Title Computer Aided Data Analysis for Geologic Problems : DataAnalysis of Grain-size Distributions of Bottom Sediments
Author(s) Yamamoto, Kaichiro
Citation Memoirs of the Faculty of Science, Kyoto University. Series ofgeology and mineralogy (1977), 44(1): 1-65
Issue Date 1977-08-30
URL http://hdl.handle.net/2433/186619
Right
Type Departmental Bulletin Paper
Textversion publisher
Kyoto University
MEMolRs oF THE FAcuLTy oF SclENCE, KYOTO UNIvERslTy, SERIEs oF GEoL. & MINERAL., Vol. XLIV, No. 1, pp. 1-65, 1977
Computer Aided Data Analysis for Geologic Problems ' 'Data Analysis of Grain-size Distributions ofBottom Sediments
By
Kaichiro YAMAMoTo
(Received April 27, 1977)
Abstract
Data analysis of grain-size distribution of the recent sediments in Chijiwa Bay, Nagasaki Prefec-ture, was performed as a case study for showing the usefulness ofcomputer aided data analysis. Multi-
variate analyses discriminated several sediment types which are confirmed to correspond to distinctdepositional environments.
Introduuion
Multivariate analysis was recognized to be one of the most effective methods
for data analysis in order to interpret the complicated geological phenomena. Its
application, however, was not realized until the recent development of the electronic
computer, as it involves complicated and enormous computations. Since then,a number of achievements of data analyses have appeared in many fields of science.
Furthermore, the computer has promoted the development of many unique appli-cations. However, the applicational techniques of the computer analysis have not
yet been settled in geology.
During the last five years I have been developing computer programs at theData Processing Center, Kyoto University, with collaboration of many colleagues.In this paper, I attempt to historically review data analysis in geology, and describe
the outline of the computer supported system which I compiled. Finally I present
a case study which shows the usefulness of computer aided data analysis.
Acknowledgement
I wish to express thank to Prof. KAMADA of Nagasaki University for his kind
offer of the sedimentological data from Chijiwa Bay, Nagasaki Prefecture, to Dr.
Kiyoshi WADATsuMi ofOsaka City University for his financial support and encourage-
ment, and to Dr. Shinjiro MizuTANi ofNagoya University and Dr. Kunihiro IsH:zAKiofTohoku Univeristy for their suggestion especially on analytical techniques. I also
2 Kaichiro YAMAMoTothank Dr. Yasuo NoGAMi of Primate Research Institute, Kyoto University, for his
help with the manuscript.
Special thanks are due to Profi R. REyMENT of Uppsala University (Sweden),ProÅí D. F. MERRiAM of Syracuse University (U. S. A.), Dr. J. E. KLovAN of Univer-
sity of Calgary (Canada), and Dr.J. M. PARKs of Lehigh University (U. S. A.),for their most fruitful suggestion and constructive criticism.
I am also grateful to ProÅí Ichiro MiyAKE ofDoshisha University forpreparation
of the SPSS program in the Data Processing Center, Kyoto University, andintroducing me its procedures.
I wish to thank Prof. Keiji NAKAzAwA of our Department, and Prof. SetsuoOKuDA of Disaster Prevention Institute, Kyoto University for their encouragement
and fruitfu1 discussion during this work.
The work has been performed by using the facility of Data Processing Center,
Kyoto University for the computer analysis. A part ofthe programs was sponceredby the center for preparation.
Historical Review
The electronic computer, which has much higher speed and more reliableperformance than any pre-existed calculator, has made the statistical processing,which involves enormous amount of computations, easier. As the statistical proce-ssing becomes more popular by the advent of computers, the geostatistics dealing
with mainly the multivariate analysis has been rapidly developed. Although theeffectiveness of multivariate analysis was already realized in geology, the analysis
was performed by conventional hand-computation (MrLLER and KAHN, 1962) andless popular than the univariate analysis, because the number of variables and the
amount of data were limited due to a lack of high-speed computers.
Establishment of multivariate analysis for geological use has realized the classi-
fication ofsediments (IMBRiE and PuRDy, 1962) using factor analysis, and numerical
taxonomy using cluster analysis (SoKAL and SNEATH, 1963). Many results fromthese kinds of studies came out in the middle of 60's (McCAMMoN, 1966; PARKs,
1966; WEBB and BRiGGs, 1966). , Although the data analysis using computer enables the composite and ob-jective comprehension ofgeologic information, the application of multivariate analy-
sis was limited in some divisions ofgeology because of the descriptive nature of geolo-
gic data. In other words, the usage of computer and analytical methods requiredthe data of numeric and qualitative nature. Lately, however, statistical methods
dealing with non-parametric data such as original scale and nominal scale have
been invented (KRuscAL, 1964a, 1964b; SAMMoN, 1969; CHEETHAM and HAzEL,1969; HAzEL, 1970), and are applied to the geologic problems (CoLLyER and MER-
Computer Aided Data Analysis for Geologic Problems 3
RIAM, 1973). In addition to the ordinal calculation and geostatistical use, the computer, which
is combined with digital incremental plotters, developed in the late 60's, is utilized
for the spatial data handling; for example, contour mapping and triangular diagrams
(HARBAuGH and MERiAM, 1968; YAMAMoTo and NisHiwAKi, 1975a). It becamecommon to display P-diagram (RoBiNsoN, l963; NoBLE and EBERLy, 1964), T-diagram (WARNER, 1969) and triangular diagram (LuMsDEN, 1973) by computer,Lately the system simulation of sedimentary processes and tectonic movement has
been advanced by Harbaugh and Bonham-Carter (1970), Hattori and MizuTANi(1971), KoMAR (1973), DAvis and Fox (1972), and YAMAMoTo (1974a). MizuTANi (1974) attempted a histrical review of the computer applicationsto geologic problems inJapan and other countries. Many textbooks and introducto-
ry articles are: (1) HARBAuGH and MERRiAM (1968) on computer application;(2) HARBAuGH and BoNHAM-CARTER (l970) on computer simulation; (3) VisTELius(1967) on geomathematics; (4) BLAcKiTH and REyMENT (1971) on morphometrics;and (5) DAvis (1973) on statisticsin geology. From 1966 to 1970, Kansas Geologi-
cal Survey had been publishing the computer contribution series which containcomputer programs and their application examples. Many leading articles in thefield of mathematical geology appear in the Journal of the International Association
for Mathematical Geology that started in 1969, and Computer and Geosciencesbegan in 1975. According to the research survey conducted by MERRiAM (1969), 910/o of scien-
tisbs who answered the questionaire use the computer in some way, and 390/. of
them use KGS (Kansas Geological Survey) computer programs for simulation,plotting, correlation analysis, multivariate analysis, and trend surface anaiysis,
and many of them want to obtain the programs concerning the items mentionedabove. The report indicates that the computer usage for data analysis and plotting
is indespensable in the geological sciences.
Data Analysis in Geology
The problem under study is usually treated through a process shown in Fig.
L
Pr6blem Data Colleetion
{2)tESma•Data Analysis Synthestze
Fig. 1. General process ofdata analysis.
In the process, the data analysis is used to draw a critical information from the
4 Kaichiro YAMAMoTO
input data, and to evaluate it. In practice the procedure consists of selection of
proper method for analysis, calculation, and evaluation of the result. The procedure
is repeated as shown by the arrow in the text-figure, to obtain the final information
several analyses. The whole procedure is called computer aided data analysis. During the procedure, the computer should not only perform the necessaryanalyses, but also manage the data file by storing the data obtained from each analy-
sis, and by feeding them selectively back to the following analyses. The data analy-
sis requires a well designed package program which is composed of a number offunctions to the required procedures and of the function of data management.This demand has promoted the use of SPSS (the Statistical Package for the SocialSciences, NiE et al., 1975), by which the data analysis ofgeologic use can be carried
out (YAMAMoTo and N:sHrwAKi, 1975b; NisHiwAKi and YAMAMoTo, 1975). SPSS is, however, not perfectly applicable to the geological sciences, for the geo-
logic data is somewhat different from that in the social sciences, and most ofstatistical
pracedures are not designed for specifically for geological data analysis. The program
was modefied to be more applicable for the data analysis of geological use.
Data analysis using computer is divided into three categories:
(1) Recognition ofsimirality (classification)
ex, factor analysis, cluster analysis, descriminant analysis, variance analy-
sis, t-test, chi-square test, etc.
(2) Trend analysis ex. trend-surface analysis, time-trend analysis, multi-pass filtering, spectral
analysis, etc.
(3) Plotting and display ex. contour map, block-diagram, triangular diagram, histogram, scatter-
gram, pi and beta diagram, rose diagram, strike-line map, form line
contour, etc.
Computer Supported System for Data Analysis of Geologic Problems
In order data analyses, it is necessary for a package program to have both data
management functions and suMcient variety of the statistical procedures. In addi-tion, these functions must be simple. I have attempted therefore to compile pro-
grams chosen from SPSS, BMD (Biomedical computer Programs Dixon, 1973), BMDP(Biomedical Computer Programs P program, DixoN, 1975) and KGS, in order toachieve more eMcient data analysis. This assemblage of programs works as a single
system in practical use. The system is called "Computer Supported System for Data
Analysis of Geologic Problems" and will be described in the following paragraphs.
The system is designed to fu1fi1 the following requirements:
Computer Aided Data Analysis for Geologic Problems 5
SuOicient procedare function: The main requirement of data analysis involves the use
of statistical analysis. For the general statistical analysis, I have used programs
in BMD (DixoN, 1973) and SPSS (NiE, l973), which are available at the Data Pro-
cessing Center of Kyoto University. I have also converted three programs,BMDPIM, BMDP2M, and BMDP3M of BMDP (DixoN, 1975), for a FACOM230-75 computer, in order to perform cluster analysis (Yamamoto, 1974c, 1975a),
because no procedure for cluster analysis is available in BMD and SPSS. I haveadded further programs from the KGS for cluster analysis (PARKs, 1970), clusteranalysis of qualitaitive data (BoNHAM-CARTER, 1967), and calculation of n-diagram
(WARNER, 1969). In addition to the programs mentioned, I have included pro-grams designed by myself in the system.
Data mangementfunction: This facility is required to store the analyzed data in data
files, and to maintain them for retrieval and processing. The result ofone analysis
will be registered in data file for the following analysis, if necessary. This ensures
the data management, selection of proper data, and transforming them. It is in-despensable, when we deal with a large amount of data, and operate the reportedfeed-back procedures. SPSS facilitates this ability.
Pregram of easl ttsage: It is important that we can prepare, withought diMculty,
the control parameters which inform a program of the contents ofprocedures. Since
data analysis generally involves a considerable number of procedures, this is anessential factor in performing the analysis effectively. I have used the programs
of SPSS and BMDP, as they are designed to require no apecific language for thecontrol parameters, and alllow free format data input.
Interchangeabilit" of data among the Programs: When we perform a series of calcula-
tions, the output of one procedure is often input into the following analysis. This
process requires the interface between SPSS and other programs. I have employed atemporary data file to establish a network among the programs, to make them function
as a system, and I have modified programs which could only use punch cards asIIO (input/output) medium, to utilize the other modes of I/O medium (i. e., tapeor disk).
Throughout the operations, programs from SPSS, BMD, BMDP, and KGS,and the other programs comprise a system. The structure and function of the system
is described in Fig. 2.
The system is composed of two main facilities, as illustrated in Fig. 2 ; the control
of data file which consists of a SPSS system file, and the three subsystems which
perform the data analysis. The data file will be treated by the data processingsubsystem, and the result is displayed by the display subsystem. The data transfor-
6 Kaichiro YAMAMoTo
INPUT
DATATRANSFORMATION
SYSTEM
FrLEMANAGEMENT
SYSTEMDATAFILE
DATAPROCESSINGSYSTEM
DISPLAYSYSTEM OUTPUT
Fig. 2. Schematic illustration ofthe computer supported system for the data analysis of geologic problems.
mation subsystem transforms the data by interpolation, when necessary. Theflow of the data among the subsystem can be seen in Fig. 2.
Since SPSS can only accept the data in matrix form, no program that employsthe data in the other forms can be incorporated in the system. In practice, however,
this limitation poses no serious diMcuity, for the data are usually prepared in matrix
form, and most of data can be described in the form.
The contents of each subsystem are follows:
The dataLlile management subsystem: Functions of the data file management of SPSS
include storage, retrieval, selection and transformation of data (NiE, et al., 1973;
YAMAMoTo and NisHiwAKi, 1975b: NisHiwAK! and YAMAMoTo, 1975). The SPSSsystem file utilized as data file storage consists of two components; reference infor-
mation about the data matrix such as size of the matrix and labels, and raw data
in matrix form i. e. variable by sample <called a case). As the subsystem can accept
data form and provide data to the other programs, it manages the data file for the
whole system (NiE, et aL, 1975, p. 80).
The data processing subs2stem: This is a subsystem to perform mainly the statistical
procedures. It contains the subprograms of statistical procedures of SPSS (NiE,et al., 1975, p. 96), and the statistical package programs of BMD (DixoN,1973) and
BMDP (DixoN, 1975). Besides, the other programs are supplied into the subsystem;
Q-mode cluster analysis using principal component analysis (PARKs, 1970) and clus-
ter analysis of qualitative data (BoNHAM-CARTER, 1967) from KGS, and Q-modefactor analysis for large-scale data (KLovAN and IMBRiE, 1971), and non-linear map-
Computer Aided Data Analysis for Geologic Problems 7
ping algorithm (SAMMoN, l969; HowARTH, 1973). I have addedafewanalyticpro-grams designed by myself; trend-surface analysis, with polynomial and double Fou-
rier approximations (YAMAMoTo, 1973a), trend-surface analysis with moving averages
and maps of summit and river levels (YAMAMoTo and NisHiwAKi, 1975c), time trendanalysis ofstratigraphic sequences (YAbtAMoTo and NAKAGAwA, 1974).
The disPtay subslstem: The subsystem displays the data in files and output from ana-
lytic procedures, according to the plotting programs, by X-Y plotter, line printer,
typewriter, and other media. The programs in SPSS, BMD, and BMDP, for pre-paring histograms and scattergrams are included in the subsystem. I have designed
the subsystem programs for preparing contour map with an X-Y plotter (YAMAMoToand NisHiwAKi, I975a) and with a line printer (YAMAMoTo, 1976), for checkingthe input data of contour program (YAMAMoTo and NisHiwAKi, l975c), for perspec-tive plotting in three-dimensional space (YAMAMoTo, 1973a), for preparing triangular
contour diagrams, columnar sections, strike-line map, and for estimation of structure
contours from strike-dip data and mapping (YAMAMoTo and NisHiwAKi, 1975d).
Case Study
Data analysis of grain-size distribution of the recent marine sediments was per-
formed by statistical and computer techniques, to evaluate the computer supported
system in practical use. The analytical procedures and results are described and
illustrated as a flow diagram. The results are also compared with those studiedpreviously.
The case study deals with the unconsolidated bottom sediments of the Chijiwa
Bay, Nagasaki Prefecture, southwest Japan. Ali the original data are analyzed
by KAMADA (1966). To show the flow and the results of the analysis as clearly as possible, the pro-
cedures adopted in each step of the analysis are not discussed in detail. Thosewill be published separately. The sedimentary environment is not considered,for it is beyond the scope of the data analysis.
Data Analysig of Grain-size Distribution of the Recent Sediments of Chijiwa Bay
Several papers have been published on the bottom sediments of the Chijiwa(or Tachibana) Bay (see locality map, Fig. 3). KAMADA (1966) classified the bottom
sediments and estimated their sedimentary environments. INouE (1970) discussedthe relation-ship of the bottom sediment distribution to the current system in the
bay. KAMADA, et al. (1973) used the statistical parameters ofgrain-size distribution
8 Kaichiro YAMAMoTo(TRAsK, 1932), and classified the bottom sediments according to the standard of
INMAN and CHAMBERLEiN (1955) into :Type 1: composed of highly sorted fine sand; medium phi 2.0 to 3.0. .Type 2: characterized by sandy constitution; median phi O to 3, sorting coethcient
1.25 to 3.0 in general; subdivided into 2A and 2B.
Subtype 2A: composed ofcoarser grains; median phi about 2.0; skewness
more than 1.0. Subtype 2B: composed of finer grains; containing more than 50/, muddy materials; skewness less than 1.0.
Both types are gradually merge into each other near the point, where skewness is 1.0.
Type 3: muddy sediments; median phi 3.0 to 8.0; subdivided by 4Åë and 6Åë median
into 3A, 3(s-s), and 3B subtypes. It is a character of the sediments
NOMO
Ortgtn
Y
t
Fig. 3.
x
o" KrLOMETERS
NAGASAKI
MOGI
" 9seSS"W
s /> ooo ooo
oo oo o
/ o
o
CHIJIWA
'$wwv o o oo oo oo oo oooo o o,
oo 5o
.e
.
o ooijjJIIilii(IAa,
. . o oe. eoe
BAY
.:,P: '•.:.:•:
eeeeeeee eeeeee eeeeee-e ee eeee:e .
ooo
oo o-vN) Ooe
c!)
o
o
o
e
Location of sampling stations: open circles
solid circles those of No. 2001 to 2160.
eeee A e
ee eeeee e eeee . eee e e .
30
li
p
•' KYUSHU.
rNDEX MAP
90,100
represent the points of No. 1001 to 1075, and
Computer Aiclcd Datu Analysis Ibr Gcologic Problerns 9
belonging to 3A and 3(s-s) that thc subtypes are less thaJi 1.0 in skewness.
The sorting coefFicient and its deviation increase, correlating to the median
phi, but the sorting coeMcient concentrate in 3.0 to 4.0 near the boundary
between the subtype 3A and type 4.Type 4: containing more than 509/, muddy materials; meclian phi more than 8.0.Type 5: composed ofpebbly material; median phi less than O.O (more than 1.0 mm in diameter) ; generally containing abundant fragments of organic origin.
In addition to recognizing these types of sediments, KAMADA et al., (1973) con-
firmed tlie areal distribution of tl]ese types in the bay, as shown in Fig. 4.
All these data were kindly offered by ProÅí KAMADA for this casc study. Usingmultivariate analyses and several kinds of display teclmiques, data analysis of grain-
size distributions was performcd on Recent bottom sedime.nts in the ChijiwTa Bay.
BOTTOM
NiSEDiMENT
AeNAGASAKI
%'a eb"S "s"s o'S l;'• .I:L;i{.l • I--
TYPES
p" ntzan- TOISHI ENOVPAAeA .. ""lirt •1.; .e ss.. R
- ---• - ---- -1 ''lt 'li'/1."/,''1111" i
"te• 't:i .= l.i .!
.iZ../.;.i,,-t/.,'/'-;,'.:./.'i..
L=z" ,-'1
•:iz
lllllilillilllzilt,ttZ;ii:iiz.iiiiiililllliilillilliiliii/lid
------i------- -------
t;'z-tii'l'FT:'2///IE•ii.t;f./.il,i,f,r..l;/s;/r/-tf,it:igi---.-l-g-trtz.t---.l{ltiilis-t-ij-t.
"' i UKI"s'-'/•,//i;?/ii/;Ti,/i//1/li•/i/11111i•/g•,///l-,i/tiilictill•i,,l'l"illll•ll-l•t/i,,/11tNill///ll/llli/-lil/rl/L/kf,,,.,SII""
Zi,,?.;l.ti-l:,.t•i'/tiTil,;lr;t;,!i•.',i.l.k.)?2,,/S,11.,i.i./lilltt'S'Si-:...-.:.-tLi-i'ill•iii:3;'iM-
.T.-'t':/ i[1 ..."
