentes into flows that pursue, each, its respective ccurse down the valleys of the Riviere Elanche anc the Riviere Seche.

They rush up its steep sides, scouring with their boulders the sices cf the cliff 65

anc leaving their fragnents poised upon its flanks.

Their cicuds of ash anc freshly e:..itted cases, sweeping over the summit, hold out possibilities of pressure anc velocity measuretents, thermometric anc barometric readings, the registering ci earth tremors anc subterranean scunds, and opportunity for the collection of gases; indeed every sort of close rarage observation."

The 1902 pyroclastic flcws of Pelee showed only crude topographic control, having been pushed eastwcrc towarc St.

Pierre, over canyon walls and intervening small ridges, by the generally higher riote west of the Riviere Blanche (see p.

103, Eolt et al.,

1977).

!!aa the upper part of the 1902 Pelee nuee been topographically controlled, the St. Pierre cisaster wculc not have occurred.

A broac sector of the slopes of Pelee was devastated in that blast, again by either the upper parts of pyroclastic flows or by related pyroclastic surges.

In the 1951 Lamington eruption, cescribed so clearly by Taylor (1958), there was Eood topographic control of the pyroclastic flows (both lower and upper parts), but these were preceded by a great " ash hurricane" (probably a ground surge) which spreac 360 cegrees arcund the volcano for distances of 6 to 13 km frcm the crater.

kithin this zone there was total devastation; 2942 people were killed "in an area where the settlements were small and scatterec."

Taylor concludes that the open ncrth rim of the crater allowed the nuee to travel further in that cirection, but that "the highly chargec ' ash hurricane' avalanches...were not strictly controlled by topography..." (Taylor, 1958,

p. 96).

The 66

6 e

selecticn of Felee and Lamington as evicence for tcpographic control of pyroclastic flows is in the one sense correct, Lecause in each case some topctraphic control was otserved; the er.phasis on topographic control at Pelee anc Lamingtcn with little tention of the upper parts of the 1cC2 Pelee cloud ano no mention of the Lamington " ash hurricane" is singularly inappropriate.

Deposits from pyroclastic flows form part of hapot Pt.

Can the Ebasco gr oup demonstrate that said flows filled former topographic depressions (rather than being cepcsited on n. ore or less their present topography)?

Ebasco incicates on p.

3 of

" Geology of Unit 1 Excavation" that "These cepcsits infillec valleys eroded in ;Le lower breccia sequence" but on p 18 of the same document they say "As shown by the nearly horizontal bedcing of ash-ficw tuff in Unit 1 excavation, the flows were sufficiently mobile to spread over wide flat areas before being deposited."

I am also puzzled by the reportec absence of coarse pyrcelastic-flow deposits so common on other Philippine ancesitic volcanoes, and wonder if some of the " volcanic breccia" at the site might not represent St. Vincent-type pyroclastic flow deposits as well as laharic debris.

Deposits of St. Vincent-type pyroclastic flows are very difficult tc distinguish from mudf1cw ceposits, especially after a few heavy rains have winnowed fine ash from the ceposits.

Cther incicators of Lct pyroclastic deposits, such as bakec anc fumarclic zcnes, charrec wccc anc abuncant l'reaccrust totbs cre recllj or.1) pcsitive ir.dicatcrs, i.e.

their presence ir.cicates a hot oririn tut their absence cces 67

r.ct indicate a cocl crigin.

Cnc good t..ethco for cistinguishing between hot anc cool origir.s cf the !.apot Pt. "vcleanic treccias" would be an analysis cf the thern.oremanent r:.a c n e t i s m ( T h !-: ) of the treccia clasts (heblitt and Kellor,g, 1979).

de Coer anc Ebasco used this technique to document high emplacer.ent temperatures for the fine pyroclastic deposits cf Cara[ man anc Cabico Points (west flank of ht. !;atib) (FSAh p.

2.5.H.3-16), but I co not finc any similar analysis of the deposits of !;apot Ft. tuffs anc breccias, despite the statement on the previous page that "For the site, it is significant to know which emplacetent mechanism der.inatec, because lahars are topcFraphically controllec, while Llowing avalanches have a tendency (especially _ lose tc their crigin) to te unaffected by terrain." I would re-emphasize this latter scint, because if any of these coarse treccias (clasts typically less than C.5m but up to 3 m) are fren..yrcelastic flows, this would pose a very much c. ore difficult problem for the engineers.

It is more difficult to evaluate the probability cf a pyrcelastic flow or pyroclastic surge reaching the plant site than it is to calculate the probability of a given thickness of airfall ash accumulating at the site.

The mechanisms of pyroclastic ficws and surges are not well understood, and sc it would be premature to suppose tc quantify the probabilities of control mechanisms being in effect (ec. as in winc directicn for airfall).

The best Euices will prcbably be a.

an estimate of the total nur..ber of pyrcclastic ficws 68

reported for the Philippine-Inconesian regicn, correctea upward to compensate for the decreasing quality of the historical record in earlier tir.es and also upwarc to ccepensate for pyroclastic surges which have only beEun to be recognizec within the last 3

decades, L.

an estimate of the number of such depcsits preservec around Natib in a determined are interval, correctec upward by an averate erosion and burial factor determinec frcm the preservation and exposure of historically reported pyrcelastic flows unaer similar climatic regimes, and c.

an estimate of the frequency of any kinc of eruption en Natib.

Macdonald, in Eclt et al. (1977), states: " Avalanches of the Ct. Vincent type appear to be specifically unpredictable, and in their case reliance must be placed on predicticn of the eruption in general, plus a knowledge of the past behavior of the volcano, as a volcano that has prcduced avalanches in the past is likely to produce them aEain in the future.

Probally volcanoes that have experiencec a long pericd of rest shculd be regarded with the greatest suspicien if they shcw signs of returning activity, because the first eruption after such a period of cortcancy is commonly of great viclence, although a long rest is not a necessity for a viclent erupticn."

he have the histcrical infcrmation; r.y estir ate basec cn the er.p n.<A pericc 195C-1969 is just under ene,pyroclastic ficws per year for the Inconesia-Philippines region.

I co not know of a compilaticn of estimates for the erosion + burial factor, but the percentage of 69

s pyrcelastic ficw ceposits eroced out of or turiec in the geolctic record on cones in the wet tropics is alcest surely in excess of SC7.

If the resulting estimates were usec directly as estimates of the probatility of pyrcelastic flows affecting the plant, they would be conservative -- this is because nct all pyroclastic ficws from :iatib wculd pass to the southwest, nor would all passing to the southwest reach the Napot Pt. site.

however, in view of our very poor understanding of pyroclastic flow mechanisms, and the extreme consequences of pyroclastic flows, I think conservative estimates are in orcer.

The point of the foregoing ciscussion is tc stress that pyrcelastic flows are a.

common on andesitic volcanoes, b.

pctentially extremely destructive, anc c.

only partly controlled by tcpography.

If a pyroclastic flow or surge were to reach the Napot Pt. plant, it would a.

exert a sudden and severe mechanical anc thermal shock on the structures, ventilation systems, exposed pipelines, electrical cables, communications equipment, etc.,

b.

kill all persons outside the plant, and c.

should any part of the structure fail, kill all persons inside that part of the plant.

A plant expcsec to pyroclastic ficws or surges wculd neec to te specially cesigned to withstanc such events, and shcule also have fail-safe automatic shutdown procedures, as it is coubtful that survivors of a pyrcelastic flow or surge could continue to function rationally.

7C

Mudflows (lahars)

The ciscussion cf mucflows is acequate.

The Napet Ft. site appears to have tcpcgraphic protection frcn any likely mucflow.

Lava flows The discussion of lava flows is acequate, and lava flows per se are not a significant hazard to the plant.

Shculd there at any point develop a vis;ous lava flow on the slopes above Napot Pcint, one woulc have significant risk of Merapi-type pyroclastic flows (formed when the front of such a flow spalls off), but such a risk will not arise suddenly.

Lirtet impact of volcanic ejecta The discussion of ballistic fragments is adequate.

The abovementioned distinction between ballistic freEments and large clasts in pyrcelastic flows mi6ht be re-emphasizec here; the latter can travel in excess of 10 km at velocities of 50-15C km/hr '; r F.o r e.

The tlapot Pt. site woulc likely be prctectea by topography fret large clasts in pyroclastic flows, but might not be protected from fragments in pyroclastic surges.

Air shock waves Air shock waves are a minor hazard and are adequately ciscussed in the documents.

Volcanic earthquakes The statement that volcanic earthquakes are shallow-focus, low-magnitude quakes well belcw the SSE is correct.

In times of 71

irminent volcanic activity, however, these r.ay number in the hundreds per day.

Are there any features of the nuclear plant which would be adversely affectec by a high frequency of sr.all earthquakes?

When large shallow earthquakes are followed by volcanic eruptions, it is not clear how the earthquakes relate to the eruptions.

It is possible that the earthquakes are triggered by macma r.ovements, but it appears more likely that magma mover.ents and subsequent eruptions are triggerea by shallow-focus tectonic earthquakes.

In addition tc the examples citec in the Ebasco documents, it is possible that the large 1902 eruption of Canta

'4 aria (Cuatemala) anc the 1835 eruption cf Cosiguina (Licaragua) were triggered by large earthquakes.

Neither volcano was previcusly known to be active.

Major earthquakes following erupticns are r.o t cor. mon except where surface caldera collapse also occurs.

Crour.d tilt Ground tilt as a volcano inflates anc deflates is, as stated in the Ebasco documents, generally detectable caly by instruments.

Such uplift (or sutsicence) is most eften a slow phenomenon, but grounc tilt associated with f ault moveraents can be rapid.

The aspect of grcunc tilt whien disturbs T.e the n.ost here is the tilt which can occur in the summit regicns.

Etasco reports shallow intrusions and asscciated tilt of up to 25 deErees (PSAE 72

pp. 2.5.1-60 to 2.5.1-61) on the upper eastern flank of tiatib.

The probability cf such tilt affecting t h e f.'a p c t Pt. site is probably lcw, but I see no reascn why there cculd not be sir.ilar tilt on the upper scuthern or western flanks (see further discussion under " Landslides" below).

Volcanic gases Toxic concentration of volcanic Eases is a rare event and is restricted to tcpocraphic lows.

149 people ciec of carbon dioxide poisoning while fleeing across a small valley on Dieng Volcano (Inconesia) this past February.

Pended concentrations of volcanic cases are not a serious hazarc for the t;apot Ft. site.