E-.•
t1 b7.'-==• •
•11e;-L'-:.F' --lli
-----:-
zzz,iiZ;,•
7i
Z/7)Z] iil'": ii•i•'i
1llil,/iZ2/ilj•i:
4'A"li, •', ÅÄ1':I
lh• tll"ii-t-- ---------t
lii{}i{}Iil
',geg .. -.
Aa"soptMAeARA PtN"SULA
ssTA-nESH
--
.,tt•:i.Il.::=-i=/'•'/
-7777/ti
.t.:.Lr=tr......------- -----------------l- - ----t-.11tT. .-....... +eEn .......".- ------- CHSJtVA SAV ......, --.------
R : Roch gs O SKM
. tsnt.::• KAZUSA:•:l•:::I: at"
LlteL':.- :::
ICUC-ptvaTSU
Fig. 4. Areal distribution of the sedirnent types defined by Kamada et al. (l973).
Procedure
The data analysis on grain-size distribution of tlic Reccnt bottoni sediments
in the Chijiwa Bay was carried out using the computer supported system. Theflow ofwhole procedures and applied methocls are shown in Fig. 5. Main procedurcs
adopted at each step are summarised in Table 1.
10 Kaichiro YAMAMoTo
START s Data Preparation I Description of data pattern I Factor analysis to extract factors as indicators
of the pattern of grain-size
distribution curves, and to
classify sarnples t Cluster analysis for grouping lTest of significance of classification
1 Description of groups defined as result of
classification r Evaluation ofanalytic results l FINISH
Fig. 5 Flow chart ofdata analysis.
Table 1. Summarized table of procedures at each step of data analysis, Chijiwa Bay.
Step l.
*Step 2.
*
* * *Step 3.
* *Step 4.
* *Step 5.
* * *Step 6.
*
*Step 7.
*
Preparation of data, original and transformed
Data transformation
Description of data patternPreparation of scatter diagrams between each pair of statistical ahd compositional
parametersPreparation of histograms on sediment types defined by Kamada et al. (1973)
Preparation of triangular contour diagrams of sand-silt-clay system
Preparation ofcontourmaps ofwaterdepth, median phi, gravel and clay contents Factor analysis
R-mode factor analysis
Q-mode factor analysis
Cluster analysis
Preparation of dendrogram
Simplification of dendrogram
Test of significance of classification
Test for multivariate equality of means among groups
Computation of F-distances among groupsDiscriminant analysis
Description of group
BreakdownPreparation of triangle contour diagram for each group
Evaluation of analytic results
Crosstabulation between two classifications, by Kamada et al. (1973) and this
analysis
Computer Aided Data Analysis for Geologic Problems 11
Step 1. Preration of orginal and transformed data
The data to be stored in the
SPSS system file should be in"case-variable" matrix form (see
Fig. 6). A case is the basic unit
of analysis for which measure-
ments have been obtained. Inthis study, the cases correspond
to the samples collected from the
bay bottom; No. 1001 to No.1075 and No. 2001 to No. 2160(see locality map Fig. 3). Fig• 6• Data form (matrix form).
Each case is composed of values for one or more measurements that have beentaken. These measurements are termed variables, and each case within a studywill have one value for each of variables. For example, the variables correspond
to the median size, sorting coeMcient, skewness and others in this study.
To each of variables is given a name which is used as index (see, Table 2).
CASEVARIABLE(Measurement)
(Sample) 1 2 3 4 5 6 7
OOOI 1 .5 A 1 12.3 6 5 AA
OO02 2 .3 A 3 13.5 8 5 AC
OO03 1.4 B 6 23.4 3 6 AD
OO04 o .8 c 2 18.1 2 4 BA
OO05 3 .1 B 8 43.6 9 9 BD
Table 2. Processing steps and data stored into data file of Chijiwa Bay.
Step Procedure Variable Name Attribute
1 Preparation of
original dataSTANOxYDEP
Station number of sampling point
x-coordinate of station
y-coordinate of station
water depth at station
MDMMMDPHsoSK
median diameter in mm
median diameter in phi-scale
sorting coeMcient (Trask, 1932)
skewness (Trask, 1932)
GSASI
CLMUD
gravel contenbs, O/. (SHEpARD, 1954)
sand contents, O/. (SHEpARD, 1954)
silt contents, O/o (SHEpARD, 1954)
clay contents, O/, (SHEpARD, 1954)
mud contents, O/o, SI+CL
VAROO1VARO02VARO03VARO04VARO05
weight O/. ofgrains coarser than -2Åë
between -2ip and -1di
between - 1Åë and Oip
between OÅë and 1ip
between 1ip and 2di
12 Kaichiro YAMAMoTo
Table 2. Continued.
Step Procedure Variable Name Attribute
VARO06VARO07VARO08VARoo9VARO1OVARO11VARO12VARO13VARO14
between 2Åë and 3Åë
between 3Åë and 4ip
between 4ip and 5Åë
between 5ip and 6ip
between 6ip and 7ip
between 7ip and 8ip
between 8to and 9Åë
between 9Åë and 10Åë or finer than 9ip
finer than 10ip
TYPE sediment type defined by KAMADA et al. (1973)
2 Data transfor-
.matlon
VAR1O1VARl02VARl03VAR104VAR105VAR106VARI07VAR108VARI09VARIIOVAR111VAR112VARI13VAR114
phi percentile of 50/.
of 10 0/.
of 15 0/.
of 20 0/.
of250/.
of 30 0/.
ef 350/.
of tro%
of 45 0/,
of 50 e/,
of 55 0/.
of 60 0/.
of 650/,
of 700/.
3 R-mode factor
analysis
Q-mode factor
analysis
Fl
F2
F3
QFLIQFL2QFL3QFL4
first R-mode factor score
second one
third one
first Q-mode factor loading
second one
third one
fourth one
Classification GROUP code of groups
analysis
defined on the basis of factor
4 Classification by
cluster analysis
CAG3
CAG3D
code of groups defined on the basis of cluster
analysis
code of groups, more detail one of CAG3
Computer Aided Data Analysis for Geologic Problems 13
In this study, the cases are diveded into two groups termed subfiles, in which the cases
of No. 1001 to 1075 and the cases of No. 2001 to 2160 are stored separately,
The data stored into the file are composed of the following variables.
GeograPhic variables: coordinate value, depth and number of sampling points. The
coordinate value is fixed by the coordinate system (Fig. 3).
Statistical Parameters (KAMADA et al., 1973) : median size (in millimeter and phi scale),
sorting coeMcient and skewness defined by TRAsK (1932). They are not only usedas indices to show some characters of the grain-size distribution, but also in this
instance, for the sake of comparison between the previous and present methods.
ComPositional Parameters (KAMADAetal., 1973): gravel, sand, silt, clay (SHEpARD,
(1954) and mud (silt+clay) contents. They are used to reveal somecharactersof the grain composition.
Sediment tlPes (KAMADA et al., 1973): according to INMAN and CHAMBERLAiN (1955).
Frequenc] scores: percentages of every unit phi, ranging from -2ip to 9 or 10ip and
both tails, by KAMADA. The measurements are analysed in this study.
Phi Percentiles: percentiles of every 50/,, ranging from 5 to 700/.. The values are
computed from the cumulative proportion by linear interpolation. The grain-size distribution was observed on particles up to 10Åë. The percentiles higher than
700/. can not be calculated for most samples.
Step2. Descriptionofdatapattern Several tables and figures are produced to extract some characters from the data
to be analyzed. With the aid of the tables and figures produced, the analytic meth-
ods employed in the following steps are to be selected. In this case study, the follow-
ing ones are prepared (see Table 1).
Scatter diagram: a diagram of data points based on two selected variables as shown
in Fig. 7, In this study, two kinds of scatter diagrams are produced using the SPSS
statistical procedure SCATTERGRAM; one is based on each pair of median, sortingcoefficient and skewness parameters and the other on each pair of sand, silt and clay
content values.
Histogram: an illustration of the absolute or relative frequencies as shown in Fig. 8.
In this case study, three kinds of histograms on the sediment types are obtained using
14 Kaichiro YAMAMoTo
Frequency
Y
.
.
-e -- -- ' t- . . -- .- --- -- , . . .- -- . --i.
. .
x
30
20
10
Fig. 7. Scatter diagram.o
A B
Fig. 8.
CDECategory
Histogram.
F
the SPSS statistical procedure FREQUENCIES; one dealing with all the cases(samples), the second with the cases of station No. 1001 to 1075, and the third with
the cases ofstation No. 2001 to 2160. Further, the histogram on grain-size distribrtion
is produced for each case by using a specific program prepared for this study.
<1an
SfiND
ssathvaZ.x,tli/llll-.S,2•tT,•{•
"fl' a
Fig. 9.
CHIJINA SEDItrIENT COMPONENTS NHOLE ARERTriangular contour diagram, showing concentration pattern of sampleon three variables.
.polnts based
Cornputer Aided Data Analysis for Geologic Problems 15
Tn'angular contour diagram: a contour map showing concentration pattern of data
points in a three component system based on three variables, as shown in Fig. 9.
Contour maP: In this case study, four contour maps are prepared using the CMPP2program (YAMAMoTo and NisHiwAKi, 1975c) ; on the water depth of sampling points,
median Åë, gravel and mud contents.
Step3. Factoranalysis Factor analysis is characterized by distinctive data reduction capavility. Given
an matrix of correlation coeMcients for a set of variables or cases, factor analysis
techniques show whether some underlying patterns of relationships exist, so thatthe data may be reduced to a smaller set of factors or components.
If factor analysis is applied to a matrix of correlations between cases, it is called
Q-mode factor analysis, while the more common variety based on correlations amongvariables is known as R-mode factor analysis.
The factor analysis that subsumes a fairly large variety of procedures is carried
out in four stages: (1) calculation of the correlation matrix, (2) extraction of the
principal factors, (3) rotation to a terminal solution, and (4) estimation of factor
scores.
The first stage involves a calculation of appropriate measures of association for
a set of relevant variables or cases. The second stage is to compute eigenvalues and
eigenvectors of the correlation matrix. The eigenvalue indicates the degree ofrepresentation of the extracted factor to the information exhibited in the original
data. The eigenvalues are cumulated in order of factor extractions and is listedin values relative to the information content of the original data, termed cumulative
proportion of total variance. The eigenvector shows the relation of the extractedfactors to each variable or case. Appropriate factors are then selected, considering
the eigenvalue and the cumulative proportion of total variance. The eigenvectorsof the selected factors are expressed in a "variable/case-factor" matrix form, that
is called the principal factor loading matrix.
In the third stage the matrix is rotated to search for simple and interpretable
factors. Each variable or case is made exclusively related to a specific factor, during
the rotation. The relation between them is shown by a large factor loading. Eachfactor will be more easily interpreted from the variable(s) or case(s) related to it.
The cases can be ciassified into the groups related to factors.
In the fourth stage each factor attribute is numerically estimated, that is termed
factor score. The score is given to each case or variable in R•• or Q-mode factoranalysis.
In this case study, the following factor analyses are carried out (see Table 1).
16 Kaichiro YAMAMoTo
R-mode factor anal2sis on Phi Percentiles: DAvis (1970)
analysis of frequency scores. The R-mode factorchosen to extract factors, which may adequatelypatterns observed in the cumulative frequency curve
BMD08M program is used for this analysis.
performed analysis on
reflect someof grainsize
the R-mode factor phi percentiles is
characteristics of
distribution. The
(?-modefactor analL),sis onfreeuencLy scores: KLovAN
fomed a classification using the Q-mode factor
this study the same method is adopted, and the
IMBRiE, 1971) is used.
(1966) and DRApEAu (1973)analysis of frequency scores.
CABFAC program (KLovAN
per-
Inand
Factor scores, factor loadings,
analyses are stored in the data file
andto be
the results of the R-
used in the followingandstcps
Q-mode factor (see Table 2).
Step4. Clusteranalysis Cluster analysis is used to classify the cases (Q-mode) or the variables (R-mode)
into groups. Among various methods available procedure was selected in this study,
in order to classify the cases (samples).
The distances (i.e. dissimilarity) among the cases are computed first. Forexample, iftwo cases (ii and i2) have X,,f and X,,d variables (j'-- 1 .vP), respectively,
the distance between two cases is:
dilt2=[t?.1
( Xiil' - Xi2j
l)2 ]2
The distances among the cases are expressed in "case-case" matrix form namedthe distance matrix. Based on the distance matrix, the similarity among the cases
are represented as a hierarchical dendrogram shown in Fig.10. The cases corre-spond to the branches at the left ends of the tree on the dendrogram, and the branches
are united to the root.
//l
Branches- 3 Å~xl
Root
/
Fig. 10. Dendrogram ofcluster analysis. Ends of branches correspohd to the samples (cases).
Computer Aided Data Analysis for Geologic Problems 17
Cluster analysis has been successfu11y applied not only to paleontology, such
as the numerical taxonomy of species (SoKAL and SNEATH, 1963), classification of
faunal assemblages (HAyAMi and NAKANo, 1968; HAzEL, 1970; UJuE, l973), butalso to sedimentary petrology, such as the classification of limestones (PARKs, 1969).
It has, however, rarely been applied to the grain-size distributions.
In this case study, the chi-square distance based on frequency score is adopted
as a trial, and the computed results are stored into the data file to be used in the
following steps. The results of classification are stored in the data file (see Table 2).
Step5. Testofclassification The results of factor anaiysis and cluster analysis are compared with the interpre-
tation by KAMADA et al. (1973). The classified groups make their own assemblages
in multidimensional space, where each variable used for the classification defines
an axis, as shown diagramatically in Fig. 1l. To test whether the groups are clearly
separated in the space, discriminant analysis is applied to the classification.
Equality ofmeans and F-value distances are examined with discriminant analysis.
Testformultivariate equality ofmeans: to test the hypothesis that all the center of groups
are same in the multidimensional space mentioned above, using student t-test.If the hypothesis is verified, the classification is, in turn, considered to be invalid
statistically.
Calculation of F-value distances among the grouPs: to compute the distance of each pair
of assemblages in the space, and in practice, to search for the distance from the
center ofan assemblage to that of the other. The larger the distance value is, the
more suitable the classification is between the two assemblages.
Discn'minant anallsis: to re-classify on the basis of the distance from each case to
the center of each group. The distance is determined taking the group dispersion
Y Y SampleP i •- '':"' 1"'N :'::' '•:';"i•l';'ill:':GroupA /N)•':ii'i,i,1::a'sil.:l.ilil:'l
--- -- . :':':"::::' .'•si!/:•: ' GroupA " GroupB ''' GroupB •x. x Fig. 11. Groups in two dirnensional space. Fig. 12. Descrimination: sample P is classified into the group A, when the group A is larger than the B, even when the distances are same.
18 Kaichiro YAMAMoTointo account. For instance, if the distances from a sample to two groups are sarrle
in absolute value, the sample should be allied to more widely distributed group,
as shown in Fig. 12. According to this standard, each case is discriminated into
the most probable group. The fewer incorrectly discriminated cases, the moresuitable the classification is. Of course, a group from which few cases are re-clas-
sified into another is regarded to be clearly separated.
Step6. Descriptionofgroups According to the clsssification, which is obtained by the cluster analysis, some
groups are defined, and their characteristics are examined. For the purpose areproduced some tables and figures, the breakdown, and the triangle contour mapof each group.
Breakdown: to compute the means and the deviations for statistical parameters andcompositional parameters, and to test the univariate equality of means for F-values,
using SPSS statistical procedure BREAKDOWN. The larger the F-values are, theclearer the differences of the parameters among the groups.
Triangle contour diagram: to search for the concentration pattern ofcases on the sand-
silt-clay triangle diagram for each group, using the TRICON program.
Step 7. Evaluation of analytical results
To compare this classification with that done by KAMADA, et al. (1973), the
crosstabulation table is produced on both Classification A abcd A
CIassi- Bfication
BCFig. 13. Crosstabulation table:joint fre-
quency between two classifica- tions are shown in squares.
27 8 2 3
o 11 5 4
3 1 12 21
classifications by using the SPSS statistical
procedure CROSSTABS. The table indi-cates thejoint frequency of each group for
each classification, as shown in Fig. 13. A
case, which is belonging to the "A" groupby classification B and to the "b" group by
classification A, is printed in the crossing
area of "A" group row and :"b" groupcolumn.
Results of ArLalysis
Description of data pattern
Scatter diagram: In the scatter diagrams based on each pair of median (MDPH),sorting coeMcient (SO) and skewness (SK) (Fig. 14), a positive correlation between
Computer Aided Data Analysis for Geologic Problems 19
MDPH and SO is confirmed (Fig. 14a), but no definite assemblage is observed. 0n the other hand, in the scatter diagrams based on each pair of compositionsl
parameters, a negative correlation between SA and CL (Fig. 15b) and a positiveone between SI and CL (Fig. 15c) are confirmed. As shown in Figs. 15a and b,the samples contain neither silt nor clay, iftheir sand contents exceed 900/,. They
are almost constant in silt quantity, but increase exceedingly in clay, if their sand
contents are less than 300/,. The SI-CL scatter diagram indicates a tendency that
CHtJltiA, SCATTER6RAM (MOPH, SO,
F:LE CHIJI-A tCREAT:ON DATE .susFILE oNiooe oN2eooSCATTEHGRAM Of CDOvN} MDPH
1,-1 2.10
ss}
HAR,26, 19TS) C-IJS"A.BAY
MEOIAN OIAMETER tN PHI SCALE
2.T) 3.36 1,99
SEDIMtNT STZE
CACROSS)
-.62
APR,24,VATA(WITH TVPE)
SO SORT:NG s,2s s,ee
1"TS PA6E 2
COEFFtCtENT
6,Sl T.1--+"----"-----------+--"+---------+------"----------"-----+-"--------------------+--e-+----+----- :
e,2o . t l 1 s1,16 •
t : i t6,52 . : i l 1S,6S +
t l 1 1-,s- - i l : r-,oo -
l l l t'3,16 . : : : 1-2.32 +
I : t l:,4e + ; I : le.s'- - t t : 1.O,20 •
.
2-t- ---O--2 -232
- •--- -2t e 2-- --- -- . --- --- --2 ;. --
-t2 --2- ----- 22?- --
.
:
--- - "-- -- . -- 2" -- -- -- 3- 3- --- -- " t2-j-2- -2t 2 1- "----- - t-t 2.2 2 -2 22 -- •- --- -
- ?32- -":- - "- 2.
.
.
--
--
'
:--
t
.
.
-:
-- .
.
:
-- -
.
t
.
2
.
2
.
.
.'
.
i I : i ..1 l ; l . I i ; .; . : l i t + : I 1 l + i l z-l • t s l i . : t : 1 + l t ` : . I i : : +
-t----+----+----+--------------+---------+----+----+-e--+------"--------------e-+---------+----+------7.-S
e.20
7,36
6,52
s,es
-'t-
-,eo
3.i6
2.)2
1,-e
O,6-
.O,20
1,IS i,le 2.-1 3.0- 3,67 -.jo -,93 5,5b 6.i9 b.e2
CHIJItiA, SCATTEHGHAM
STATtSTtCS,.
CORRELATION CR).
STO ERR OF EST .
PLOTTEO VALVES .
Fig.