Landslides and cctor collapse Large landslices on volcanoes are not necessarily associatec with eruptive activity, but they are usually related in cne way or another to their volcanic environment.

Cn ancesitic cones, large landslices are not cor. mon features, but where they do occur they can usually te related to a.

eruptive activity b.

upwarping frcm near-surface intrusions c.

hiEh seismicity in volcanic areas, of both tectonic anc volcanic origin d.

hydrothermal alteration within the volcanic pile e.

radial and concentric faulting in the volcano, and/or

f. widespread, cutwarc-dipping fine-grainea ash layers which can act as glide planes.

Cccasionally lancslices on volcanic 73

cones are related tc uncerlying karst terrain, to sector collapse (dcwncrop cf a pie-shapeo piece of the cone alcnE racial faults),

cr to calcera coll,pse.

There are historical recorcs of 2 major landslices or sector collapses on volcances in the Philippines, althcugh the sirilarity of dates suggests that these reports refer to only one event.

Jagor (1873) writes:

"I was informeo by the priests of the neighboring hamlets that Iriga Volcano, until the c o mme nc er.e n t of the 17th century, had been. completely conical anc that the lake (Euhi) die nct come into existence until half of the mountain fell in at the time of eruption.

On the 4th cay cf January, 1641, according tc witnesses, all the known vcleances in the area began to erupt at the same hcur and a leafy hill in Camarines Cur inhabited by heathens fell in,

...anc a fine lake sprang into existence... henceforth called Euhi Lake." Pagisa (1641) writes:

"...in the province cf Ilocos...five cays further inland to the east...

the third volcano erupted...the earth underwent cn January 4 (1641) a terrible and frightful quake which was preceded by a furious storm.

The earth devoured 3 mountains... The entire mass, torn from its foundaticns, flew into the air together with much water so that the gap fortued a vast lake without leavinE any trace whatsoever of either the villages or of the hiEh r.ountains which had been there before.

Ainc anc water blew up the bowels of the earth with such m.onumental fury that fragments of trees and mountains were hurlec skyward..." It is nct clear frce either acccunt whether a true 74

eruption accompanied the landslides, but the worcinE suggests that one cid.

Peports from throughout southeast Asia of louc cetonaticns on this cate may refer to this ever.t or may refer to cne or two eruptions reported in the southern Philippines -

Celebes regicn (Awu kolcano anc a possible eruptien near Jolc).

The PSAR notes a lancslide block (Mapalan block) immeciately east of the plant site.

Characteristic landslide tcpob"aphy ana up to 1-2 km of estimated lateral movement tark this as a major landslice.

It is essential to detertine the age of this lancslido.

The report suggests wave cut benches crossing the slice argins which, if correctly identified, would in. ply a relatively old age for the slice (PSAh p.

2.5.1-61).

I wculd like to see positive evidence of the wave-cut origin in the form of relict marine littcral deposits.

!<ccern uncisturbed seciments are also reported to cverlap the toe of the slice; how much time is represented in these sediments?

Cn the east sice of Mt !;atit is a "geolcgically complex" r

_ which appears on 1:250,000 topographic coverage to be ancther large landslice.

No discussion of this feature as a landslide appears in the documents I examinec.

It is discussed as the core area of intrusive upwarps (PSAR p.

2.5.1-c0 to 2.5.1-61), a possible locus of eruptions (PSAS p.

2.5.1-64), anc as a distinct morphclcgic regicn ( a r e a " C , PSAE p.

2.5.1-42 anc Fig.

2.5.1 ul).

If this is a large lancslice, was it tri,terec u,

the t

upwcrp' If sc, was the rapcian tlock sir.ilarly trictcrec?

uhat is the likelihocc that still another large lancslide coulc tove 75

down the scuthwest flank?

Since a similar, new lancslice coulu present cifficult er impossible engineering problens it is impcrtant to ur:derstand rore about the location, causes, rates of mo v er:e n t, anc timing of landslices on Mt. Natit.

Calcera collapse Caldera collapse is a relatively rare event which, on andesitic volcances,is most frequently associated with vcluainous explosive eruptions, partial cr complete emptying of a nagma chamber, and subsequent collapse of part of the volcanic edifice.

Eruption can cccur frcm summit vents or flank vents, but the region affected by collapse depends on the volume, fort anc location of the magna chamber anc the volume of magma erupted.

Most cften, collapse is restricted to sumn.it regions, even when the eruption is from a flank vent.

The fatib caldera (s) suggest that there was formerly a single or twin cone rising above the present caldera wall, which after a major explosive eruption ecllapsec to form the present caldera.

'ahen and frct which vent aid the calcera-formins eruption (s) of Natib occur?

Is there any reason to suppose that a major eruption froc Center 3 could nct crain a summit magma chcaber and result in summit collapse (as cccurrec at Eattai)?

Cuttit collapse resulting frcm a flar.k eruption coule easily trigger suttit eruptive activity.

The probability of ancther Katmai-scale eruption frcm Gatit is relatively low, but shculd such occur, the probability of 76

collapse of portions of F.t.

Natib, with consequent earthquakes anc landslices, wculd be high.

The 1968 collapse of Fernancina calcera was accompanied by 16 lccal earthquakes of - 5.C-5.4 within 1C days, plus several hundred smaller shccks (Cir. kin and floward, 1970).

Any gravitationally unstable bicek en the sices cf Natib could be loosened by such shocks.

The point to be r..ade here is that there has been at least one major caldera forminb eruption in Natib's history which, were a sir..ilar eruption to recur, would pose major hazards nct consicered in the occun.ents I have reviewed.

Cne shoulc not discuss hazarcs associatcc with Katmai-scale ashfall and neElect to mention the caldera collapse which often accompanies such eruptiens.

77

Vcleanic hazarc maps Gne wicely used approach to the pr2sentation of volcanic hazards is to incicate on maps 1.

the maximum areal extent of each type of volcanic

ceposit, 2.

the relative frequency of each type of deposit at a given point (eg. at a Eiven city), anc

3. on the more elaborate maps, a quantitative estinate of the protability of a given volcanic phencmenon affecting each spct on the map.

These maps take into account not only the recurrence frequency of each type of event at that volcano, but also such factors as prevailing wind directions and topography.

Each risk can then be portrayed as a zone on the volcano, eg. the zcne insice which there is a risk (or a relatively high risk, or a probatility greater than "X"")

that the area will be affected by mucflows, pyroclastic flows, airfall in excess of 1 C c:;., etc.

Examples of simple volcanic hazard maps may te found in Cranccll anc

ullineaux (1978) and Hyde and Crandell (1978); an example of the more quantitative hazard map may te founc in Mullineaux anc Peterson (1974).

A similor type of map is one in which all the cepcsits cf single eruptions are pcrtrayec (as in the isopach map of Katmai 1912 airfall, but with the pyroclastic ficw deposits alsc shown).

I ce not find any volcanic hazarc maps in the site safety dccuments, and it is puzzling anc disturbing that such a useful tool (and one which has been usec extensively in the 78

Philippines and Indonesia) is crittec.

I can only guess that the field work cn !!atit, was tcc cursory to prepare such maps.

79

IV.

Ponitoring and Prediction There are two kinds cf volcanic preciction -- general anc specific.

General preciction focuses on review of the historical behavicr of a volcano failing that, en the establishment anc careful interpretation of the stratigraphic record of that volcano.

From the stratigraphic record one can see the nature and approximate frequency of eruptive activity in the past, and use that as a general guide to what may come in the future.

Most of the documents relating to volcanic hazarcs at Pt.PP#1 are contributions to general prediction, but they give so few harc cetails of the eruptive history of Matit tnat they are very general contributions inaeed.

Hac the Ebasco team preparec detailed geclogic maps and used stancard tephrochrcnological and stratiEraphic techniques we would have far more informaticn on that eruptive history.

Specific prediction involves preciction of when and where a volcano will erupt, with added inferences alcut the probable nature of the specific eruption if pcssible.

Ecst specific predictions rely on geophysical anc gecchemical monitoring.

I have not seen details of the proposed monitoring system, but a few general comments might be useful.

There is general agreement that state-of-the-art s e i s tr i c, tilt anc other volcano monitors will cetect increases in activity from baseline measurements taken in quiet t irae s.

There is no general formula for deciding hcw to translate specific increases cver baseline into probabilities of actual eruption.

Each volcano appears to have SC

its own premonitory patterns, determinec in part by unique characteristics of the volcano, eg. sizes, shapes and depths of magma bocies within or beneath said cones, local and regional stress patterns, varying influence cf grcuncwater, and so on.

Harlow (1971), cited in Eceker (1973), reviewed the literature of seisnic monitorin6 on volcances and found that in 71 cases of volcanic earthquakes and eruptions, al cases showed an increase in seismic activity followed by eruptions, 27 cases showed an increase in ceismic activity without eruptions, and 3 cases showed eruptions without premonitory seistaic activity.

Studies of other pretor.itcry n.ethods show sin.ilar patterns of successful predictions, false alarms anc failures tc scund alarms (UNESCu, 1971).

Even on volcanoes with known premonitory patterns, at what level of increased activity shoulc cne shut down, and at what level night one take further safety measures (eg. evacuation of fuel)?

hhat should be the basis for setting critical levels at Natib, where no eruption has been recordec in historic tine?

How much weight should be given to each monitoring method or combination of methods?

oho shall set and enforce these critical limits?

These are crucial questions raised by the IAEA Mission.

I expect there will be considerable disagreer.;ent among volcanolocists on what are the critical levels for each form of pretconitory activity and for each phase of shutcown.

The present state-cf-the-art in volcano prediction really doesn't tell us what critical values to adopt.

Should we, in the interest of conservatism, shut down and/or evacuate at the first increase in 81

premonitory cctivity over baseline?

Or should we guess at higher critical limits, in which case we run the risk of failing to sound the alarm soon encugh?

Perhaps the discussions between scientists and government officials re: when and how to make earthquake predictions will be of sore help, as will the study of

arrick (1975) and the experience of volcanolocists whc are entrusted with the responsibility of recon.niencing evacuations around active volcanoes.

An important difference exists, obvicusly, between precautionary actions for nuclear power plants and notification of the general public.

I strongly support the IAEA ana PAEC requirement that the applicant be required to maintain a state-of-the-art volcano sv-veillance systen, staffed by properly trained anc experiencec personnel, and to present clear and satisfactory criteria fcr deciding, in the course of premonitcry activity, at precisely what point they will shut down anc at what point they will take further steps (eE. evacuation of fuel) to further safecuard against a volcanic-nuclear accident.