CMDPH, SO,ss)
e,7o3ss
"ssgte2)S
t---------
HSVuAHLD . tNTERCEPr (Al .
[XCLUDEO VALUES.
IS PRINTED IF A COEFFtC:tNT
(a)
O."9-Y9'
O.913T-
o
CANNOT tiE
14. Scatter diagram among statistical parameters:
(b) median against skewness (MDPH-SK) ;
APR,2-, 191S
StGNIFICANCE .
sLopE {e} .MISStNG VALVES -
COMPUTEO,
PAGE 3
o.ooooi
:.t3s2a
e
(a) median against sorting(MDPH-SO);(c) sorting against skewness (SO-SK).
20 Kaichiro YAMAMoTo
CHtJtWA, SCATTEHGRAN tHOPH, SO,
FTLE CrlJl#k (CREATION DATE .sueFILE oNiooo oN?eooSCATTEHGRAM OF (DO"N) -DPM
e,29 ",62
so}
MAR,26, i9TS) CHiJ:-A.BAV
MLDIAN D:AMETER tN Pttt SCALE
e.g6 i,30 1,b3
SEOIMtNT StZE
"CR05S) 1.97
DATAC-ITH
SK
2,jO
APrt.24, i"7S PAGE -TYPE)
SKE"NESS COEfFIC:ENT
2,6- 2,9e S.Jl-+---------+-----"--+---------+----+d------"+---------+----+----"--tt----+----+---------+-----------
s,20 - t t 1 l7.S6 . I. 1. [ 1-6,S2 . 1 1. I 1s,be . t l l -- :-4-S4 - :" ; t
l4.00 . I t t l),lb . 't t s i2,S? . I 1 t :i.-s . 1 1 l le.b" . 1 : t l.O.20 .
--2:- ;-
.
--.
-2--.t)---
. --
. 2.---- t2 2-- -2
.2
-- --.
2-
02.
.
.
---2
" --
.
u2-) ---2j-3-
2. -. 2 --
2
.
:
.
2:
.
-- - .
-
-- -
. - -2 2 -- - . ---- - : -: :-
t ---
.
--
.
.
- :: j--- - --2 -3 - -2 -2 - - -t-"- -- - - tt :2 -- : - t- - -2 t--- t-
.
-t-
-
.
. i t t `
. I l I :
+ 1 J
t
r+
tt
tI
. 1 1
1
;
+l
t
t
t
..l
i
-tI
•i
i
l
;
•+'t
!i
I
'1
t
lt
.-+-------t-t----+---------+"------t-----"-+-------+----+----+-"------+"--+----+---i+----+----+tO,l2 O-b u.19 i.tj i.-b i,so 2.i- 2.-1 2,el ),1- 3-e
e,2o
1,36
6,S2
s,6e
-,e4
-,oo
j,l6
2,32
1.-e
e.s-
.O,20
CHIJ:-A, SCATTLRGRAM CptDPH.
STATISTtCS..
COHHELATiUN CR}.
•STD ERft OF EST .
PLOTTED VKLuES .
SO, 55)
.o,sle61
2,oes"23S
t---------
Fig. J4. Continued (b)
R StsuAREv
INTLHCEPT
EXCLuDED
{A) .VALUES-
:S Plt:NTED IF A COEFFtCIENT
o,uy6-e
S.ISSII
o
CANNOT 6E
APH.24, l97S
StGNIFtCANCE
sLopL (e)
MISStNG VALUES
COMPUTED,
PAGE 5
e,oooe:
.i.?sala
o
the samples concentrate on two lines, which cross each other near a point of 350/,
silt and 150/. clay. The silt contents are mainly increasing, so far as the mud con-
tents are limited within the point. On the other hand, only the clay contents re-
markably increase, if the mud contents exceed the point. These facts suggest that
the bottom sediments are classifiable on the basis of grain-size distribution, and
safely grouped into several assemblages corresponding to sedimenary enviroments.
Histogram: In the histogram of frequency distribution of sediment types (Fig. 16),
(loniputer Aided Data Analysis for Geologic Problems 21
CHIJIWA. $CATTERGRAM CMDPH, SO,
FtLE C":JitA tCREAT:ON DATEsubFILE oNloeo oN2ooo5CATTEHGHAM OF (DO-N} 50
O,2" O,62
ss)
. MAR,2-,
50RTi-G
O;96
1"TSi CHtJIWA-eAY SEVIMLNT SIZE
COEFF:CIENT ("CR05S) 1,)O 1,6S LVI•
OATAC"ITH
s(
2,30
APH,24, 191S PAGE 6TYPE)
SkEWNES5 COEFFICIENT
2.6" 2,9e ).3i---------------------+---------+----+----------t----------"+----i----+---------t------------------+-
T,-S - l I 1 I-,6Z - t I l 16,;9 . 1 1 1 1j.S6 + I I l l.4.93 . t I I I-,oo + I 1 I ;3.6T . I t I 13.04 + t 1 1 12.4i . I 1 : :i.ls . I l i 1i.:b .
.
.
.
.
'
.
-
2
.
.
.
-- . . 2--- -- .2- -) - 2- --- . .2 -")- .-- -- 4 2 2 -- - 4
.
.
.
. --.
.
2-
-- 2 . '2
-- 2 --
.
.
.
-
-
.
.
-----2- -- -
:
-- 2-- S --- -2- . .
::
2:': -; 2:
a2-2:.
-
.
.
-
--
-- --2
-- ---j 2-2--- -j2 .2S2- -"3-2- - .--22
.
`
..
.
.
. t I l 1 . t t t :
. I t l l
. E `
tt
+l
i
`
:.ril
l
.Il
t
I
.I
1
Il
+-t
I
z
1
.I
r
sl
.-+"-----------+----+-"-+-----------"----------""------------tt-------------"------+----------t
2,Si 3.1- }..s
T.4S
6,e2
6,i"
5,S6
-,90
-,so
3,sT
3,e-
2,41
1.IS
i,IS
O.12 O,-6 O,T" 1.1) lt-e i,ee 2,14 2.-T
CHIJIkeA. SCATTERGRAH (HDPH, sO. s))
5TATi5TICS..
COaRELArtON (H). .O.j12o) STD EHR OF E5T. 1.2b"e PLOTTED VALUES . 235
---------- l$
Fig. 14. Continued (c)
H5-uArtLD . O,1)S"2iNTLHCEPT (A) - J.76-i)[XCLUDEO VALUES- O
P"INTED :F A COEFFtCIENT CANNOT SE
-P".24. 19T)
SIGNIFKANCE SLVPE Cb) NTSS:NG VALUES
COMPUTED.
.
PAGL 7
o,eeoo:
.i,29UIT
e
Fig. 16c indicates more variety ofsediment types than Fig. 16b does. The collecting
area ofsamples No. 1001 to No. 1075 seems to be different somewhat in environment
from that of the samples NTo. 2001 to No. 2160.
The figures on grain-size distribution (Fig. I7) illustrate that the samples may
be divided into a few groups on the basis of the curve patterns. Then, the curve
patterns are compared with each other by using the multivariate analysis, in order
22 Kaichiro YAMAMoTo
SCATTEHSRAM OF cHtJIh'A CSA,Sl,CV APR.IS, 19T4 PAGE 2FILE CH:eOIVi tCRtATION bATE s MA".Z2, ier4)sueFILt oNlooo oraoeoSCATTERGR-M OF COOIN} SA SANO R-TIO tACHOSS) $i SILT R-TIO 3.)" iO.62 1'i,To 24,Te ]1,-6 3e.94 46.u? So.10 60,le 6T.?e -------------t------+---------+--t------------------"+--"-----------------------t-+----+-"-------- ieo,oe .g I l . 90.2S .2 . E I + t'"' 1 ; t t.U2 t [ t I'...- I : t 1 2.. 1 1 1 1. ....t 1 : l, j. I t r :- --2 ..-1 t 1 TO.e4. - ----1 t + :.-..-.-...--..-.".---.9.gg..-9.7.-...h"....-..-..---.-.-.."...-.."""".--."--."-".""1 1 - ---- 1-- t I -2-.- t sl,:2i 'i/-i :-::- i si,., 1/ 'r..: ;: 22. I
: l-1 l t"a .. ; t I -- I-- -"6e! l ----- i- I l " l.• 1----.-----.-----.----.-----.- - -.-.".-.-.-.-.---9.--.-.--V-.-...--..--"-".".--".--."."":
I : . 1 t 1 t . s l : ' . I : t . ' l't 1 . t l l :
-- r : : : .
ioo,oe
90,lt
eo,s-
le,s4
61,12
si,4o
-:.6s
r1;.96 ;
l?2`?4 i
1 i12.S2 i
l 12,eo !
Fig. 15.
1
l
1
t
l
lT
fi1
1
l1
t
!
r
.
.
- ---.
.
. ..
4-.
..
::
:
:
It
t't
l
l
l.tt -'1
t-1.r- --r --: -- 2.3T---- --
.
.
.
.
.
.". I---t
--- - : -
.
.2
.
.
t---------------+----+---------+t-------+-------+----+---------+-"-+----+----+---------+"---------e.o 1.os 14,lb 2i,24 2e.Ot )b"v -2.-S -9.S6 S6,S- el,T2 70,eO
(a)Scatter diagram among compositional parameters: (a) sand vs silt (SA-SI);vs clay(SA-CL) ; (c) silt vs clay (SI-CL).
ls.ge
22,2-
i2,S2
2,tO
(b) sand
to classify the
size.
samples objectively according to the frequency distribution of g rain-
Triangte contour diagram: In the sand-silt-clay contour diagram (Fig. 18), some
concentrated area are confirmed. Fig. 18b is somewhat different in pattern from
Fig. 18c. Namely, the both figures show a common concentrated center, but theother center different; one is situated on the side of less clay content, while the other
is on the side of less sand content. It suggests that the collecting area of the samples
No. 1001 to No. 1075 is not the same in environment as that of the samples No.2001 to No. 2160.
Contour maP: The bathmetric map constructed by computer contouring (Fig. 19)
Clomputer Aided Data Analysis for Geologic Problems
ScATTERGRAM oF cH:JrvS (S-,sl,cL) ApRnS, igT" P-GE 4FiLE CHIoolvi ccREATtoN' DATE = -k.22, lgTg)sueFILE oNloeo oN2oeoSCATTEftGRAM oF (eowN) sA SN,O aATtO CACrt055} CL CLAv RATIO 3.9e il.9S lg,9T 2T,96 3S.9b 4j-94 51,9J S9,92 6r,e: IS.OO -+---+--"-".-.----"---..--"--+----"-+----.----.---+--"i----+.---"-"+---------+-------"--
23
10",OO• .9 r) :4 Il- l- eO,!S .2 1 : I : e",s6 . s l- 1 I Te,e- + t r l el,12 + r : l I s;,4e + ; ; 1 I 41,6S •. r 1 I li.96 . I r 1 r 22,24 . I 1 i I 12,}2 . 1 l 2,SO i
Fig.
shows a ri
and ato No.
from theof the
Themudin
to that
certaln
--•--22-- ----- - -- -- --- - -2- -2- -- - u-- ---
I---.".""-""."-."-9- . -- -• -- - . -
t+----+----+----+----+-------!
I
lt
I
t
lI
l!
lI
I
l;
l
r
I
l
:
t
l!
1I
1I
t1
it
:
l
li
.I1
l1
+l
II
r
.1
l
s
l
•+
l
:.-."....-v99g..-..99.....-..-......h-...".--...-....-."....."-.-.-.-.-"."-..-.".".".-.-.-.I -- ---2 l : 1 ------ ] i l -- ---- l t l }- -f --- 1 l r --2 -- t 1 -: 2- -2 .. I I : --l ; - -,1 1 : ..---1 r : .. .- t I : -2 1• I . ---l I T .-"9.-".."......t
.
2
.-I-
i
1-1-i'-i'1
-:vI.I- ---1
:-I--I-i
l -e1.t-+----+--
----------------------"t-------------------------: 1: I+ I: 1: I1 Ir :1 it 1l 2.-- J !-t--- -! 1 - -2i- -2- --1 l --- -2 --a 21 --+-------+---------------------+--t------+-t---------
ioo,oo
ge,2e
eo,s6
10,e-
61,i2
Sl' ,"O
4i'6e
31,9S
22,24
:2,S2
1,eo +. e.u 7,gg ;).98 2J.91 31.9b j9.9S 47.9- S5,9S 61,g2 li.9! 79,90
15. Continued (b)
dge structure in the central portion of the bay, surrounded a flat bottom,
submarine canyons also along the Nomo Peninsula. The samples No. 1001 1075 are collected from the submarine canyons, and those No. 2001 to 2160
central portion of the bay (see Fig. 3). The geomorphologic defferences bottom for both sampling areas should have influence on the characteristics
of sediments.
contour maps on the distribution of median phi scale, gravel contents and
contents (Figs. 20 to 22) show that the mud contents (Fig. 22) are rather high
the nearshore area, and indicate that an spatial distribution pattern is correlatable
of median phi (Fig. 20). The distribution of gravel contents are pathy in limited areas along the seashore line and the Nomo Peninsula (Fig. 21).
24 Kaichiro 'I'Apt{AMoTo
SCATTERGRAM QF CHrJ:WA CSA,Sl,CL)
FILE Crileelvi cCREATION DATE tsu"FtLE eNlovo ot,2oooSCATTERGRAM OF {:OWN) S[
S.9" 11.9e.++I
ItI
+rt
l!
+I
t
T
:
.I
+ -+- -+-
lApt,2i, 19T4)
S:LT RAT;O
19,97-+
21.g6 3S.9S+ +- -+- : 1 ! t s I t '-- i l ; It- T-- l- - -: - L- -l - I. o
(ACijOSS) CL
.3.9" )1.gS-+----+--d---- +----+i-
APR,iS,
CLAY
S9,-+----+- -
lgT4
RATIO
92 eT,el
PAGE'
IS.90
e
Tu,eo
6j,T2
Sb.6g
49.SS
l1
1
.r
1r
1
.1
::
1
+I
r
I
I
+1!
I
1
.tl
!t
.'l
1
r
+9
o
.
.
-
.
.. .e9
-
•-
.
-
.---. 9
.
-
-- .j
--t 2t
9-.-
2- - -2- - .
2--- ----s9
ts
II
1
I
tl
st1
1
1'
II
,
+- -------+----+----+----+- + I r s I . r ; 1 l + I : : i + I
42,4S
3b.40
2S.12
2- --
. t---.
--99
.
--
-..t-
9- .---
.---
-- 2t-. .
2
.--
.9
.
"
t -"--
.
--2
-- . .
"9 9
t I- I -1 2- 2I .1 "r- :- i t l I
-I I :
.-
.-
.
.. 1
't2t l : 1 s I
I1
t
1 : I I
-d--"--
-- -----
"----------i--------t•:
t
.t1::
.t::
r
+:r
21,?4
14,16
1.ee
e.o
2
. . . . -
-p2- )--2 2- .
".
.2
.
.
.
.
t2
--. .
.
.
-- -+----+-- -
i
s
i
i
i
1
!
:
1
:1
r
I
!
1
I
:
- - +"--
i
t
11
I
I
!
i
ll
1
l!
!
1
l
1
-+--------+--"+i --+--"+- ----
-----------l
-+-
-1 +
srtt
.1tt:
.::
l t .
Fig.
+--+--+ + .O 7,99 1).9S15. Continued (c)
2j,91 3i,96 )9.VS -T,9- S),9S 6), g2 71,ei-------+ ,79,90
TO,eO
eS,T2•
S6,S-
4e,S6
-1,4S
ss,4e
2S,S2
2:,24
i-'i6
r,es
o,o
Computer Aided Data Analysis for Geologlc Problems 25,
CHtJtWA TYPE HISTOGRAM
FtLE CH:JtWA CCHEArlON OATEsueF"t e",:eoo eNaaoeVARIASLE TYPE SEDtMENT CObE t 2
tA
2A
?e
3
SA
IB
.
s
VALtO MISSING
Fig. 16.
. SEP,IS, 1914) CHIJtu.8AY SEDIMENT
TYPE
--------e------- { ;S) s,- PcTtl:
-- C 1) O,4 PCT,ttl
---------.- ( ta) 4,) pcr111
-----------t----t------t--------tt-t------------tI1l
-------------------t---------------"----- c 4e)
5tZE oATA'CwtTH
c -e} ?o.4
tT,e pcr1t
l
----------------------e---------------"-------------------t1r
--------"-------------"t---------------------------------- (
PCT
( ss)
Sl)
SEP,26,
rYPE)
a4,T
24,;
PFT
PCT
19T- PAGE 3
111
--"-- c1t'
--- ctr
1.".,."oFREeVENCV
oesERvATIoNsOe5ERVArlONS
Frequency(b) samples
-} i,T PCT
2) O,9 PCT
.1..".""L".".";"""...1,"."."""..",.1",.."..1"..""10 20 3" 40 SO 60 70
.n).o (a) • of the sediment types defined by Kamada et al. (1973): No. 1oo1 to 1075; (c) samples No. 2001 to 2160.
.1so
".-""u"",,.,s go leo
'(a) whole samples;
26 Kaichiro YAMAMoTo
CH:JIWA TYPE HISTOGRAM
FiLE CHtJIWA CCREAT;ON VATESUBFLLE ONZOOO
VARtAeLE TYPE SEDIMENT CODE l 1 ...{ 1) 1,3 1 T 1 2A --------- t 4) 1 : :
2B
. SEP,iS,
TYPE
PCT
),3 PCT
197-) CHIJIbuA.SAY
---------------------------"--------- (t1
l
-----------------------------t--t tt
t
t
;6)
SEDtMENT
le) 2",o
21,3 PCT
PcT
S:ZE OATACwlTH
SEP,26,
TyPE)
Z9T- PAGE e
1
IA
VAL:D M:SS:NG
Fig. 16.
----------------------t---------t---------------'------------------------- : : s""."":""," Db FREauENcY
oesERvAnoNs .OBSERVATtONS .
Continued (b)
--I---------l---t----tl-t--t-t--:--t------ti-e-
io ts 2o 2s 3eIS
o
c
-----I-----t--it )s
36)
--i40
4e
-t
,
,
o
.
PCT
--:---4S
"""T so
CHIJIWA TYPt H;STOGRAM
FILE CH:JtWA (CHEATiON DATEsuBF:LL oN2oeo
VARIABLE TYPE SED:MENT COOE 1 1 --------------- ( r s l IA .-{ 1) O,6 : : : 2A -------{ b} r : l
?B
. sEP,18,
TyPE
PCT
1"
3,7 PCT
19T"
S.1 PCT
------------------------------- c:
1:
--------t------"M---"-- c 24)::
:
----------------------- c 22):1
:
C•.IJIwA.BAY
30)
x),o
iS,1
PCT
SEDIyENT
PCT
SIZE DATAC.:T-
sEp,26,
TyPE)
1914 PAGE s
3
SA i3,7 PCT
]S -----------------------------------------------------t---- ( 57) IS-6 pcT r t i " -.---C 4) 2,s pcT : 1 : S ...C 2) ;,2 PCT : : T",",",I"","":,""""i.."."":,. O lu iO SV 40 FREOUENCY
VAL:D OeSERVAT:ON5 - 160 MZSSING OSSERVATIONS . OFig. 16. Continued (c)
. ,,.1,,".",.1.-."".1"""".T","H"1"".""t so 60 70 eO 9e :oo
Computer Aid ed Data Analysis for Geolog ic Problems
Åë
o
5
10
o
5
to
o
5
ro
o
5
le
o
5
10
o
5
:o
o
5
ro
o
s
le
o
5
10
o
5
to
IWI
teei
10/1 1011
/e/t lett
ln17 lelt
leo]
1
10/]
leo- TOP-
/OTI
--/et-
/pst
.L.-:oll
10]-
Ie4t
1011
tees
lee-
lels
L---•----.
tei-
iots
:e/-
ts/ots
/o]s leis
xTols
teTl letltoet
tNt \OTt
se]1
lel- IOIt
loe- 101- lpt- 10Å})
:edi
--le-l
lete
tett
le-l
lnst
LOI)
-
-
les-
Tess
-L..-Tt.-
1-li l-tt 1-1- l"-
tes-
-10Sl
10Se
101t
-
10 se m le co m
1{" 1-i-
Fig .17. Frequenc y distribution
to co se to sm to s s" w (a)
of grainsize, samples from the bottom of the
eo og
Chijiwa Bay.