Those criteria shoulc be given wide hearing and the data from such monitoring shculd, I ily by the appropriate Philippine would think, also be reviewed t government agencies (COMVCL and PAEC?).

Scientists may encounter strong political or business pressures in times cf volcanic and seismic crises, anc clear criteria and shutccwn plcns thought cut in calmer tines will be neecec.

If such a surveillance system is established at Natit, it will not only be an impcrtant safety neasure for PHPP1, but it will also be one of the few if nct the E2

only volcano monitoring ststien on a volcanc which has nct erupted in recent years.

As such it can gather a Ereat ceal of heretofore unavailable baseline information about such volcanoes.

83

Concluding remarks While the l a y:n a n r..a y be reassurec to kncw that a volcano like Mt. Matib has not erupted in histcrical times (approximately 40C years), a volcanologist would not.

Given a setting in a currently active chain of subduction zcne volcanoes and preliminary indications of a lon6 history of eruptions en Mt.

Natib (with pseudo-repcse periods averaging 66,00C 3 67,000 years), and suspectinr that some of these eruptions were extremely explosive (eg. the caldera forr;.ing eruption (s)), the volcanoloCist should undertake a prcgram of careful geologic tapping and stratigraphic work to establish a proper eruptive history of Pt. !:atit.

This point is fundamental and cannot be overstressed.

The procedure of performing reconnaissance raapping and essentially random sampling provides no real framewcrk for even the most sophisticated analytical.rethcds, and no ancunt of aanipulation cf the latter results will provice the i n f o rma t i o r.

I believe most volcanologists would require bercre presuming to evaluate volcanic hazards at a nuclear plant site.

Geology is inherently an inexact science -- but we need more than a mass of a n a l y t i'. data perched on a very weak geoloEic base.

It is not at all unlike building an elaborate structure withcut a proper foundation.

I have no more confidence in the final risk analysis than I have in the founcation, and that little conficence is not augmenteo by either the conflicts with vcleanologic comron sense (eg. low estimates of eruption recurrence frequer.cy or the :r.odel of a single volcano formed above two subducticn zcnes) cr the 84

e e

apparently erroneous anc/cr self-serving sections (eg. incorrect use of statistical nocels, failure to provice C-14 dates fcr

'.a t i t, circular palecn:agnetic reasoninb, contradictcry interpretations or. the age of the " youngest" rocks at

.atib, etc.)

I have reviewed these docurrents carefully, anc I fino no reason to reject the original, comc.on sense hypothesis that Mt.

Natib is a dormant member of an active chain of volcances with a probability of serious explosive activity as hi h as that on b

its neighbor volcances, including Finatuto which is known to have erupted explosively as late as 635 years ago.

The exact prcbabilities will vary with the type cf activity in question and r.:a y be adjusted by the results of a careful stucy of the eruptive history of Natib, but I am confident that they are substantially igher than those presented by Ebasco.

Despite specific insights anc a mass of generally useful analytic cata, I cannot describe the Ebasco volcanologic work as either careful or objective.

I'ac I reviewee this work for publication in a reputable scientific jcurnal, I wculd.. ave recccrenced return to the authors for rajor revisions anc souncer, more realistic interpretations of their cata; for the purposes of a site safety evaluation, I reco xenc that Ebasco shculd be called upon to defend its conclusicns anc, if deer:ec necessary by the appropriate regulatcry bcdies, be required to improve its fundamental geclcgical infcrmaticn anc revise its interpretations as necessary.

The ultic. ate solution to questicns cf :xactly what cata is needec to evaluate volcanic risk for E5

nuclear power plants, anc what levels of r' are acceptable or within the art of the engineers, lies in ;he establishr.ent of a set of volcanologic g,uidelines ard agous tc the prescnt seist:ic r,uicelines.

E6

References Cited

Aranaki, S.,

1963, Ceology of Asama Volcar.o:

Tokyo University Fac. Sci. Jour., 2:1u, 229-443

Eolt, E.A.,

W.L. Horn, G.A. Macdonald anc R.F. Scott, 1977, Geological Hazards; 2nd edition, Springer-Verlag, hew York, 33c pp.

Scoth, E.,

1977, Papping volcanic risk, New Scientist, v.

72, pp. 743-745.

Carr, M.J.

and Stoiber, R.E.,

1977, Geologic setting of some destructive earthquakes in Central America, Bull. Geol.

Sec.

Am.,

88, pp. 151-156.

Crandell, C.R.

and Mullineaux, C.R.,

1967, Volcanic hazards at Mt. Ranier, Washington.

U.S. Geol.

Surv. Bull. 1238, 26 p.

---

, ----, 1975, Technique and rationale of volcanic-hazards appraisals in the Cascade Range, northwestern United States; Environ. Geol.,

1, 23-32.

---

, ----, 1978, Potential hazards from future eruptions of Mount St. Helens Volcanc, Washington:

U.S. Geol. Sury. Eull. 1333-C, 26 p.

Cecker, R.W.,

1973, State-of-the-art in volcano forecasting, Cull. Volcanol., 37, 372-393

Harlow, D.H.,

1971, Volcanic earthquakes; unpub.

P.A.

thesis, Cartmouth College, Hanover, N.H.

Hay, R.L.,

1959, Formation of the crystal-rich gicwing avalanche deposits of St. Vincent, B.W.I.,

Jour. Geol., 67, 540-562.

Healy, J.,

1963, The broad approach to volcanic prediction: Eull. Volcanol., 26, 141-151.

Hoblitt, R.P. and Kellogg, K.S.,

1979, Emplacement temperatures of unsorted and unstratified deposits of volcanic rock debris as cetermined by palecmagnetic techniques; Bull. Geol. Soc.

Am.,

Pt.

I, 9C, 633-642.

Hyde, J.H.

and Crandell, D.R.,

1978, Post-glacial volcanic deposits at Mount Eaker, Washington, and potential hazards from future eruptions:

U.S. Geol. Surv. Prof. Paper 1G22-C, 17 p.

87

International Association of Volcanology and Chemistry of the Earth's Interior (IAVCEI, various authors), 1951-Catalogue of the Active Vcleanoes of the norld including Sc1ratara Fields, IAVCEI, pts. I-XXII.

---

, 1973

,Cata Sheets on Post-Miocene Volcanoes of the World, IAVCEI

Jaccr, F.,

1873, Riesen in den Philippenen, Berlin,

'A i ed m a n n s c h e Euchhandlung, 361 p.

Macdonald, C.A.,

1972, Volcanoes, Englewood Cliffs, Prentice-Hall, 510 p.

Magisa, R.,

1641, Succeso raro de tres volcanes; Manila, imprenta privaca, 15pp.

Filler, T.P.

and Smith, h.L.,

1977, Spectacular n.obility of ash flows around Aniakchak and Fisher calderas, Alaska; Geology, 5, 173-176.

Miyashiro, A.,

1975, Islanc arc volcanic series: a critical review: Petrologie, 1,

177-187.

Mccre, J.G.,

1967, Base surge in recent volcanic eruptions, Eull. Volcanol., 30, 337-363 Mullineaux, C.R.

and Peterson, D. '<.., 1974, Vcleanic hazards on the island of Hawaii:

U.S.

Geol. Surv.

Cpen-file rept.74-239, 61 p.

f;akamura, K.,

Volcano-stratigraphic study of C-shima volcano, Izu. Eull. Earthquake Res. Inst.,

Tokyo Univ., 42, 649-726.

---

, 1977, Volcanoes as possitle incicators of tectonic stress orientation -- principle ar.d prcposal, J. Vole.

Geoth. Res.,

2, 1-16.

Newhall.

C.C.

and Self, S.,

in prep., Vcleanic Explosivity Index (VEI): a simple estimate of relative explosivity for historic volcanic erupticns

Nielson, C.R.

and Stoiber, R.E.,

1973, Relaticnship cf potassium content in ancesitic lavas and cepth tc the seismic zone: Jcur. Geophys. Res., 78, 6Ee7-6892.

Ferret, F.A.,

19 37, The eruptier of I.t. Pelee in 1929-1932; Carne 6ie Inst. of Wash., Pub. 5452, 126 pp.

Perrey, A.,

186C, Cocuments sur les tremblements ce terre SE

et les phenomenes volcaniques cans l'archipel des Philippines; Extrait des Memoires de l'Acacemie de Lijon (1860), 110 p.

Ranneft, T.S.M.,

1976, Understanding segmentation of islanc arcs aids exploration: Cil and Gas Jour.,

114-118.

Simkin, T.S.

and Howard, K.A.,

1970, Caldera collapse in the Galapagos Islancs, 1968; Science, 169, 429 437.

Simkin, T.S.,

L. Siebert, L. McLelland, W.C. Melson and D. Bridge, in prep., Volcanoes of the wcrld: a regional directory, gazetteer and chronology of volcanism during the last 12,000 years; Smithsonian Institution

Sparks, R.S.J.

and Walker, G.P.L.,

1973, The ground surge deposit: a thirc type of pyroclastic rock; Mature, Phys. Sci. 241, 62-64 Steen-McIntyre, V.,

1977, A Fanual for Tephrcchronology, published by the author, Idaho Sprincs, Colorado, 167 p.

Stepp, J.C.,

19'3, Analysis of completeness of the earthquake sample in the Puget Sound Area, in Harcing, S.T.

(ed.),

Contributions to Seismic Zoning, U.S.

Cept. Commerce, NCAA Tech Rept ERL 267-ESL 30, p.

16-28.

Stoiber, R.E.

and Carr, M.J.,

1973, cuaternary volcanic and tectenic segmentation of Central America; Eull.

Volcanol. 37, 304-325.

Taylor, G.A.M.,

1958, The 1951 eruption of Mount Lamingtcn, Papua: Australian Eur. Min. Res., Geol, anc Geophys.,

Bull. 38

Tazieff, H.,

1977, La Soufriere, volcanology ana forecasting; Mature, 269, 96-97.

Thorarinsson, S.,

1967, The eruption of Hekla, 1947-48, Pt. I: The eruptions of Hekla in historical times, a tephrochronological study; Visindafelag Islendinga, Reyjkavik, 170 p.

Tsuya, H.,

1955, Geological and petrological studies of Volcano Fuji. Pt. 5. Cn the 17C7 eruption of Volcano Fuji; Eull. Earthquake Res. Inst., Tokyo Univ., 33:3, 341-393 UNESCC, 1971, The surveillance and prediction of volcanic activity: Paris, Unesco, 166 p.