27
28 Kaichiro YAMAMoTo
Åë
o
5
10
o
5
10
o
5
le
o
5
10
o
5
10
o•
5 1 1
le
o
5
10
o
5
10
o
5
le
o
5
10
)nil
/e-1
TPSI
t.
teTl
tn)r
tee;
bnll
tOCI
lo"
loes
lnel
-Tf
leT-
10Ts
;oed
lell
.-- -
/ce-
i{[IIII
li)J],..,
lei-
"
"
IOS-
Fig.
lele
:O so 5017. Continued
lvt
-aot]
10tS
-ny100S
lu/-
'iI
ELL-.
tOII
tnlt
telt
Ntexl
,-10tl
'
il lo]t
lr1 te"!
L...-
?Ol-
kzT-
tets 1
h-"e)s
teol lo/t
tooe lnle
lnot t-lt
t-le
IOtP
tot-
ttl-
-
•N
-
tet"
'
t-l- tt"
le so sc
toli
"blt
t-lt
1-"
10 se co lo eo so
t"-
10 sc se k
Computer Aidec l Data Analy sis for Geolog ic Problems
Åë
o
s
10
o
5
10
o
5
10
o
5
10
o
5le
e
5
10
o
5
10
o
5
le
o
s
:o
o
5
le
Fig
te-1
to-t
10Sb Te-1
test nil
le-t
roTL
lo-4
<S-.lieTi
Te"
10S]
-+IOS)
LlfL•"]
tett
IDri1
le"
lntl
JO"
'
loss
lo--
lets
?o--
)e-?
10-S
t"4-
nrm-
.
-- ---
!i
loss
F
L.-.. 'e•:.
-
eot-
tnfs
T .ttt.tTt."
losr
t--o
les-
.-
;
L
'fl
tnTt
toel
-
10tl
1011
let4
-T
)"TS
10-T
?e--
4
loi-
10
. 17.
se se leContinued
lelt
Ies"
--•-.-3e se
t-To
le se
zelT
tp7i
tt/
toti
loel
ie-s
to-s
totl
.
tt-t
nl-
ttle
se
lb"-
l-t-
IQ se SO to so
tbt-
tet-
tL"
se 10 SO se k
29
30 Kaichiro YAMAMoTo
Åë
o
5
le
o
s
10
o
5
10
o
5
le
o
5
10
o
5
io
o
5
IO
o
5
10
o
5
10
e
5
le
Nel tlTl illT
t:ql vn t/n
tlol
1:O-
tlO-
111] iell
,11b-
1114
t:et
ITbsLT. ..+L.-L.
t::-
1/IS
veT <]
ve- 111-
tl--
E.-rr---
jL lbll
..T fiL-
t:11 ?1-1 tl-1
tllt tl-t lbS?
:llt
tile
r/1]
tu-
rp-1 MSI
17ii
10
Fig. 17.
111t
ES:L=-- v-s
11?-
lltt
vÅ}e
se se le se seContinnued
le eo
11--
?bSS
;ll- 11-s tLSf
I
ll]l
iblt
' , , , t lsl
? / -- 1 b st
1/1- tli-
Il-e
se
tlS-
10 !O se :O 30
Il--
ti--
se le ao se .
Computer Aid ed Data Analysis for Geologic Problems 31
SAND
C)•/kZ&-$sw
1
?
'
[1'x<
c2
1
7
1
SILT CLRY
Fig.
CHIJINR SEDIMENT COMPeNENTS inHOLE AREg
(a)
18. Triangular contour diagrams, showing concentration pattern of samples diagram of sandsilt-clay three components system. (a) all sampies; (b) 1001 to 1075; (c) samples No. 2001 to 2160.
on triangular samples No.
32 Kaichiro YAMAMoTo
SAND
<atÅíxl$lisx
-y fZziLltsZf7/zk21?t
!s //
/
"
SILTCHIJINAONIOOO AREA
SEDIMENTCOMPONENTSCLA\
Fig. 18. Continued (b)
Computer Aided Data Analysis for Geologic Problems 33
SA Ni D
N
Å~
XXkx
'NS,g.
va
ISiuilSl,,,,lliZZ/,,,,t,•t".' }'gg'N {"s}slil; %Z N,
•/ts
Lwwwa't//,i'i/;,x
/
x
Nss
//
'
r'
T
f
/y
,.t
fr'
fr
y
/, i 11 /i"" i' rf X
l<
,1 .4`xk/init,7'/
'<' 2>
(Å~
ssll "vll,lli"l"
xÅ~
SILT
x
CHIJINAON2000 PREA
SE [] IMENT coMpertiENTsCLAY
Fig. 18. Continued (c)
su KaichirD YAMAMoTo
taI `r
NOM8
NAGgS"Kl
"o r' T
Fl
AB"
c--y.it
ICTSH!rk: 'j
t. N>.F.-
"Ar )
4t.
4V
PENIrlSUTAM[5H!
'
js.
J--•ao. $Sisi
.4S-
. r' 42•
,)
CH!
b,5.N.
cO)FJ'
ot: t
JIN"
)
,J-
5.dn.
lo•
gEil7
,s•'s,
t)o
CH
id
IJI NA
<ill.
<
)t-
BRY
rt-ius
KVCHttCT5U
Fig. 19. Contour map of water depth (DEP) in meter.
Computer Aided Data Analysis for Geologic Problems 35
o s"p
NOMO
/
NRGRSRKI
MO I
PENINSUTRpt sHZI
TOISHf
ulRBR d..
Åëb5.
&a-if
1•
8. s?"•
by
f3
CHIJIWA
I'
BAQ•
Y
tr
EH
.
1,
6
KPZUS
KUCH
Jrwp
NCTSU
Fig. 20. Distribution of median ip (MDPH).
ss Kaichiro YAMAMoTO'
e-w
NeMo
.
NRGRSRK!
MO
PENINSTR" SHI •
--.lli;/-
5•
I
PBR
of•
TessHIgiii},i
CH
'
CHIJINR
(il}>
,,iiei;-
ng •
i?''
'"v.
BAY
KRZUS
Kucn
JIHR
NOTSU
Fig. 21. Contour map of gravel contents (G) in weight percent.
Ciotnputer Aided Data Analysis for Geologic Problems 37
1"V rt#
CH JrNA
V NPGRspKI
! MONeMO PENINSU TRpt SHI/
.Z'X.
"o
I
n
gBg
TOISHI
o
`ep
o
0
20
o
.
CHIJINA BAY .
KAZUS
KUCH
.
NOT•SU
Fig. 22. Contour map ofmud contents (MUD) in weight percent.
Factor analysis
R-mode factor anal)sis: Three factors are extracted, judging from the eigenvalue
and cumulative proportion of total variance. To obtain the final factor loadingmatrix, three principal components are rotated (Table 3). The values are diagram-matically shown in Fig. 23. As the result suggests, the factors seem to have the fol-
lowing attributes on the cumulative curve pattern of grain-size distribution.
Fi...Size factor. Position of curve, for the factor Ioadings are positive and almost
same.F2,..Shape factor. Range ofthe curve.
F3,..Shape factor. Distortion of the curve. The negative score means that thecurve is distorted on the larger side of the proportion (O/,). The positive means the
contrary case, though the distortion will be smaller than that of the negative.
38 Kaichiro YAMAMoTo
The correlation of factor scores to each statistical parameters (MDPH, SOand SK) are shown in Fig. 24a-c. Fig. 24a indicates strongly high correlationbetween Fi and MDPH, but Figs. 24b and 24c do not illustrate any one. Factor scores are plotted to investigate the relations among newly extractedfactors, Fi, F2, and F3, and among the samples based on the factor scores (Fig. 25a-
c). On the basis of the scattered diagram of Fi and F2 factor scores, the samplesare tentatively classified into four groups named A, B, C, and D, which seem to take
their positions parallel with each other on the diagram. The cumulative curves of
several representative samples of the four groups are shown in Fig. 26a, and their
positions on the scattered diagram of Fi and F2 in Fig. 26b.
The cumulative curves are reconstructed using the model obtained as the result
of the factor analysis. Although those of the samples which show unimodality offrequencies of grain-size distribution are well reconstructed, those of showing multi-
modality are not well reconstructed.
e-mode factor anatmsis: The computed eigenvalues and cumulative proportion oftotal variance are shown in Table 4a. Four factors are extracted, and they explain
94.70/. of total variance. The eigenvectors (the principal factor loading matrix
Table 3. Results of R-mode factor analysis using correlation matrix.
Variable F,Initial
F2 Fa Quartimax rotated
Ft F2 FsVARIOI (di 5)
VARI02 (ip10)
VARI03 (Åë15)
VARI04 (Åë20)
VARI05 (e25)VARI06 (Åë30)
VARI07 (ip35)
VARI08 (Åë40)
VARI09 (Åë45)
VARI10 (e50)VARI11 (ip55)
VARI12 (Åë60)
VARI13 (Åë65)
VARI 14 (e70)
O.892
O.950
O.974
O.987
O.993
O.995
O.996
O.994
O.992
O.988
O.983
O.978
O.970
O.962
O.409 O.313
O.219
O.146 O.087
O.028-O.023
-O.os8-O.107
-O.141
-O.173
-O.202
-O.219-O.226
O.104 O.059
-O.OOI
-o.oro-O.067
-O.086
-O.087
-e.o74
-O.022
-O.022 e.oll
O.055
O.103
O.119
O.885
O.945
O.971
O.985
O.991
O.995
O.996
O.995
O.993
O.990
O.986
O.980
O.973
O.964
-O.435-O.332-O.229-O.149-O.086-O.025 O.026 O.067 O.101 O.128 O.153 O.172 O.179 O.183
O.020-O.O03
-o.o"-O.067-O.082-O.088-O.080-O.057-O.029 O.oo9 O.048 O.096 O,I46 O,164
EIG*CUM•PCT**
13.32
95.2
O.55
99.1
O.07
99.6
(l3.32
95.1
O.53
98.9
O.09 )
99.6
***
Eigenvalue. Sum of squares offactor loadings in the parentheses
Cumulative proportion of total variance (O/,)
Computer Aided Data Analysis for Geologic Problems 39
1.0
o.
o
Fl-t
: : : :i::::"'----- ':":'::i:': ;•::-:•::••• :- :-t::s.----lt•:,f,:.:. :.:.:,::d--. .i:I.:.:i: :t---:-:--:-;:-i-i-- -:t:-:J:-
::----:::::J:.:'
:•::;• :•;•tt----
::':i:'::I':-t-t
::;:J-t::::i:::.
o.o
-O.5
F2
:-, :.;:.:;)
i.'•IJ:;t:i -;.;.::.: ::;----.
;r•:il:.1:
lt-L::::
F3l.}•,::/ ::;• 1•
:.t::.:
i:::::':"tt t t+ tt
i.::'U
O.1
o.o
-O.1
:"D.2ts
>
.9gs6 NeE 'et-
gsFig. 23.
u) oeomomo Ln omo"N pa th mev ut ut eo "`gsvngvnsg" fi amc u} eN co aOrN th eeoeooeor"-r-e. e e e- r-. e- e- t- t- r- r e-ct ae cx a ee ac et ac et tr at ntq(<<q(((q rt qq)>>>>)>)))>> Final factor loadings of R-mode factor analysis.
of these factors) are rotated by the varimax method. Then, the samples are plotted
in a coordinate, where each pair of vectors define X- and Y-axes (Fig. 27). Fig.
27 indicates that most samples are concentrated near the factor axes.
KLovAN (1966) employed the Q-mode factor analysis for grain-size frequencydistribution data and classified recent sediments into four groups on the basis of
grain-size characters. According to his procedure most of the samples are wellclassified into any of four groups, named 1, 2, 3, and 4, each of which is related to
one of four factors. There are, however, still remaining several samples which have
not any significant relation to each factor.
4Q Kaichiro YsxMAMoTo . By factor scores of Q-rnode analysis, variable(s) is/are selected, which is/are
closely related to each group, as shown in Table 4b. The group 1 is especially re-
lated to the 7th variable (frequency of3 to 4ip), the group 2 the 13th' variable (fre-
quency of 9Åë +), the group 3 the 4th variable (frequency of O to 1ip), and thegroup4 the 6th variable (frequency of 2 to 3Åë). Namely, the groups 1 to 4 areconsidered to be constituted from the silt, clay, gravel and sand rich samples,
respectively.
Ce"HELArtLN bF.T-Et- F4LTe-, ;CL'-tS
FkE r[- -P CCHtAT:"N vATE .SUeFILE "NIOOO O,.2ijCi"SChTTEU3FAH OF tvc-N) Fl
O,il v,9t
iNt• srAT:}T!cAL v"uJtS
JAN,le, :vtS) CH:J:.t,.ti-,v
Flk5T FP.Cl'.h'
t.bT 2,"l 1.11
StvSMtNl S:ZE
tAcaoss) j.vi
VATAC-!Tlt
yoplt
",bT'.
JAK,29.
TYPOMtblApt
b.42
i"1)
olAMLTLK IN
b,IT
p-ee 2
V,tt SCALE
6.9it----------t-t--i--"---------"--t-----------+----+---t+"t-----------"---------t----+---t-----+------
1,it
1.ss
1,21
e.rb
e.st
.O.14
.V,59
.x,oa
.;--e
.1,g)
-2.je
tt
Fig. 24.
.
::;lil;;
l;liIl;1
l
1
:
:
:1
tf
'
l
!l
.
1
lill
l:t
I -- -t
1- ?- -l --ul.
.
.
.-- -
:- .
. - -- -1 i2--2
--..
1."....."....."....--...".-...r.."-...--....".--..:..9;";:1-';""-.'.."""".":.""....1
' ': --? -t-.
--
l .. .'21 44- --: lb 3- . I -S-" l -- --; 2} .i --t j2..t'-l--2211.- -st
t
l
tt
'tt
lts
t
1--------------------------t----g------------------------------i-----"---------------------"--------l
i--t-- 1i .. "1 ' .: • ' ' 'l. :: ,. I •lil• x ,"t "'T.21 te-,l
--
. l::t
tt
t
lt
li,
l
`
,
-----t--------i-+-----------------------------+----+----+---------+-------------------"--------ny+------.c.2o u.)) it)u 2.eS 2.ev 3.)ti -.je ).oS .).eO (a)Scatter diagrams between each pair of factor $cores of R-modeparameters: (a) first factor score vs median (Fi-MDPH);coeMcient (Fs-SO).; (c) third vs skewness (Fs-SK).
d,)) . T.ju
analysis and(b) second
+ 2,11l
ll
. :.6e:
:
s
l
• 1.alttI:
- e,?61
:
l
. O,Slttt
s
. .O,1-st1:
- .o.sgti
s
• -1,O-:l
li
' 'i'-9tt:t
+ .1,e3l:. .2,)- ' '
statistical
vs sortmg
ComputerAided Data Analysis .for Geologic Problems 4I
CO"HtLArlCJN tifT.t'[N "LJO" SCOPeS
F:LE !F.1,P (CgtAT;eN bATe .sveFILE ouGoo e,.2oooSCATTE"5H-Y GF {OO.N) e2
• :."T U.1"
Ar.v STATISIIC-L VAi,JtS
JAN.le', !9TS) chlJI"A.eAN
sEcQNv'FAcroR
2.TS j..se 1.Sg----------+-------------------+
SÅíoSMtl-T SJI[
"CSJSS}
-,6a
JAN.29, 1'91)
DATACbJTlt rtPE)
SO SvkTtNG COEFNCIENT s,kb s.es 6.sl
---+------------------------ti---------+----+---------------
PAGE 1-
1.14 ).rl
s.i)
a,ss
1,gg
i.-2
e.e4
O.2T
.e.Oo
.O.ST
.1.-S
-?.o?
Fig.
.t:l'
.1:
t
1
•1t1
l
.I
'1
t
.1
:I
l
+t:
:
r
•I
;
:
•:rI:
+lI1
1
.:tt'
--.
---+------ t-1 + tlT lll l1: 11 -+ S 1- . t llI ' t11 .i .r : -s i -- ; l. 1 l11 i.1 1 t lt1 -l 1 . ;rlt"."g"..".-".""."..-."-"--.".".".9.-.""".-.".-"".-"g""-".""."".""""-l . .
-- -- .
.2e - --2 - - 2- . 2?
,.
.
.
.
.
.
1 .l l `-l : T
;- s. 1 -;
.-l -- I- 1.
.
. .-- . - --
:
l' l' I' l.'l' l' . 1 -s -
l
t
.::
l:
.t
:
:
t
.Is
:---.----.e-"..-------...y.3e......"...-"9..-..-".-.t--g..--.-.--t-.."-.--'.-....--.-..:...--.--..-t'
-l -2-2
t-- 2--- --- 1
.
--- S,lb
24.
.
.
-- -- .
-2.
?- .
.
i- . -----" -• -t 1- -- - -1-22-.- 2 -v - .
.
.
.- -2----
.
-------"-------k-------
.--
1 - l :
i
: I
l
1.1 1 : : l s 1 :
.
.
-----------
.
..
.
.`
---
.
tl I
l
l l
l
.l l
l
I
.I1
l
t
.tll
:
.1t
tl
+---+-------------------+"-t---------------+---------+"-j-+e,
j.Tl
J,13
i.S6
i'e"
1,42
t e.s-
e,2T
.o,lo
.O,ST
'i'-s'
-a,o2
i.Te
Continued
2,4i
(b)
j.e- 3.bT ".JO -.?3 b,)6 S,1" -.bZ T.-S
42 Kaichiro YAMAMOTO
COIIRELATION bFT.Ef.N FACTCk bCOHES
FtLE IEMF, tCRLATION L,ATt .sue-ILE vN!ooe ot,2eooSCATTEHgNAM OF CDO") VO
O,29 O.ts2-+----+----+---------
-hv STArl.SrtLAL V".uEb
JAN.1ts,!,)IS) LHIJI-A.-AY
rHl-•i fl.CTvl:
e."" i.3,J "bS-."--"+ -------
1 t 1 t : t l I 1, t : 1 : : : : t
-----+-
SEDIktLNT btl.E
CA[HtjSSJ
1.g'l
D-TA(tiITh
bk
2.30----------+--t
JAN.29. i91) FA3E 2- TTPO
Skt-NeSS LVEfFiC:LNT
a,6- 2.9s 3,)i),gT
),2S
2.4"
i.IS
r',
,:
I-•!-I
t-.t
l-r
l ---•l -- --;."9-.-
.
--.
.
-ti-F---"- - -
1
i
t
t
:
t:
1
t
i
''
1
1
1:1
----t+--"+---+------"-+----i-.t
t1I
+it
t
l
.11
1
;
.t
----- ---"----- - -- -t
3,9T
3.2J
1.-9
1,IS
' i 1.0: + 1
1 : e.27 . I l ' '.O.41 . : 1
:.t,21 + , 1 l l.i,9S . t l 1 1.2,ee . t r t t.),-s -
.
.
.
---.
---"2- . -- - . - --- --
.
--
..
.
-- .2
-- )- --
2
i-
2
.
?
.
.
. -- -- -t ---- ---? -- -Nt --2 -i-- -2- -2 -t ) --- -- - -•t -2- -2-
..9.-"..O.9.v"9..""v.t)y"gy.
-- . --
.
.
:" L 1 t : `'t 1 t 1 t 1 1 t t
.e""."...".9.
.
.
1 i t t i i i I-1 1 i i
l i
1 t . I i l t + t 1-t : . 1 t
.
--
------------------ ----"--t-----:
1.el
e,2T
.OAI
.
.
------------
.
.
.
.
..1- -i -i --1 -i- -1 -- t- t ' l l : i -: t : , -: -.".- ".-•
;
i
i
t
i
s
t
1
;
J
t
i
i
"------------------"-------- --
:
.tl
t
:
.1t1:
.Iit:
.-+-----i-----"--i
.i,21
.1,95
.2.69
.).4J
Fig.
o.iu
24.
----------- e.-e v.'tg
Continued (c)
1.i3---------+
1.-e i--O 1.1- i.-1 i,el 3,1- s.-e
ComputerAided Data Analysis for Geologic Problems 43
CHIJIdA.CU.H.FA
elLE TCMPsueFILE oNieooSCATTERGRAM OF
2,li
i.66
1,?i
O.T6
O,Jl
.o,14
.o.sg
.i.e-
.1,t"
.i.9J
.a,le
..I
l:
1
. t'
:1
l
.:l
:t
•'
l1
.:
1 1 t.
1I1
1- . I l
1 ' I 1 1 ! . I I t T . t ; : 1 .
.2
F.SCORES AbDeTION
tCRE-TION OATE . oNaeoo (OOtN) Fl
-i,13 -i.i6
JAN,tS, 1"IS}
FIRST fACTeR
-O,S9
CHtJl.A.eAV
.o,o2 e,ss
JkN.ie, 1915
SEVtMENT StiL bAT"-ITH TVPU "eRoss} n sLcoND -AcroH
i,ij 1.10 a,2e 2,1}
VA"L
3,-2
?
--"------+------"-+------- ,--+-"---------------+---------+--"--------t----------------------------
.
2
-"-
-•
.
.
.
.
:- 1.-1--2: - 1
-- t' 1. :-- i- t-- i 1- t
t-- I. I.
1- 1; .i l lI .I t t- .I I.I l; :-+ i; l- -1 .t l 11 1. 1 .tt".........9.."...-".-...".""."""""".".".."""..""...""."."-...--.."""."".1 - -- '
..
.
, -- -"-22 S2- S22- --
. --
. --?-2.2S S ..
-- -- --- -? ---
..
.
.
11 -. 12-.-1 -- t L I : : 1 l-2- l l
t- I
.
.
.
.
.
.
l
i
li
I-1:1
t
1l
lI
1
1
.
t l . t :-i : . 1 1 l : . r l
I-"...."-".."..g-.--"9--..".........-."-.-"..".-"--."....-.".-."."".""."-."......:. .
-- : t : T I-I I t t I 1 I i l : t s
1
.
.
.
--.
-- . -- . 2- --
--
.
.
.
..
.
t t i : 1 ] -1 1 s t i 1 1-i t 1 :
.
.
.
.
.
---------------+------"--"--+-----"-+------"-"-------+--"-----+------------------------------ ,v2 ..i.iS .O,Sl .",10 O.2T o.S- i.R i,V9 2.)b o.i3 '
CHtJ:dA,CU-H.FA F-5COReS ADb:TION JAN,le, 1915 PA"E 3 ' STATISTICS..
CeltRELATION tR). .e.O03eT HSuuSRLD . U.OOeUi S:GNtHCANCE . O,-T"-- STV ERR OS EST. 1,Oe292 SNTEHCEPT ") . O.SeSOiO.Ub SLePE (B} . .O,OOM5 PLOTrED vALu[S. i9- EtCLvDED vALuES. O MtSSiN" vALutS- o ,...-".., IS pR:NTED IF A CeEFFLcSENT CANNeT bL co-puTtO.
(a)
Fig. 25. Scatter diagram among factor scores of R-mode analysis: (a) first vs second factor
(FrF2); (b) first vs third (FrFs); (c) second vs third (F2-Fe).
I . i l l i . I i :. : . : : l i . .J.Ti
2tl:
1!6e
:,21
o,16
o,ji
.O.1-
.o!s"
.i,e-
.x.-e
.i,9}
.2.le
score
44 Kaichiro YAMAMoTo
CHtJl-A,CU.R.FA
FtLE TEMPsveFILE oNloeoSCATTERGnAM OF
F.SCORt5 APDrrION
tCREAT:ON DATE . ON2000 (DOhN) fi
-S,06 .2!S2
JAN,le, i91S}
FtH5T FACTOR
.1-SS
CHIJ:wA.SAV SEO:HENt'S:U
"cHess)
.o,e4 .e,iO o,6-
VATACt:
F3
1,)S
JAN,ie, i91STH TVPtS
rH:HD FACTeH
2,12t--"--t------------------+------e"----------+---------------"------+
2,es---------------
p-Gt
S,6e
`
2,11
i,S6
1,2i
O,16
I t O,Sl . t i t l.e.1- . 1 i 1 t.e,s9 - : :
st.1,O- - 1 1 1 I-i,49 . : 1 : l.t,gs . t 1 1 l.2,St .
t
i'r
tt
1
.s
l1
:
.1
l: .1 t - -1 -- i :. s -.l 1- - l• -- t t -': 1 .-: -l- t s- .: 11 1: - t. 1 t. 1-- - 1 l- t. I t---t----------------t----"------"-t----------t--------
.
-
-
.
--+----------
--"p--"---"---------t- -- ---
.
1 t . - 1-- t- ?-s--" 2-- l I 1 --- , L -"J .1 a- t
:---------------------------t--------
-2----2 ---- - -- .----- --
-- . 2 ---2. 2
-- ---- - ..
t-
-- -t -.2i-t .l -l --I 1 -1 at {l•-t l i 1
-i.."..--"g""9.9.-...."
. .
.-- .
.
.-
------s
.
- -l- I- - I - - 1 l l. - 1- : 1 : i 1. i 1- 1 1 j
. ..
.
. -- . . -- . .
. . ..
. .
.
----.
:1
`lt1
l
t
tt'
t
i11
;:
:
t------------------------------i-------t--------------------------------------- --- -- i,rb 2,4y
JAN,IS, ;gTb
S:tsNIF;CANCE
5LOPE {U)
HISSINS V4LUtS
CoHPUTtO.
--"---- j'ij
PAtsE
...1
. I i I t . I l s 1 . I 1 l t . 1
t 1 . i s-: l . : l t 1 . 1 l
l.II
i
i
+'t1
:
.i
l
`i
.
.j,-s .2.sg .i,gb
CHSJI-A,CU-R.FA F.SCORE5 ADblTtON
STATISTIC5,.
CORRELATION CR). O,eOl-6 STD ERR OF ESt. 1.00a"V PLOTTED VALUES - iV-
"---------
Fig. 25. Continued (b)
.i,2i .o," u,21 i'.ui
RSeuAREO . e,ooOob INTERCEPT {A} - e,-93b2D.OS
EXCLUDEb VALUES. OT$ pRINTED LF A COEFFIclENT "kNol ot
"---- S.VT
'
v,4Sbll
O,OOT]O
o
2.:1
1!ee
i,2i
O,ls
o,)i
.o.:"
-ors-
'i,04
.i,-9
.1,9S
.2,je
ComputerAided Datac Analysis for Geologic Problems 45
CH:JI".CU.H.FA
FILE TEMPSU8FtLE ONiOeeSCArTERGRAM CF
F.SCORES ADVITSVN
(CREAT:ON vAlt - ON2000 tOO-N) f2
.1,06 .2,J2
JAN.le, 197S)
SEtONO FACTOR
.i,se
CH:Jl-A.BAY
'O,S4 .e,io
SEv:MENI SIIL
(ACH05S}
e,6-
OATAcktTH
t3
i.3S-- -- - -+---------+
JA-,les
TyPt)
TfitHD
2,t2
19Tb
fACTOH
2,e6
vAtst
- t-- -+ ----ny"-----+-3,60
b
-+--].71 . : 1 : :3,1) . : : t :-?,5e . I 1 1 l1,9g . 1
-+-----
l.. ..
-+-"-+--- ---
.
.
-t
+----t--1-t
1t
1
:1
1
i1
1
1-I
l-ts
1-;
i i i 1 1-L t l l i 1 l ; i 1 l ;
h .9..9.....9.-..........
--+--t + l : 1 1 + i , t t-+ s l i 1
. t
""i
3.li
3.:)
2,Sb
i,99
l 1 1,-2 . I l 1 i o,t4 . 1 l 1 lO.?7
1 i 1 1.O.10 . i ; 1 1-e.sl . : i i l'i'-5 1 1 r [.2.0?
.
.
.
.
.
.
t-
.
s l i : 1- 1 s i- 1 i- ;-- 1. i-'i
lt '
.
.
..
.
-? ---? . ..e ." .
t] -- l-- s-- -t --2 tt- -- 2
l -2 ---- 1-t ----- --2 -t-- 1
i i l i i 1
t
1
.
. .--
-- -- ---- .-t .v.-2t t . -- --"? 2 t ---t2- --
-- --
9" -9 99
i
i-i
i
1.I
:
i
si
l
iti
t-t' v.
.
'
.'
.
.
t i
i 1 i z . i l i 1
I 1 i ` 1 i i 1 . I ` i i . ` t l i •---).Y7
2,-2
e,e-
O.27
- ---
..
22
.
+ ---
.
--) 2
-- -- -- - .
.
-
- --2
.
2
.
.
..
t- - -t-- -t
i i , 1 i.1 i- 1 i t'
1
. 1 l l t i
-- .
+- +- - +--
.
---t ----
.u.]o
.o.e7
.1.-5
.?,02
'j'-] .2,bV .1.gs -:.21 -O.-1 O.2T 1,Ol 1.75 l.-9 j;U3
CHIJtptA,CU.H-FA F.SCORES ADD[TloN
sTAnsrtcs••
CORRELATiON (R). .O,Ob3S" STD ERR OF E5T. LO]3eO PLOrTED VALUES . i--
--------{
Fig. 25. Contuned (c)
RsteuAREo . e,ou-be INTEHCEPT (A} . .e,2]eS9V.Ub
EXCLuDEV vAL"ES. O
, ]S PHrNTEP IF A COtFFICILNI LANNVI et
JAr,,ls, lg7S
St"N]HCANCE SLOPL tv)
MISSiNb VALUtS
CvMPblLD.
eAGt 't
O,ISS2h
.v.ebl:e
u
46 Kaich iro YAMAMoTo
dtemeter(6)
8-
7-
6 --
5 --
4-
3 --
2-
1-
o --
-1 --
t
.
LEGEND
.-----•----• e2 Group
------ A2 Group s•- c
GrovpGroup
. -- it- -.- -- -" 4-- --.-. i- tt- J-- -- -t --- --- -e- -- -•" -- ".r -- -- --i J-- t. -"- --- -I- J- - -- -t- ---tJ- - .- .- t ." " .;•. t "t'- tt ..--111.tl;;.:-;.-•.--:..
:.:.;.:',.-E{{.Jvii-:':'..----".-
' /::
.- ----- i-lt- - e-" --- --- .. t- t--- .-t . e- - .- -- -i -- ----t - .- .--l -- --tT- --i-
l / tt t- t -t tt /t tt -t t-tt t- tt tt';"'t::';"':';t..t".F :-tt "--
-.
/
-
!
•2104
.2oel""2016--
1
-
t
xle23
tlO19
-.1048
t tl064
103]200B1021
c2053 2156
Z-7 //2-/
.
-x- /-:/-
!• •/-
.
7 / /` '
-1040 2147
// 2013 !.-:-.2iS7'
/• t---2oe6
Fig
•/••/•l
10
/. 26. (a) Cumulative curves analysis, and (b) where arrows show ascend
ltltll 2o 3o 4o so se 7o cumulative propartien {X)
(a)
of representative samples of groups defined by factortheir points plotted on Fi-F2 diagram of R-mode factor scores,
ing order of cumulative curves in (a).
Computer Aid ed Data Analysis for Geologic Problems 47
2oete 2 e21e4
2013e
1 LEGEND
eD26roup
OMGroup
102 OB Group
ec Group
1019
-1 104S 1
2ooe
10 1 i064
1033 21S6
21S4
2147 lo4e-1
21S7
20i3
-2 2ee6
F2
Fl
Fig. 26. Continued (b)
Table 4.
(a)
Results of Q-mode factortion of total variance; (b)
analysis: (a) Eigenvaluevarimax factor scores.
and cumutative propor-
Factor Eigenvalue Cumulative Proportion
1
2
3
4
5
6
7
133.12
53.76
25.65
10.05
4.46 3.17 1.45
56.65
79.53
90."94.72
96.62
97.97
98.58
48 Kaichiro YAMAMoTo
Table 4. Continued (b)
Variable1 2
Factor3 4
VAROOIVARO02VARoo3VARO04VARoo5VARO06VARoo7VARoo8VARO09VAROIOVAROI1VARO12VARO13
O,OOI
O.O07
O.O19-;-O.O12
-O.139
O.344
O.840
O.318
O.208
O.053
-O.O13
O.053-O.077
-O.oo1
-O.O05-O.O13
-o.oos O.044-O.042
-O.039 O.155
O.233
O.327 O.287
O.146 O.8tro
O.031
O.140 O.433 O.741 O.439-O.158 O.073 O.096 O.087 O.Ol2-O.Oll
-O.OOI-O.057
-O.O02 O.O19 O.080 O.036-O.525-O.808 O.150 O.154 O.118 O.O02-O.023-O.Oll
-O.056
C-1"--, e.FA AeDtllON
ML[ CHIJIV- ICREATIew- pATE .sueFtLt oHiooo oN2oooSCATTERGns" oF COO.N) eFLi
e,enb e,ns-
MAy lu, teT5)
f:RST f-cToft
O,21"T- -- -e- l ila.- -- :- :- --- . " -:.
--2
C"IJtbeA.tiAv SEOT-ENT S[tE
LO-OtNG oF ---oOE f "cHoss)
e,J21e o,-?-e -.sllq
JATAC"IT-
aFLI
O,elil
Juh. 1, 19TS eAGE aTrPE)
sEco-p FscToft LoAolHG oF -."eeE
e.11-2 e,t}TJ O.e-O-e,esto
o.-se}
e.k3o
o,ssss
O.e210
e,-leJ
O,)110
e.2es}
e.eele
-e.oe-}
TItll;ll-;
I.I-;t--l.; ---
l.:l
t':'Il
-2l'l S•
;-l: :;2)
l "'1:---t--
.
.
.
.
.
t
.
----
.t
- --
.
.
.
.
- 2--
"i . .
.
--
---- -- -t- -- +--t t t-t-
-- t:
.. 2 :. :. .. . .
--
.
'
--
.
.
---
.
..
.
.
:: --:- -
..
.
-t -- --- --- -- +-P--t-
.
.
-
.
.
.
.
:
..
-- 2-
. --- f ,:2 2; 3;T . ::l 2t r : t---------------
2: . -
-1 .
--
.
;{:::;:i::;Itl:;:il:
e,t}so
o.tfos
O,T410
e,-sss
e.s2eo
O,-10S
o.S"o
o.ioss
o.ettn
.o.oo"
.entlo - .o,za7e ---"-----t--- - -- ----------- t- t-t- --- -t-- ---+-- ---- -t--------- .o,ol-e e,a-" ,o,i-r2 o"UOI o.jT3- o,-T-t o,sT-s et"21 a,11sl e,-e-t eitt2o
(a)Fig. 27. Scatter diagrams among factor loadings ofQ-mode analysis: (a) first vs second factor loading (QFLI-QFL2); (b) first vs third (QFLI-QFL3); (c) first vs fourth (QFLI-QFL4); (d) second vs third (QFL2-QFL3); (e) second vs fourth (QFL2-QFL4); (f) third vs fourth (QFL3-QFL4).
Computer Aided Data Analysis for Geologic Problems 49
CHIJIvA. e.FA ADDIT:ON
FILE CHIJt-A (CHEATION PATE .sveFtLe oNleoo os2eeoSCATTE-GftA- OS tDevN) eFLI
.o,o-s- o,osss--- -----
-2 .1 )S . 22 S2.2 --1}- al--1.- -z- 1------ . -- ---t . . • - 1-
. ---- --- 2- -2-- . -- --) - -2 -- 1)- }) 2. )- 2
-Ar 12, :sTS)
FtNST PACTVR
e,isfo
CHTJt#A-eAV SED;"ENT SIIE
LoADING OF e."boE F "cROSs}
o`eTol o,}Ts- o.-eoe
oATAt-ITH
eFLS
e,ses--t---- -- -- --- -- -- ---- --- ---
JUN, 1, IStS -AGE -TvpE)
THI-O fACTOR LeADING ot e.MOOE F
e,sglo o,Tg-a e.gos-o.-seo
o.lses
e,nso
e,-sss
o,s2te
e,-2ob
e.11}e
e,2ess
e.oeee
.o,oe-s
-e.:11o
-----•tt
l
l
.1
l
:'
lt'
:
1
ll
t
t
.1l
II
Tl.
tIi;lill;lij------- --
.
.
.
.
.
.
.
.
.
----
.,
.
.
.
..
.
-t
.
.
.
.
---
.
t-
2-
.
.
.--
.
--"
•
.
.---.
--
------------
..
-t
--.
.
.
.
.
•
.
.
-,
.
.
---
.
.
------------- -- "-+----+- ---- . 1 1 l t • t I ' l • 1 1 : 1 • t t t . t t t t . I ' t t . -i t : l : --- l -- : . -; 't -1 -- "t -- l . 2 ---. -ell -I l 1 --+- -------t--i---+ -- - ---- -- -+ "------ -"--
e,-seo
9,ises
e,T-se
e,"ss
e,tlto
e.-2es
o.3iso
o.less
e,e-en
.o.ees-
.O.1:ln
.o,o"te e,oen O,ll2- o,2zl- e.s22s o,-2-o
(b)
O.)};1 o,-J-- e.T-3s o.e-e- o,t)-o
CHIJI-A, e.Fk ADOIT:e"
F:LE CHlt:-A (cRtATsor- OATE .suBFILE eNiooo os2eoeSCATTERGnAH OF Cco-N) eFL;
.o,1-eg -o.n"
-Av 12, telS)
FIHST FACTen
.O.S-67
C"SJ:tA.eAY SEDt-ENr 511E
Le-D:NG eF etMouE F cACHOSS}
.o.sS-s .a.4-2s .o.]lo-
DATAC"ITH
aFL-
.O.2tSS
--+ -- +
-2 Q-- -- --2--S
JuN. T. :"rS PAaE STypE)
.Fou"TH FACTeR LOAD:N6 ek e-MooE
.o.tasi o,ebse o.tneb."seo
e.tses
o,T-so
o.6sss
O.f210
'e;-2es
b'.iiso
ti;2ess
o' ,oslo
.o,eets
.e,111e
---.1
tl,
t
.:t:•
t
tt'
t1l
.tT,
tl
.•s:
1/
t'
'
rlt'
:/
l
,kktt'"StL,
,t•
lt:
.s11'
.