89

s

Walker, G.P.L.,

1974, Volcanic hazarcs and the prediction of volcanic eruptions, Geol. Soc. London Misc. Paper 3, 23-41.

Warrick, R.A.

1975, Volcanc hazard in the Unitec States:

a research assessment; Inst. Behavicral Sci.,

Univ. Colorado, 144 p.

'a ic k ma n,

F.E.,

1966, Repose period patterns of volcanoes, Arkiv fur Mineralogie och Geologie, v.4, 291-367.

Williams, H.

and McEirney, A.R.,

1979, Volcanology, San Francisco: Freerc.an, Cooper, 397 p.

Yokoyama, I.,

1956, Energetics in active volcanoes, 1st paper: Eull Earthquake Res. Inst., Tokyo Univ.,

34: 1E5 195.

---

, 1957, Energetics in active volcanoes, 2nd and 3rd papers: Bull. Earthquake Res. Inst., Tokyo Univ., 35: 75-106.

90

s t w r el Yo{cs lL Er}.%s a 1g, esa. 4e Pl..fy.; es To. b te_ j

( ajg7,)

1700 - /970 yct,

yggs AH i.

(t a 1 Lu. eF I. )

3 1, s~

r,,1, y!'

i DeSh VS1 =_Q L

h

_3_

3, 5 auces qu. eg Lao,_Q t T oo's 7

Ill I o's

, Lo'5 30's I

10 '

11 Il

.I CD I

I I

I

.60

'll I

l 70 I

. to Illl I

I l

9o I

II I

16. o o Ill I

t Io I

II LO 3o IIll I

'I o M

fil l

)

.. s'O till ll

(,o Illi 7o iffl 11l l

l l

to Illl i

l 90 lill till I

'I7 00 Il to lub II l

l 2.0 Il 50 I

llll 1

_ Illi

..,Yo Il Il I

To Ill I

i

',co Ml Il I

il 70

. Il fttklitiIlll

\\

I sll 90 I

Ittb Illl I

.I 1o l

M

[ Ifr o o ftti titk 5 lli l

l to THL IHL73,Illl Il l

I ll l

l to I

rttf.TM MM'M 7t& l I

I i

i til p

50 fttifttkM M sritL ltS ll it/

g si is il

,.. ! Yo I

IItbiN M M M.M ilff,l%!ll l

till

.iM m, CO 11 i

E MMM,M M11 I

i l

! ' (, o I

itR M M M M Mkiitt M I I

I ny hl

'7o Il fttkMIltk KM,M ll I

I til

/g 70 till ilth'IttklitLM %EEh fl I

1. 1l 1

tHI A

1. 40

.I IlI

'5 'N %M 7ts it% Mlill Il I

III m, it Mllt!

19 o o i

Iti{ rra M r>(m tsiltg E a q jrl l

)

. Io 1

ItX ltil Itn % M M M IMll!

If I

fl il M ot l

Lo l

Il tHL M mL A N N tt8.ltiLJtK nlllll y

og in,,

50 it$,Il

?HI till M N N lill lifL ll Ill

TtH, IM s.

' 10 Il llll TS M MII$ 5 M ttK II)

I llIl HQ II

' f0 :

ill i

M'.rts, M 7ttkTN M M E ill il stil M

Gds ill ill!;

11$,'Ilt R1tli% $

Iffl, Illi ng si 51 In N i l*

~

r i

Tc. 6 /(

~2 -

{

1 I

1.

Maximum column height (above crater)

Column beisbt VE1

?

l

>25 km 5

10-25 km 4

3-15 km 3

l, pgg, 1-5 km 2

0.1-1 km 1

1 (0.1 km 0

2.

Volume of tephra

}

(rarelu siven as separate estimates for different i

I twPes of n9roclastic deposits, es. airfall vs.

l ashflow deposits. a:id therefore considered here as the sum of all nuroclastic material or "tephra")

,,...% ~.

u

' w, 4W '

Volume VE1 10' 7

m 5

1 8

10 -10* m' 4

10'-108 3

I m

3 10 -10' mS 2

6 10"-10' m 3 1

4 3

<10 m

O

-e L~

3.

Qualitative description L""a:- a w; p*

vM Duncrintion VEI N

Catac19smic 4-5 Parox9smal, t,e r r i f i c,

3-5 severe, violent E:: Plosive 2-3 1

i Erfusiva, sentle 0-1 j

OP F;'T.. U l

Plinian 4-5 g

i Vulcanian 2-4, seneralls 3 Strombolian 1-2, senerallu 2

,,.a**

a Hawaiian 0-1 j.,

a l-4.

Most e:: Plosive activitu noted;

f. ' # p,,'

i Descrietiou VEI

}-

E::Plosion 1-5

' C' '

Nueo ardente 1-5 Phreatic e::Flosion 1-2, rare 19 to 4-5

• X'"}'+

• h,.t '^f,$,

Dome buildins, mudflow 0-1

>~

Lava flows 0

L..-

r,,,.

9 '-

6 p

5.

Duration of cont,inuous blast,

('

' ~

(considered on19 for a handful of e:: Plosive eruptions wh'Q.!

for which this is reported)

]

Duratico VEI

'>12 hr 4-5 6-12 hr 3-5 1-6 hr 2-4

1 hr 0-3

~.,,,.n.~..<...