--- -- ---------- - - "--- ---
,-
.2-.
--- z- ----- - a.t---
.
.
.
.
-- e.
.--2-t--
.
.
.
-.
.'
2
------ - ----
. -- -e2-.
.
.
-- .
"-- -- .
.
.
.
t2 -- --
.
.
•-
.
--
.
--
' --
-- --
--
.
-.
t- ---
.
.
.
.
.
---t- --
2 ...
-- '
'
.
.
.
.
.
.
.
.
- -}
- -- -t+ - +---+-- . .
.
.
.
.
--.
1
.
.
.
.
.-- t .-t-
.
.
-- . -- - --2 . -- a 1- -1 -- ---t 15 - --2 -S-- -- . .
h+•t
t
tt
.tt
tl
.t
tl
-t.t;tt
.1t
't
•t
l1l
.1IIt
.1
t,l
.t
slt
.tt
tt
.- - ---"-"------------------
n.-sea
e,tsel
o.Nlo
o.-ess
e,"lo
o.-2ef
O,S"n
o.ioss
o.e"-a
.o,"o"s
.e,uTo, i-----+------ab.e41o .v.-}--
Fig. 27.
--------.e.lns
+--- -------------"-----+"------e.-lal .e,-ee- .e.le6s
Continued (b), (c)
(c)
.e,214- .o.1-2J .e,ese2 o,ee:e e.iko
se
CHtJltiA, a-fA APOIT:ON
F:LE CHIJ[tA CCREATLON uArE .sueF:LE oNsooe oh2eeoSC4TTERGR-H oF CDOv-) eFL2
-o,o-)4 o,ossl
kAV 11- 1"TSI
SECONu FACTOR
".IS50
'Kaichiro YAMAMoTO
CltIJI-A.tiAT SEOI-ENT sttE eiTAt.tTti
LoAerNfi OF in.-oeE "CHuss) aFLj
",?lo2 u.}ls- g,"sts ",S"hs
JuN, T, lgTO P-eL ITvpE}
THIRo FAcTOH LoAolftG eF -.-ooE f
O.b910 o,19S} ",9Di-e,"e?o
a."e-
O.TeS-
e,"11
o,blss
O,-lsb
O,)1:-
e.nos
O,1-7t
e,e--i
-e,o)-e
:
1L
'
:
::
1
s
1
;1
isl
J
1
:'l
'l
:1
'
11
1
ls1'
t
s:l
s1L
1
S"TJ2 -22---- - 2-t --2 t- 2- -1- 2
-- - }. .
-- --
' --- - 2.)2 - - a-- - --- - - - )-- - - -- -- - 2 -- - -- ---- 't 2-- . - -1- --- - - --
.
.
. -- t--
1 I i t 1 1 , ' 1 ' t 1 1 t s i s t ' ' : t : : t t 1 -1 -: : 1 1 . -1 l -: -l - - I -- t- -- - -- -- 2 S . 11)l
o,II2- u,nTi e,)22- e,-2-o o,ssll o.-Sl- o,T-)s o.t-IS e.-e-e
(d)
JuN. T, ISTS PAGE 1" t"lf) C-tJt-A-"Y SEDIMENT S:IE bATAt-ITH TvPE) fACTOe LeAbtNG Of -.MOPE tAcROSs) oFL- FeuRTH FACTOR LeAO:N6 OF -.PvDf
.o.Ss-- .o.--2s .o.ljo- .e.lleJ .o.:o-1 o,eOtl u.:il-
e.--2e
o.l-t-
e,rlsi
e,sl21
v.Sl-s
O,-lss
e.11s-
O,1101
O.lbll
o.es-1
-e.olee
.o,e-ee o,oe72
CHIJ:"A- O.FA ADP:TION
FILE CHtJlvA tCHEATIeN OAtE . HAV 11.suaFILf o,tteoa o,,leooSCATTERGRsM OF cOO-N) "FL2 SECO"O
-e,-eo" .e-ITte .e.---re,t-ao
o,let-
",}tsl
e,-t2T
O,57f-
e,-lis
O,31}-
O,210}
O,1-TZ
e.o-4i
.e,e3-e
,t
t
L
lt
,t
l
t1
s
I
t
1
I
stt
•l
.I'
:1
s,:'
.t1sl
1l:t
1t,
:-•
-----
. . .
-- -2- ---- --- --- --t-- - -t---s------s- -
. . .
.
. ' . -- . - -- - -- - . ------- -- - --- -- - .-e .- -- -- -- --
. . . . .--
. ' .. -- ---- -- 2 .-- --'.
.
- . . .
..
.
.
.
.
.
.
.
.
.
.
-2
--1-2-J. 1 2Pt .2. t -2- l --- l -2 - -- - i -t ; •--1 I -t -- t . : l 1 7-1 -- -t ' 1 el . t -I l 1 e- t -1 s 1 -- t l 1 tt . -t -- l i t l : 2l- t - t --- ---"- -- +-
o,"ele
o,ll;-
O.llS"
o,ent
o.eTti
o.-T"l
e.sTM
O.ITOs
O,;-lz
o.e--t
.o,esto-+
e,-"o .e,ls-e
Fig. 27.
.e,na- - t ----o,-iol .e,--l- .o,;"s
Continued (d), (e)
(e)
.a.tT-- .e,l-ts -e.o)el e,e"t e.u-o
Computer Aided Data Analysis for Geologic Problems 51
CHtJl"A. e.FA ADDITION
FtLE CHtJl.- ccaEATJON DATE .sueFtLE oNieao oN2oeoSCAT;ERGaAN OF CbO.N} efL3
.o,seog .etllSe
vAy 12, 1-IS)
TH[RD FAcTeR
.O,bSeT-----------------------;l1ll
l
l1
J
l
1
l
i/ ?: -:l.!: ..]2l-- -:-;- --; tll--1 ---- - --l --T.I-- --:----I
:.--- ---- -- - ----- -
ClttJldA-S-V SEDtMENT StlE
LoAoJbeG oF a."ooE F "cRoss)
.o,SS-6 .o,-"25 .o.Sle----------+-------------
-- -- . -- -- . -- .
.
. . . t . . . -- t -- 2e - -- -- - tt
t------- ------+------
DArA(lttTH
aFL-
.o,1:e)
JuH. 7, 1-1) PAGE 12TvpE)
FouRTH FACToR LOADING OF e--ODE
.o,loeJ e,oese o.liT" o,g}-o
o."ts
O,1-lh
o.6}l-
e,s)n
o,-aee
e.Jllt
O,211b
e,u2-
e.pola
-D,09IO
.
.
..
--
'
:
:
.
.
.
:.22
-------t------------v-----t----------t------ t- -t -t- : -- 1 ' : -; •l : l l ; l : l l i l 1 i I ` : t l -- : -- - -- -; t --- -- - -2 J --- --- t s'S' 'i::: ]l:?: '.li.;;i'n:" :
.-2 - }i:IS2. I l .- -----------!-------- +-----s---t---------t-
o,-s-e
o,"II
O.1-IS
e.il"
e,ss)7
OA7-o
e,}11-
OilLl-
OiU?"
e.oo7p
-o,e"-n
.O,--10
Fig. 27.
.o,S]-" -O,12?S
Continued (f)
-O,tlol .e.-9jb.o,le:;'
.O.2k- .e,lb2) -e,o}c•2 o,o-;g e,;1-e
Cluster analysis
Several methods of the cluster analysis were carried out, and it was clarified
that the cluster analysis based on chi-square distance of frequency scores is the most
suitable in this case. The analyzed results are discussed below.
Four major branches are observed in the simplified figure (Fig. 28). Theupper branch is further subdivided into three subbranches. As the samples be-longing to the lowest branch are rather limited, the branch is omitted in this case
study. The second and fourth major branches are subdivided into two subbranchesrespectively. Thereby, seven clusters are observed in the figure, and named as the
Ai, A2, D2, Di, C, B2, and Bi group in descending order. The group names corre-spond to the classification by the factor analysis. By using the dendrogram of the
original outputs (Fig. 29), the more detailed classification is attained, as shown in
the figure.
The convergence of each cluster is very good, and the separation of one cluster
from the other is also good. It is rather easy to recognize the cluster structure.
52 Kaichiro YAMAMOTO
A.i
A2
D
Di
c
a(
Bi
Fig. 28. Simplified dendrogram of output shown in Fig. 29.
Q-modecluster analysis, compiled from original
Computer Aided Data Analysis for Geologic Problems 53
llt i ttt t itt tm# :, r:r:'r ,' ,'ra#ti r::r ,, ,' ,'tltp" -lt'pl t tlah .ti N" S::I;-;t ,- t'ps.Ek'utlli7" r" 1`ItT-" it,-- t tl""" ",- 1 tAS- -l i.t- " -tts"r, ,r:L..-.t rsk tiN -,----j 1S--" -.hl Ii- --N -.I- -EIZ, r:•F' /:me::::--m- -h.H--ny--.-..ilt--"- ",rt t t li-"- T-,-t t J -.t#:: ,:::t-,' ,' r-"" -t,it J itsk": r, .",,IF' 1 /tU:tt ,r,:r-;,t ,tR-"" -1.--"t P=:: it:rr xi--" ",v tIS E Ntu lllll ., , -.,-,•t
:ln n r.:r,r ,, ,r ,ti)- 1- et IL-,jt t Ii jt--S-" S--,iJ tttI-S"" :--..-l N tl:g: :#];' ,r rptr: l::lrt ';r :::: # tr,:L:, .,ral' tt tiigr/"- -i i-t'-ji"" -i i".Ir-i u: iel--s ve:r,::-"--..".
i- ti ---- -"-"- --------------------i--H-""H-------H--------"i;i it lr:lll / .: ,,' r'Sttl t--j J l -t-li i-i't t t ti;:: ;::rr,, r,=' ,' ,' ,'Ir:IN-l:irl•;1,::,l,,' / / /--i Å}it.tl- i i t"-i -h.Jt I iJIrer: ::[t.I- tj -t tt,g/r-l:-Fal/il';,;-;---r;1. ,•" !'
rc"' rP ,.ts,:,r•'r' ,•' ," ,,itn: r, :u:=r ,, r ,'tfiIF "•l::.-.--",•' ! 1FltrAn:1•fr•' .i /' /'ee S,A•,'ik'•-"-];,;i' ,/ /tEel :"rl:-:,;,t' ! 1
t-t "tGillt.-.-."H--H-...-H..H.ti.t /I
--- "'t-t- t t:h'Il :tGELI ! :'-s- ",lt j tu:t ,.,:: ,, ,.Itl- tH.- t -N-l "-. ------ J -"s- -,tl-tt t t"-i t-wt-1 - I---ii -t'-t j- t tiJN- -.-t ii t -"-t- L-t i- t tL'"t l"':L.-,',' r ,'$;i.Ai,ttr,,lil,'1/ // /")-i s".-.I tt tj"ui "o.-t 1 lNit: ,::;' .' r .tI---i ltS..---.--.- t i""- n,tflt tptgi -.--t Jt l)-t-i IT.-t tj -?,El ti,•ilL---/ ,j' 1teel iti,ll;•:;,'1/ / !-"i -..tl J --- i-- s"'l- t l::: ri::' x ,jsk/ll-l,iH, li/l;Tr-;;i ./;U ,' ". t:::tr .. .tic" .r Fk lEif;-;•' r'r,t::, t:;:Ll ,'ta'irg-iti'i•i:';, "ra: r, #:r
-----------t,
j
,
''
'
''
''
'
''
''t
't
t
j
't
''
,'
''
''
,
'',
''
',
t-- tll tt tt lt tl ttl"' Illl tl tt lt Jl -tJ- tl"j jtl ttl }l IIt t-t It iJ ' I"tt ttt ltj jtJ tt-- Ittet-
' ' t ' ' ' ' j ' ' t- iltlt
j , Iltti-
1"l t-x ' tt ttt- it ttt i" ttittt
'
''
''
tt
J
'
't
''
tt
'
'
''
',
'
mr.: :#Ir ,' ,' rMEi M.lcr,;'/ .":EEi kti.I7,! !tSNt ---,-Hl t,mrt IL':v ,':r: rc:' .--.rm:Ic ,t::-.,r ,tÅ}::: Ltt:L.,, 1 "ME{ ,111;:' / /mn: r,l::::.:=t rre.;'i n'ill':;.1 ,,'
-"C -b-r i-IT-t --''- -:ve':t•----ny-• •-----.--.-•,'-- --j :ti'I- lt 1 t:h l-t :T-- -t l jst:t LU:;-rr .t ,t"-- Å}".-t e t-ki :-,it - a'EEl& lii•iF•' / ,rilEi •rliy"':;' ,x1-1- 1",-.-t -"Ei irlf:,:t,' ,rEuar,I::"..--...-.kH..-.!
-kl- -"J- t -":k: i-.ltr .' xt::: ';::r,r:j .jikiC -S."- J-"sc "w, tra:{ 1::;;' ""Ht Jli,i 1pt: - IiT.---+----.I:M :Bt ",::r r-iSl- S--J --"t lt1'"-"-"1-i"- li-,t IMId - tcl."H+---t"- I- - I-,ltt -:c:.: ,::1:' rt:t ; : :::::::;t
,nr: ,:::;r-"- t-"
''
'.
''
t'
,
,
''
'
J
Fig. 29. Dendrogram oforiginal output from detailed classification is attained than
computer.in Fig. 28.
More
54 Kaichiro YAMAMoTo
Statistical significance of classification
Classi cation b] thefactor analysis: As shown in Tabs. 5a and 5b, the distance between
each pair of groups is significant at 1O/, significant level, and the equality of means
among the groups is rejected 10/. significant level, so that the multivariate means
of the groups are different each other. As shown in Tab. 5c, the discriminationof each sample is excellent, for almost all samples are discriminated into correct
group.
Table5. Discriminant analysis of groups defined by factoranalysis with frequency scores and phi values: (a) Test for multivariate equality of means; (b) F-distance matrix (D. F.=14, 172); (c) Re-classification by discriminant analysis. (a)
U-statistic
Approximate F
O.02927 ,17.72478
D. F.=14, 4, 185
D. E = 56, 671.22
(b)
GroupA B C Dl
Group
BCDID2
15.24
47.29
14.44
tlO.05
9.66
30.02
62.26
58.37
104.13 7.94
(c)
D2Number of cases classified into group
Dl A B c
Original group
D2DlABC
22
2
o
o
o
4
21
4
o
1
o
1
67
1
o
o
o
3
28
2
oo
o
2
32
Classi cation by the cluster anatmsis: As shown in Tables. 6a and 6b, both the distances
among the groups and the rejection of a null hypothesis on the equalty of means are
significant at10/, significant level, The discrimination of each sample is also very
good (Table 6c). Several samples of the Ai group are, however, discriminated into
the Di or A2 group. The fact suggests that the Ai group is situated between the
Di and A2 groups in the space, based on the variables used for the classification,
and that the Ai group is a transitional or overlapping of the latter groups in the dis-
Computer Aided Data Analysis for Geologic Problems 55
tribution pattern of grain-size. Judging from the distance, the groups may be ar-ranged in the order: the Di, Ai, A2, Bi, and C group, on the basis of the similarity
among them. Examining each neibouring pair of the groups, the distance betweenthe B2 and B! groups is especially small (4.14), in comparison with the others having
usually 11.85 to 18.11. The division between the B2 and Bi groups is more detailed
than those of the other pairs.
Table 6. Discriminant analysis of groups defined bycluster analysis with chi-square distance of frequency data: (a) Test of multivariate equality of means; (b) F distance matrix (D. F.=14, 174); (c) Re-classification by discriminant analysis.
(a)
U-statistic
Approximate F
O.02048
l5.02178
D. F.=:14, 5, 187
D. F. == 70, 832.49
(b)
Al A2Group Bl B2 c
A2 BlGroup B2 c Dl
l1.85
30.72
21.39
82.12
18.77
12.68
10.90
54.29
34.43
4.14
14.41
64.34
17.12
46.39 127.73
(c)
Dl Number of cases classified into groups
Al A2 B2 Bl C
Dl AlOriginal A2 groUP B2 Bl C
25
4
o
o
o
o
1
41
1
o
o
o
o
8
39
3
o
o
o1
2
13
1
o
o
o
o
2
25
o
o
o
o
o
o
27
Ctasst:tiication made b" Kamada et al. (1973): As shown in Tables. 7a and 7b, a null
hypothesis on the multivariate equality of means is rejected at 10/. significantlevel. The distance between each pair of groups is significant at 10/. Ievel, except
for that between the 2A and 2B groups, which is not significant. Combining the2A and 2B groups into a group, five groups are tested by discriminant analysis.
The result (Table 7c) shows that many of the samples belonging to the 2A and 2Bgroups are re-classified into the group 1, and that some samples of the group 3 are
56 KaichiroYAMAMoTo 're-classified into the group 3A. Therefore, this classification is indistinct especially
in the separation between each pair of the groups 1, 2A and 2B.
Table 7. Discriminant analysis of sediment types defined by Kamada et al. (1973) : (a)Test of multivariate equality of means; (b) F distance rnatrix (D. F.=8, 215) ;
(c) Re-classification by discriminant analysis.
(a)
U-statistic
Approximate F
O.03000
29.03202
D. F.==8, 5, 222
D. F.=40, 939.96
(b)
1 2AGroup 2B 3 3A
2A 2BGroup 3 3A 3B
5i13
7.38 5Lll21.14
106.51
2.42
27.18
11.67
60.90
82.21
35.58
216.34
22.88
40.25 109.80
(c)
1Number of cases classified into groups
2A,B - 3A - 3 3B, 4
Original
group
1
2A, B
3A3
3B, 4
14
12
o
1
o
2
ca
o
o
o
o
5
39
17
o
o
o
1
381
o1
o
2
56
Description of groups
Breakdown: The general statistics of each group, which is classified by KAMADAet al. (1973) and the cluster analysis, are shown in Tables 8 and 9. The tables describe
the mean and standard deviation of the statistical parameters for each group.
Table 10 illustrates the difference of each parameters among the groups. Themedian, sand, silt, clay and mud contents are remarkably different among the groups.
The difference between two classifications is tested by the F-values of gravel contents
(G), sorting coethcient (SO) and water depth (DEP).
Triangle contour diagram: A
is produced for each group,
the original dendrogram (Fig
triangle
which is. 29).
contourdefined
diagram ofby the cluster
sand-silt-clay
analysis and
(Fig. 30)
shown in
L
Cornputer Aided Data Analysis for Geologic Problems 57
All
mean
the groupslocations are
except for the
different fromgroupeach
Ao show good iso
other as shown
lation
in Fig.
in the figures.
3LThe
Table 8. Description of groups definedcompositional parameters.
by cluster analysis, concermngstatistical and
SaEAmol- ot C"IJ:-A YAV 1. 191) PAGE -FtLESU-FILE
C"IJtvi ONIOOO
-----c-:rEntoN baptEN-----
CCHE-rlo- OAtE e-loou
vSnl)bLE eovN sy------
. -a-.2e, lvT))
---- +.•McP- EDiANCA") GROLP
t5cF: DtA"Åíltkt4;f" '"-
{-tJt.A.e-.v
pr:v IN PHIF".Cb.CA----
StUl--T
N UpSCALE
SrtE "AT"vtTH TVPE)
subpopuLtrloNs
-------------------t-----------------VARIAeLE
FOR ENTI"E
CAG)CAGSCAG}csfi1C4")c-o"-j
PoOvLiTION
ceLti
Ali2et-2cDlbl
siLL,E LleEL su-
-S-.-itjb
2.b.]oO14T.iOe )1.eoo S-.aoo 10.42b2jS,tCClo).aeo
ME-N
..i2Sl
i,s-)J.Jol1,g"22,700i.blie.S)1T,see
STD OEV
1.:es2
1.ul"e.2slo.bi)O.?Tle.T25e.1-2e.loe
vA"IANCE
4.Tt21
l.:6-e.o-)O.2sSe,ol-O.S2So.sile.ovs
c
t
(
(
{
(
c
t
N
2SS)
s"-1)as}ae))e)s-)al}
CRITEgtew tiee`EN
------vA-t-eLE oo-t- sv
------so"eJ
.... otScArpTloN oF suepepuLirlONs ........-........ SURrCNG COE-FIeSENT e,?bep f-sEo :,. s--Lo.t-
ViAl-eLt
FOe fNTIHE
ci-}c46s"G)CkScAeS"6)"Gl
-JPuLATI:n
;o"t
"uele2c
Jt02
v--Ui LA5i.L 5u-
bb4.Sloo
iY).)10 9i.)10 ]s.O)u lr.elO )),21ol'J-.e"uSJ.SSO
-Eit.
2.tlT:
3'-tT:'i9i1.]sT;.e"o:.11--.)3b),i-s
ste vLv
1.S)Sl
1.-2t)e.b-sO.1)-c,-ilc•.)l)t'i?)o,)7-
vi-t-kCE
1."o92.elea.4sTO.Gi-C.;1-e.21e'i.2h:o.ies
t
(
t
c
(
(
t
{
N
tls)
}"-1)a-)to)Jo}"}aT)
-----C-JTtalow eqo(EN'r-'' Lva-lieLE
Fpn ENTseE
CAG]CAeSCA"CA-JC4s)cAes"GS
V-nlASLEb:d- ev
PJpo"rp3N
s4CAS)
.... DESCglPrlO- bF SUsPOPULATiONS ................. StL-NESS C:tFFIctENT c.Ro"" -4sEb :N ig.ce.c-
CODt
ilnSle2c
D:Ol
vA.VE "eEL SUM
tTo.-noo
)-.ebO2a.-3o1-.-Totl.ISo}o.)ao2].)eot4."e
HE"o,T2SO
c.s-'te.)4i1.e2sC.eS2l.02"u.s-ge.si)
sTD DEv
o.S-ii
o,4isO.116o.illo,4s)o,snO,26)e,:11
v--TANCE
e,:S29 te.lst {a.o41 (o.ets ce.ls" te.21) ce.ost to,elt t
H
1)})
s-)-))?s)}o})e))s)ll)
CNtTfntov eebtE-
ViNl-eLE eo-N lv
------GCiGJ
.... L 'sk-VfL ""OJP .
EscgtPTI)N oF subpePuLiTloNspirJ:eAsEo ON Fn-:e.cA
VAAIAeLt
fe" fNrlRE
ci"]c4"1"GSCAblCA"SC4eJCAGS
psu"rlos
CObE
il
o"ts2c;lo2
vhLJE LAtitL SUM
1"V.4JOO
c.e o." b.O tb.loe1--.eoo2C,tOd
S-3DO
-E-N
O.S-T)
e.oe.ec.o:Ol)"t"7O.Sli;,1))
srb otv
saliib.Oo.oe.o2.-SS-.sil1.sfo1,OOO
vARtAecE Sle.tl-2 t e,e t e.e t o.o { s,o-s c-1.S-- til,aa2 t o,9-- t
N
IJS)
s-}-t})s)to)oo)s-}lr)
CelTEalON eaptE------
vA"t4eLE DodN eN
stCA"1
--.. o S-N,) G"OuP
tSLHLPTIUN JF SVbPOPULAt:ONS ........-........ftiT[) IASEv :H ilt-Cb.C-
------------------t--t-------------viH:AeLt
FOR E-TLHE
"""G3C-b3CAG)C"G3C-G3C4G]
PO-,uLATIVN
cvvt
-1-:b:dt
o:vt
v4LUE LAaLL 5U"
lz)gl.jv""
1-so,zeetTll.•jOu2h2e,oooISIe.leulbe-,;OU SJ?,-"U 1)i-lou
-EAN
sJ,fei-
-b.11-b-.902-1.2-)ls.SiUsg.Se]1-,Tev -.-le
STD 1)EV
S2.;IS4
i).910 e.)}o 4.g)1 1,'tol:a.t-o t,)11 i-}-s
vARtANCE
to-i.eiso
1--,e41 -xl)e 2-.S)- 1"-)"-nl,-Si -1.}9t 1.--T
t
t
t
{
c
c
c
c
N
IJS)
si)-1)1-1le)oo)3-)alJ
cntTERso- eeocEH-----
v-t-eLt Do-N ev-----"
SlcA-)
.... D SILT GROoP
tsce;PlloN oF suepop i.)LATToNsA-TIO eAS:U Oli rH.CV-C-
VA"1-eL-.
fe- ENTIRE
CA")CiGSC-GSc-ejC-6)Cie}Ca"s
po"uL-TrJN
:OvL
A:A2t,1
eltJL;t
JAvJE L-"t'. s"y
eT:e.-oot
in19,eoo --T,b:e "),loo?lr,-oa
"-.Sccle16,Seeil)-.10u
YEAN
1-,Slt}
)d,}ISli,5-i 1,Tsdi).t7e 1.ITJ)i,11)ee-}Jl
.STO oEV
lo,evi2
11,li. b.OLi 1,jSle.-e2-nsT
e,e994.ol-
VM:ANCE N
iOS."1-: t 21S)t2-,lgl t 1-1 )S.11) t {2) il,;eo c ls)-1.e)) t 2e}-e.-e- ( )o) IS,-II ( IS)
iS.lgS ( 11)
58 Kaichiro YAMAMoTo
cR;Te";pt, e"o"N
vAal-bLt
FO" ENIi-f
ch")cse)c--3C-GJc-"JcAa)Ci6S
v-"l-bLC CL VOt- bV Ci"S
P'-ULAIiON
..-. J' (L-V tsROIiP
LOOE
i:il-i52L
OID2
i. 'j .11 ,P+10- 1. JU ti OoPvLitl:NS-iT:j b-SE/, JN -'ny.Cu.C-
---.-------------------------------------",.JE ,.t,fl:.L 3JH
.05),)bOO
se9.)oO bib.lb,] n,eoo ;sl.eoo ei,-oell2-,2bO12"9,"Oo
gEAN:T,ISb"
lb,i02u,b-o o•ese T.S"O 1.-?O1}.e9-i-.Sli
STD oEv
:s.1}-O
9.11) ).ii- 1.Tle 1.l"S 2,-S) lt,1"- -,:os
vtglANCE
2.S.11jS c ee,Oel C 1).101 t 2--s- c -,tle ( e-l"O {11",;l- C i-'lsl t
N
1)S)
si) ill 2" 20) se) s-) tn
-----CRITER:oN S"OtEH
VAal-tiLE
Fos ENr:HL
CAkSC-GjCA"JCAts)c,la)
CAWCA6j
vSRt4SLE "uJ o:-- et c4"1
POPULAT:ON
---- t )
-u" "- r,eOuP
coot
i;
nele2
viJ?
L.scaTO-t:N oV suePePuLitlOhSiw,s-SE. :N g".C,).CA
--------------------i--------------------vALJI LieL4 SVH
;ollb.aoot
lo-".tna1"4,200 11.JOO -l-,loO ib-,IDOSO",SO,J2Sfi-.)OO
-EAN-).-OOb
5-.ell)}.)]d 1.1112i.-eO ).e?1--'elt95.t44
STO OEv
))')st)
tl'-5t t'i"s -.g2- t.2-1 ".ae- 12,ISt 1,2"1
vA-t-NCE
il12,SiOt
lt-.lli -1.iS- a-,lsi -e.lss t2,-Stt-S.e)1 i'-ei
N
t 1)S}{
c
(
c
(
c
t
s-J-1}ts)jo))o})s}tT)
CRtTE"]oh eecktt{
vAal4bLE OLP J.".E by C"h
.-.. u btvrh GsO•1-
LSL-LP'!a-o,IS.w 1)N -g.Lu.cA
v'- bueP] -• u ,. {t1ONs
VAFIatiLE
ruu E-Tl"c POPeLVSJh
CPIIc-t,)CA{,)cAltlC-rt]citljCAG)
TOTLL "bc5. ?j
cevE
IL.aa
/-:
di
.u•)a
iALUE L.',ll[L S'JM
ITj),eooo
:o}1,ncolgll./.100 ie",OUb:Odi,DCO !S-,D:Uilll.•)COIJo".oed
Mf--Jl.11e2
)T."ei-).eelLs.e)t>-.f,:L',Jy.e•/•oJb,-:TST.)ll
Sro eEv
is'-ie-
sl.)lg1).)ttIS.S.eS21h15to.--i e,-t) :.gi2
vA"tANCt
IT-.2iel
i}r.Tsi IT".eel li4.S)e Sll.il- -11.t)- 11,ebi )--"1
(
t
t
{
(
t
t
(
N
?)s)
)" il) 2o 10) )e) ss) tl}
Table 9. Description ofstatistical and
sediment types definecl by
compositional parameters.Kamada et al. (1973), concernlng
PltE"OU.N OF C":Jt{A
FrLt cHIJh4 (cStATteN )41E .SudFILL oklouo eN2oco
-----CRITtp:oN SPO"ÅíN-----vA-liLLt
FO- tsTILL
TYPtTYPErvpETvPETVpLTvPETVPETVPtTvpe
------vAR:A-LE vO-N ey
"OP"LATIUN
pSk.2L,
-1"- .,tLllrVPU SLLTiceut
1
iAi42sjs•-
le.b
:' t
-t J'IL-"r
;vT':) CHtJ:k".tiAY bLot-LNI
M-Y 1.
5:Zt V-T"be:TH TYPL}
iylb PAJE -
sc"L-ttJN ov su"pepvLAr;JNs ...........t...-.I,YL:bH N PH: :•C-Lt TtPl
---------------t--------------------d-v-LVF- L-ftE: bU-
9-s."leJ
1).eOo t.rpt" ?1.Yt,U et.-tu191,o"ai"4.k•a-ek.-eo 12.-C,u -o.ic,u
YL'AN
-'t?sv
1.leo 2Jco U.ltte L.etis ".Trs 1.)11 7.0-1 b.lbe.o."Q
srD tjf.v
2.ls"Z
o.121 n.tJ o.T2: 1.t91 O.-6tJ e.JJ- e.5)o e.1=J' aiall
vAHt-NCt
i.T"uT
O,1,)- e.D r•.el" S.-li e.-ls e.tll e.)dl o,els a.o4s
c
t
t
{
t
t
c
{
(
{
N
7)s}
:s) 1)lo}4e)"o))e)ST)•
") 1}
-----CelTER:ON eRO(EN
VAlttAeLL
FOR
TVFErypETVpETVpiTVPCTypeTNPtTVpEtypE
ENrlqE
-i--t---v-"IAeLE $e oovN ev TvpE-----d--
POPvL4T12N
.... Df SORITK" bEOI.•ENI
coDe
1
:-t-`' L-
]:A)b`)-
st"tcCLIFT: Tt"E
VALJi
PTIbN OFtEsT
LAti:-L
suovepLL-TloNs -------------------t----- '-tt------------------------------
Suy,
tib-.)TbO
2;.liO :.-lo :{.-so "1.eeu ibU,ilC ISS.-b" 7)b.}"e iS.ouo s.-7o
-EIN
i.E2rt
1.]t: 1.eec ;.egs 1.tc- 3.lbJ 2.Jbl 4n)e J.2S: i'T))
s!e etv
1.)b-e
e.11u e.o o-::e e,Si` i.1)2 e.e-- e,1-" e;ag- o,Olt
V-H:-NCF.
i.14e-
e.ot2 o.o e.ji2 O.2Y- 1.-2T e.Tll e."ez o.oe" e.oee
t
t
(
c
{
t
(
(
t
t
N
2J}1
il) l) 10) 4e) -e) s-) sl) -) 1)
CetTERItw SnyOcEN-----
v-"tAeLt. St DOvN SV TVPE
.... JE StE.NESS SEOIytNT
t-ca;ptt"H oF C:jV-ICIINT Tt;i
SUSPOPULAt[7kS
VAP:AeLE
FOH e"TlltE
TyPErVFETvpEh"eTvp5TvpErl;f'TYPtTVFE
P)F"L-rpt.
LOVE
l
lililtj
y1-->
vALdiE . -h..t 3U-
ITO.eotJo
Sb.-e L.S)O11.69u"b.S""tT.e;ojV.liLJJ,bb" t.VJ :.T-U
stit.
v.T2So
i.Cbju.el)i.ss",/.e-9;.eTeO,b21/,.s"]e.li)e.e7o
$Tr) oEv
c.S9tl
c.110e.uo.e)1o,}loe.)11e.igJC.19so,o3-e`li-
NARtiNCE
o.t}iS
e.elle.OO.12Jh.09bO.2ble.ellr,.o)le,oelo.eJ4
(
t
(
c
t
t
(
{
t
`
N
ljS)
:]) :) 10) sl) -e) sl} Sl] 4) v
ComputerAided Data Analysis for Geologic Problems 59
CRITEqtO- enokE-i
vAVIAULL O:ptr- By
GTVPt
.-.. bESCgTPTTON )F SubPePuLATTON5 "-iv:. L -iTIJ SF"t[Nl Tt;E
viHL-ILt CJJt vi..L -s:" :.L- YE-N srD oEv vAvl-NCE h
seR ENT]HE PIPJ.AT:JN ig".looj ,/.e412 1.lltl le.11V? c lss]
TvpETypLTYPtTypiTvPETYpETypETyeLTyPE
i
t4i-2•1
'
IA)3.
c,9"o c-:Jl,leJeb,-eo b,C --?eJie..c,O o,u4j.SLJ
c,ee: o.o ).Tr3 1.e:3 c,o e.cli L,-6J o,ol1,T)O•
c.21ie.Je.STe1.129c,oe,-ob2-1)so,o1.:-1
v,eb- o,o-1.2bg y,191 c.e t,,lsT T.-As o.e 1.12:
t
c
`
(
c
(
c
c
c
!}1 1)le}-s)-e))e})T} i} 1}
CRITEHION U"O-EN
V-nlAljLf DO-N tsv
s-TtPt
.... DL)C-IOT:.N :F S"bP:•PuLAT::r-S .-.-............. 5-Nv -iTt. StV:"ENT T.-v.
v--iA-LL C')OL v-' ]E' --,J'- )JY y[-iu ST) JLv v"IANCE N
Fos
TvPETvpETYp[TyPETvpLTVpErvPLTYP i.
rypt
EtsTtae P:PULirlJN
1
Lli42s,
)-)]`s
.iSOj,-OJ"
:-tb,-Ol TT."dO ei-.)bc.2tj.ICJisid.10u.,6Jl,"jJ bTd,}Od 19,bJO 1-j.bUO
b].)eeg
9t.19)Tl,-e"o?,-bO-T.9el)r,"bt62.Tl"11,lsc 4,"soli,TSc
)1.)1)-
).IOI O,Jlq,Tls1),-1?:e,-lh e,)1-n.oon 1,brl -.i12
lnbl.02bS
10,21e o,on-.c--p)T.))"11b,9bT lv.illi21.eoe ?.-ST --,i2e
c
c
c
c
c
c
t
{
c
`
2]J)
ls} 1)le)-l)-e}s-}oT) 4) v
C-ITEHISN e"o(tN
viklAdLt Jovt- ey
.... o ., )c lt rp'loN :t. buJpJpJL-rllNsSl SIJ II-T[:'VPt SLu:-U" tv".
vAHIAeLE
FOH
TvpETYPtTvpErvpttvpETvp[TvpETVPtrypc
[NTI"L "OPv--tl)N
.e,jl
1
lit-2B]
)-)tl
`'
J- .E .-lt. Suu
blJV..JOi
le.TuJ i).1)o et,ToD )et.1:u;Tt9,10o:.9-.!OO2-T",e:e ITI,IJv e.loe
LtF4N
2d,tR) i.11JIS,)Ou t,21v ti,i;e-),1?e2),eiv)o•)s>--.iT) -.oso
STD )fv
20,""St
1.dV- C, rJ
s,-2" ",)22i?•o)) 1.lll t.20S e.-Tl s.Tit
v"IANCE
-]).gSel
1.)IT e,o "L2Tl "e.elf1-S.}1- bs.i-s )t'-sl e.1.e" )2.eo}
t
t
t
c
t
`
(
`
(
`
N
21Sl
le) 11ie)-tl"e))e;ST} -) 2)
C-ITEelON eeecEN
vAnlASLE v3-N ey
CLTVPt
.-.. JE CLAf P"T SEuLM[-T
b.-tDT,-N JV 3UePePU-Arta-Sio TtnS
vAH;A-L[ tout v-L'J; L--tl ;Ud -E-N STO OEv vARTAhCE -
fos
rvp{TVPtTVPtTVPLTvPtTVPLTvpETvPLTvPt
ENTt-t )OPv.-TIU,.
:-t4irl
)JAJ"
.o}S.lbej
-.reo T,)Ol }!,4oo ,el.SdO T)).aou -bt.-JOlidd.VbO :O?,),1J -.-oo
IT.2)--
L,-LT 1-leo ).1.c J.ltSIS.S17li'1)?se.-o):o,e?) 2.-)e
:),Taso
1,i12J,O4,JO"-,}SJg,J"1).12Se.)51t,Ses1.--}
1-".Z9--
1.Sle e•ele.o.Sio.-"il.1)i;b'-tiVl,l)O 1.---i2.oaS
`
t
`
t
t
t
`
t
c
`
t3t)
IS) 1)ie)i-)ie))i)sl) i) 1)
CelTEfi:'J-, SAO-EN
vik:AeLE bo-h e-
.... bLSH",: vLt ait:1TtCt .tuiVit" 1c , t J r t J s
TPL
JF SJtiP:PuLaT:OhS
VAktAeLt CJpL v--Jt L -bL "vy MEAN iTO Jtv vAHIA-CE N
FOR ENT;ltE "oOLL-T:os lolib.vU:v -b. )tc,] )).J)1) :li2.)10J ( 11S}
TvpETVPtTVFLTVPtrvFiTvpLTvFtTtpF.TvpL
.i]-;"'
)-}b.,
2).1.JO 22.boo 11-.leo --S.-OJt--1.2o"i:es.lc"eool.so" )ho,ia. 11,)J"
1.).e21.-OttIJAt:10.l-Set.oos)I.)}T-r.SSL-b.IC•)-.Se;
).Ol: e.ole.i"-:}.-osie.T"1 -.1-"l2.tTi 1.-ST ".IS2
Y.O-T e.oieS."io2-).SeSliS.Srt --.o--l-S.-2- 2.T-T t-.)oo
t
{
`
t
c
c
t
t
(
IS) 1})o)-e)"ol51)5T) -1 t)
calTEHsc", aPJtEN
vAbtAeLf VtL' Dbdlt ey TvpE
.... DLS. ,l t;;l;- O- 5obPeO-LiTIONS .....-.....-..... OtPT" SEolYENT TT;E
vAHIAtiLt ceoL ViLL[ LlbU su" -E-H STC utv vAeli-cE -
se" i-T:HE PcPuL"IUK tn).edov )T. ITCt 1-.-leb 2T--11o2 ` 1)J)
TVPtTT-[TTpETYPErvptTVPtTvp5Tvp,.rvpe
TOTALLAStS - 2J)
t
1-t-lej1-j-`'
2b).oev t).vuu ly}.•)uoih)b.jLllbO-.JO-tSby.bceioiT.oan iS-.coo bi.;o"
:T.bel)x{,e"J-.See14.ees-c,:ec-4.121)}.S-1J-.bOO)1.eeo
2e.bll e.c)o.T2.21.7TL11.-lili,Ot) S.91" 1.1-1 o.e
Sl-.elO e.e-"."ss-T4.iesIJe.-Sl:te,-i4 J).TSI :.-sl c.o
t
`
t
`
`
t
`
t
t
IS) 11io}-e)4e)Jt))II il 1)
60
Table
Kaichiro YAMAMoTo
10. Univariate variance analysis of compositional parameters.
statistical and
Procedure TYPE CAG3Degrees of
Freedom
8
226
6
228
MDPHsoSKGSASI
CLMUDDEP
251.1
39.6
ILI 24.0
224.4
131.7
117.5
213.2
4.9
341,8
51.6
11.5
l2.6
399.8
228.2
166.4
389.6
17.2
t-kn snNo 5A,,O
#t.