.,.c.-

,.p..

d 4^s daaLil fI

'11 d

1 i.

+

} 'k..}L Mkr Skyp Ops), ' 4'e uMTf~Q G*-f EP ble*15 C[

/d[tdes'C -

b h E..!.

ht.

?$ b ff Ad19' hh*G$ *'4L l% ~lh

.h.

id I

i I

s I 1g O.

s I

i

,' f'lf 21.* 4

, I i

I

' ~.'1A s.

- 1 j e 7 -g l 4, o q' a. I i j. (' 6 ' C r O e-O '*f.. n A,. .m l o g he LO o }' 'Y/.*% O e 'l N,

- l, -

- e ~ .'j, ' = l'* ,( A ,{ l i t i i i l

• e Os a

I l-a. t o' ' " r. . aj. (8 I j I ,y .= .,.- }. 7 's. t a rc a O o l '~ l V. I I I 0 -x j. O 's e i. .- 7 4 j i i O ~ i f 4 8 ,q l 6 e. i j + i ! 4 ol i t a 4 -8 tal t t I~ 6 'O l i

• • r, D e

s .s. l l 'I + .o, + i , Oil + I, .l. I l l t 9 { j.' g! I i t l l I. .0 t h -5 I l 71 m .l p l I ,fI i l a i . l V i GONI N N EI I - ~. '.. 0 0 0 a 3 4 J J4 0 00 ao cc so so b"3 s...s ) (.~/ ( # 7 #c) IsrosOn'7) t py 6, fe,, /37o (3e A.D. ska... pu c.%

• s)

V 'T e ( v. p o., s ' (, v e z 3,,n q g,, 93 g,,,, 4 o ' S e e.c. s e vt*: L

8) ' * * " u & 3) ht.

g,n ; till Me su't 4 u s,N + - N ecro N Wj m- .., ~ -, . '. s c7 <} c ph N,y c,.ht. e,< e 4 cis 3es

Relevant experience, cont.) Aaaorted field tripa with volcanologiat 2 in New Zealand, the Philippinea, Japan, Hawaii and the western U..>. (1968-present) Preparation of the Dartmouth computer-ba ;ed catalogue of historic volcaniam; cooperative baring of information with the Smitbaonian Institution. Related publicationa:

Newball, C.G.,

1979, Temporal variaticn in the lava of Mayon Volcano, Philippinea: Jour. Volcanology and Geothermal Research, v. 6, pp. 61-83.

3. delf and C.K. Paull, 1979, Preliminary Geological and Geopbyaical atudie: in the Atitlan C Idera, Guatemala: (abatract): ab.5 tract-of the 1979 meeting of the Pacific Northwe.:t ;ection, American Geopbysical dociety, Bend, Oregon and d. delf, in prep., The volcanic explo;ivity index (VEI): A aimple estimate of explc.1ve magnitude for biatorical volcaniam, with application,

in volcanology and climatology

e s_- 'a I ~ s" 'N U n

• h

_d 4 )Y1 es =>wLT*W }y \\ \\ s ,5 U ), fa ,.. y n \\% Q . n 0 O ( M .e 8 7 1 a y i 4 !a 8d r h', O O e4 i . r g { D' lt t a b h 50 h 3 x e J i d 1 . 2 4 % +d (a t { n\\ j ~% 3g

• I i

a luL a X Mt> ms 1. =6 g p ff f T 3 -es l ~$_ t e bl i

• t e-p

'l I -a.} .1. C....,,, l a o,n, o o 4 O3O ','t s .7 j C*y p / ** ' ep - n. j 'y 1 I a G i .p,. g.. i I l l I i

m Reviewer's Cualificationa Name: Christopher G. Newball Addreas: (Office) (Home) Dept. of Earth Sciencea 70 3achem V111ege Dartmouth College We_t Lebanon, NH 03784 Hanover, NH 03755 (603)-646-2373 (603)-643-4326 Born: October 30, 1948 Married, two children Education: B.S., University of California, Davia, 1970 (Geology) M.S., University of California, Davia, 1977 (Geology) Ph.D. candidate (final year), Dartmouth College - (Geology, with specialization in volcanology) Current professional interests: Basic and applied studiea in volcanology, including documentation of large-scale, caldera-forming eruptiona, uses of petrology in predicting the nature of future eruptions, evaluating geothermal potential of volcanic areas, facilitating return of fresh volcanic deposits to cultivation. Professional memberships: Geological Society of America American Geophysical Union Sigma Xi Relevant experience: Geologic mapping and petrologic atudies of Mayon Volcano, Philippines, and reconnaissance studies of other volcanoes in southeastern Luzon, Philippineo (4 years) Study of the 1978 eruption of Mayon Volcano; additional observations of eruptions of Taal, Ngauruboe, Fuego, Pacaya and Santiaguito. Geologic mapping and petrologic studies in the San Franciaco Volcanic Field, Arizona (with the U.S. Geological Survey,1h yeara and c ntinuing) Geologic mapping and petrological studiea of the Atitlan. Caldera, Guatemala (2 years, continuing) Geopbysical atudies on Central American caldera lakea (1 year, continuing) Field trip co-leader, Central American volcanoea (3 yeara)}}