1/.
't
.tt/-' i7•""
t
=-'
S/LrS:tSirJ-".l -EomENT, tecH ceoup e-sEo eN cL
s--o
SILTSraiJUn SEPS"EliT, E"H OReue e"sEo ok cL
snNo
SILT L esJy- sEoleEHT, tAcH onyoup eqsiv eH cL
S-NO
Sl{T L, iH.:J;,-.t SEOLHtNl, cncH pteup tAsto o" [L
SnNO ss as li'x
tl.
SILTS:SiJl,-- SEOIMENT, EgcH atovp e-sco o" tL
S-NO
S:LT L scSiJL"n SEU:"E-T. EAcH a-OuF epsED e- tL
sn-o
Sl:t SUt E• :Si"1Jit-P S[Dr"tNT, t"CH Gtou- egsEo ek cL
Fig. 30. Triangle contour diagrams, diagram of sand-silt-clay.
Al-1; Al-2; A2; AO;
t StLT Tsnl-Jwa sEolHE-T. Eat" oeeue lAsEo oN cL Åí:sti.lw- sEvmEe4T, Ea[- a-c:" eqsEo o- [L
showing concentration pattern of samples on a triangleProduced for each group defined by cluster analysis:
Bl; B2; C; Dl; D2group.
Computer Aided Data Analysis for Geologic Problems 61
SAND
Al-1
1-7
,N
e
BtB c
e-
c
s.
A2&
Al`2
AO
"
Ol
&
Fig. 31
Dl D2 SlLT. Configuration ofconcentration centers ofgroups defined by also Fig. 30.
CLAYcluster analysis, see
Evaluation of analytical results
The crosstabulation table between the two classification by KAMADAetal.(1973) and by the multivariate techniques demonstrated in this study are given in
Table 11. According to this table, the sum of the groups Dr and D2 in this study
corresponds to the group 3B by KAMADA et al., the sum of the groups Ai and A2to that of the groups 3 and 3A. But the groups containing coarser grains have no
clear correspondence between the two classifications.
This may influence the interprepation of the sedimentary environments of the
Chijiwa Bay. As shown in this case study, the cluster analysis and factor analysis can classify
the samples easily, definitely and in more detail, and the multivariate disciminant
analysis can reveal the characterisrics of classified groups and the relation among
them.
62
Table
Kaichiro YAMAMoTo
11. Crosstabulation table between two classification by Kamada et al. (1973) (across side) and by cluster analysis (down side). Numbers in each cell indicate count, row percent, column percent and total percent in descending
order.
CHIJt-A
f:LLSUtsFILE
----- CAG3""---
CAts3
A;
A2
-ts--
e2
c
Ol
bl
CH:CRAMER,S
KENDALL,SKENDALL,SGAMMA .SOMER,S D
C-RCUP ltASeLj ON CA, VAH ALVJTILts MAH.26, 19TS PAGECh:JikeA CCttLATILN OArL . Ftb. i, iS7bt LnWt.A.bAr 5tL)SMENT blZE DATAt"ITn TyPE) 9NICQ" cNtoeu " - :li su:, ;A;t :. :.-F :.:• D:cA L F v s S T A t; U L Ael iTep: O FsEbrM:N; ;v;E- - - - - - t - - - - -
t---{--""-"g•"-t---l.--"-t"-f-----"-""-----t-t PAGE 1OF TVPt CC,JNT i HD- PLT I ROw CoL PLTi TCTAL TOT PLTti i iA S ZA S 2b tb : 3A 1 Se 14 1S l ---------t--------i--------t--------1"-"---i--------t-------i---"---l--------l--------l
SeuARE vCONTIN6ENCY
1 Vi L•1 Ut ;t 30t i?: b: O; Oi t b.O l u,U l U.U I 1,9 l )b,6 1 )1.5 l Zl,i l O,O l O,O l 1 u.u l J.u 1 V.u i 2.; 1 7),U i 29,S l iU.S 1 e.O l O,O 1 t u,v t v,e t O.e i v." i 22,e : T.2 1 2.b 1 o,O t o,o l-t--------t---------1---.."-l-------l"------.t--------1-"-----l----"---:----------I
t et c; tt ls -l 3bt o: ol ol i U.O t -.C, l -.b t ?,4 i V.b t b3.) : O,O t O,O i O.[) :i u.v i c,.L/ l L'o,u ; u,1 l io.o t 6o.3 1 n.o 1 o.o t o,o ;
i u,e i L•o l b.9 l o.- l i.1 t i4.9 i O,O l O.O l o,o :.l......-..:.."....i.f-"-"j"---."i..----.t."...-.T.-."".I".--...l.-.-"d.I
s i4t "l il iit "t ut ot oi es l )s.e t b.e t .1.e i -2.) l O.b t O.O l O.Q t o,e l e,o l S V3.S t U.u i 10.u t ;2.V i V.V i O,V l O.O t O.O i O,O 1 s b.o i u.L t e." l 4.7 t e,o 1 o.u l o.v l o,o I o,o l .t----i-p.:-----.p-:--"----l---------t-------:----"--I--------:--"----1--------t
S lt lt 2t )IL Q: bt aL Ot eS i ).e ! ),L l le,b t s).e 1 u.v i 2S.u 1 o.o 1 o.o l o.o t ; e.T Lio:a,L t 2o.e 1 22.g 1 u.v : e,b 1 e,O t O,o I o,o t i c.- t v." l u.g 1 -.7 1 u.o t 2,1 l v,o : o.o t o,o t .t..""--L.-.-"":"-".--l-.-.....S..-.."1".".---l.-..".t"...---S-.-.""t i vl L,I bl 2)l O: Ol OI VI 21 t u.O 1 U.e i lb.1 i 7b.1 i u.O i O,O 1 O,U l O,O l 6,T t s u.o s v,e J bu.o i 4T.v t b,u 1 o,u I u,o t o.o tioo;o t i ".o s a," 1 2.l s s.- : ".o t o.v 1 o,o 1 e,o.: o,e t -1--------t--------t-----;-l"--"--1--------t--------l--------:--------i--------l
t Ol Cl Oi XL 6t ll 2ol Ol Ot 1 u.u ; v.o t u,u i 2.e l lb,1 l 2,b t ?T.e l O.o I O,O 1 i u.o ; e,v i v.v 1 2.1 i i).o 1 1,? l -y,i l o,o i o,o 1 i ",U S Q.O l {hU l ",- l L,6 i O." l 11.9 1 O,O l O.O : -s-----.---:-"--.---1------.--l--------1--------;--------:--------l--------1--------:
l "i Ll u; eE ei vl 2Sl -1 O: } v.o L o.c, l o.{J i e.e s v,o t o.o t es.2 t 1-,e 1 o.o 1 i v.o i u,u l b.e l v.u t L•.o l o,o l -o.- liuo,o t o,o 1 I u.o l u,L 1 u." i e.o : o,v : e.u ; g.s i 1.7 : o,o l .i.-..."-;..-.""t".--..1.--..".l-"--".1--".-.-l".-.."S"--...-1-.""..lCOLUMN i> 1 i" -e -O )b ST - 2 TCTAL b.- O.4 lt,3 i"." IT,O 24,'t ;-.3 "7 O.9s )3b.yg7u6 w;Th -6 VLGREtS vF FNEEVcM Si""IFiCAI"cE . b,O. O.PiT;3
CCtFFICILhT . O,e)-V2TAv ti . O.2v)bl S:GNiHCAr`CL . o,euouTAu C s O.IV54S SIGNIFtCANCE i O.eQooo.22elT
. o.2ue5b
S42S,O
42;1.9
26S;,L
2D -.s
30z2,e
3615,S
2'rix,)
235100,O
ST
t
1
Conclusion
In the case study mentioned above, the grain-size distribution data are analyzed
using the computer supported system. Namely, factors are extracted and theirmeanings are examined by R-mode factor analysis; samples are classified into several
groups by the Q-mode factor analysis and cluster analysis; of their significance, the
given classifications are tested by the discriminant analysis; and the results are also
shown in several types of diagrarns using the display programs. In the course ofthese studies the data file system is used forthestorage, retrieval, selection, transfor-
mation, and display of data.
The results are well concordant with those which are concluded from the ex-
Computer Aided Data Analysis for Geologic Problems 63
cellent studies of ordinal geologic techniques. Therefore, the validity of the data
analysis in geology is confirmed. Simultaneously the usefulness of the computersupported system presented here is shown in the smooth processing of thses studies.
The size ofgeologic data is generally larger than those used in this study, and the
data analysis for such data is hardly performed without the help of computer. The
computer supported system presented here can treat such a large amount of data,and is available for any geologists with a little knowledge on the computer.
In this study mainly numeric data are used, though large amounts of geologic
data are nonmetric, nominal, and nonquantitative. The computer supported systemcan treat only a part of such data. By the improvement of the system, however,nonquantitative geologic data will be treated in near future.
I have studied not only the computer supported system, but also the mathematical
and statistical techniques to treat geologic data. Some of the contributions on these
aspects will be reported in future.
References
ANDERBERG, M. R. (1973) : Clttster AnalJsisfor APPIications, Academic Press, New York and London,
359 p.Bz.AciinH, R. E., and REyMENT, R. A. (1971): Multivariate MorPhometrics, Academic Press, New
York and London, 412 p.BoNHAM-CARTER, G. F. (1967): FORTRAN IV program for Q-mode cluster analysis ofnonquanti- tative data using IBM 7090/7094 computers. ComPuter Contr. 17, Kansas Geol. Surv., 28 p.
CHEETHAM,A.H., and HAzEL,J.E. (1969): Binary (presence-absence) similarity coeMcients. Jeur. Palaeont., v. 43, n. 5, pp. 1130-1136.
CoLLyER, P. L., and MERRiAM, D. F. (1973): An application of cluster anaysis in mineral explora- tion. Jour. Intern. Assoc. Mathematical Geol., v. 5, n. 3, pp. 213-223.
DAvis,J. C. (1970): Information contained in sodiment-size analysis. Iour. Intem. Assec. Mathemati- cat Geol., v. 2, n. 2, pp. 105-112.
(1973) : Statistics and Data Anal7sis in Geology, John Wiely and Sons, Inc., New York, 550 p.
DixoN, W.J. (ed.) (1973): BMD Biomedical ComPuter Pregrams, Univ. Callfornia Press, Berkley, 773 p. (ed.) (1975): BMDP Manual, Univ. California Press, Berkley, 785pp.DRApEAu, G. (1973): Factor analysis: How it copes with complex geological problems. Jour. Intern. Assoc. Mathematical Geol., v. 5, n. 4, pp. 351-363.
FoLK, R. L., and WARD, W. C. (1957): Brozos River bar, a study in the significance of grain size parameter. Jeur. Sed, Petrol., v. 27, n. 1, pp. 3-26.
HARBAuGH,J.W., and BoNHAM-CARTER, G.F.(1970): ComPuterSimttlation in Geolag2, Wiley-Inter- science, New York, 575 p. , and MERRiAM, D. F. (1968) : ComPuter aPPIicatiens in StratigraPhic anat7sis, John Wiley and
Sons Inc., New York, 282 p.HATroru, 1, and MizuTANi, S. (1971): Computer simulation of fracturing oflayered rock. Engineer. Geol., v. 5, n. 5, pp. 253-269.
HAyAM!, I. and NAxANo, M. (l968) : A consideration on the numerical taxonomy in palaeontology, taking the Trigoniidae as example. Sci. Rep. Fac. Sci. K,7ushu Univ. (Geot.), v. 8, n. 4, pp. 191-
236 (in Japanese).HAzEL,J. E. (1970): Binary coeMcients and clustering in biostratigraphy. Geel. Soc. Amer. BulL,
64 Kaichiro YAMAMoTo
v. 81, pp. 3237-3252.HowARTH, R. J. (1973): Preliminary assessment of a non-linear mapping algorithm in a geological
context. Jour. Inter7i. Assoc. Mathematical Geol., v. 5, n. 1, pp. 39-57.
IMBRiE,J. and PuRDy, E. G. (1962): Classification of modern Bahamian carbonate sediments. Amer. Assec. Petrel. Geol. Mem. 1, pp. 253-272.
INMAN, D. L.,.and CHAMBERLAiN, T. K. (1955): Particle-size distribution in nearshore sediments. Soc. Eco. Paleont. Minerai. SPec. Pub. n. 3,, pp. 106-129.
INouE, N. (1970) : Submarine topography and grain-size distributions ofsediments in Tachibana Bay. In Report ofinvestigations on the hydrographic conditions and fish shoal distributions in Tachibana Bay. (Contr. Seikai Region Fish. Res. Lab., n. 256) pp. 23-32 (in Japanese).
KAMADA, Y. (1966): Coarse fraction studies of marine sediments along the Mogi Coast in Chljiwa Bay, Nagasaki Prefecture. Sci. Bull. Fac. Lib. Arts Educ. Nagasaki Univ., n. 17, pp. 45-54 (in
Japanese). , and HoRiGucHi, Y., INouE, M,, and WATANABE, H. (1973): Bottom sediments ofChijiwa Bay, Nagasaki Prefecture, with special reference to the distribution of muddy sediments. Bult. Fac. Lib. Arts Nagasaki Univ. Natural Sci., n. 24, pp. 61-79 (in Japanese).
'KLovAN,J. E. (1966): The use of factor analysis in determining depositional environments from grain-size distributions. Jour. Sed. Petrol., v. 36, n. 1, pp. 115-125.
KLovAN,J. E., and IMBRrE,J. (1971): An algorithm and FORTRAN-IV program for large-scale Q-mode factor analysis and calculation factor scores. Iour. Intern. Assoc. Mathematical Ceol., vol. 3,
n. I, pp. 61-77.KoMAR, P. D. (1973): Computer models of delta growth due to sediment input from rivers and
longshore transport. Geot. Soc. Amer. BuU., v. 84, n. 7, pp. 2217-2226.KRusKAL,J. B. (1964a): Multidimensional scaling by optimizing goodness of fit to a nonmetric
hypothesis. Psychometrica, v. 29, n. I,• pp. 1-26.
(1964b): Nonmetric multidimensional scaling: A numeric method. Psychometrtca, v. 29, n. 2, pp. 115-129.LuMsDEN, D. N. (1973): TRI: A FORTRAN subroutine to plot points on a triangular diagram.
Geol. Sec. Amer. Bull., v. 84r, n. 5, pp. 1765-1768.
McCAMMoN, R. B. (1966) : Principal component analysis and its application in large-scale correlation studies. Jrour, Geol., v. 74, n. 5, pt. 2, pp. 721-7S3.
MERRiAM,D.F. (ed.) (1969): •Symposium on computer applications in petroleum exploration. thmPuter Contr. 40, Kansas Gco1. Surv., 41 p.MrLLER, R. L., and KAHN,J. S. (1962): Statistical anatlsis in the geological seiences, John Wiley and
Sons, Inc., New York, 483 p.MizuTANi, S. (1974): Mathematical geology. Earth Science ("Chik2u Kagaku"), v. 28, n. 4, pp. 128-
149 (inJapanese).NiE, N. H., BENT, D. H., and HuLL, C. H. (l975): Statistical Packagefer the Social Sciences. McGraw-
Hill Book Company, New York, 675 p.
NrsHrwAKr, N., and YAMAMoTo, K. (1975) : Computer processing of geologic data specially on com- positional data ofsandstone. Jrour. GeograPh7 v. 84, n. 6, pp. Sl7-335.
NoBLE, D. C., and EBERLy, S. W. (1964): Discussion; A digital computer procedure for preparing beta diagrams. Amer. Jour. Sci., v. 262, n. 9, pp. 1124-1129.
PARKs,J. M. (l966): Cluster analysis applied to multivariate geologic problems. Jour. Geol., v. 74, n. 5, pt. 2, pp. 703-715.
(1969): Multivariate facies maps. ComPuter Contr. 40, Kansas Geol. Surv., pp. 6-l2.
(1970): FORTRAN IV program for Q-mode cluster analysis on distance function with printed dendrogram. Computer Contr. 46, Kansas Geol. Surv., 32 p.
RoBiNsoN, P. (1963): Preparation of beta diagrams in structural geology by a digital computer. Amer. .rour. Sci. v. 261, n. 10, pp. 913-928.
SA.MMoN,J. W.Jr. (1969): A nonlinear mapping for data structure analysis. IEEE Trans. Com-
Computer Aided Data Analy sis for Geologic Problems 65
Puters C-18, pp. 401=i09.SHEpARD, F. P. (1954): Nomenclature based on sand-silt-clay ratios. Jeur. Sed. Petrol., v. 24, n. 3,
pp. 151-158.SoKAL, R. R., and SNEATH, P. H. A. (1963): Princi les of Numerical Taxonomp, Freeman and Co., San Francisco, 359 p.SpENcER, D. W. (1963): The interpretation of grain size distribution curves of clastic sediments. Jour. Sed. Petrol., v. 33, n. 1, pp. 180-190,
TRAsK, P. D. (1932): Origin and environment ofsource sediments ofPetroleum. Gulf Pub. Co., Huston,
Texas.UJu6, H. (1973) : Sedimentation ofplanktonic foraminiferal shells in the Tsushima and Korea Straits betweenJapan and Korea. MicroPaleont., v. 19, n. 4, pp. 444--460.VtsTELrus, A. B. (1967): Studies in Mathematicat Ceolog7, Consultants Bureau, New York, 294 p.WARNER,J. (1969): FORTRAN IV programs for construction of Pi diagrams with the Univac 1108 computer. ComPuter Contr. 33, Kansas Gcol. Surv., 38p.WEBB, W. M., ancl BRiGGs, L. I. (1966): The use of principal component analysis to screen miner- alogical data. Jour. Geol., v. 74, n. 5, pt. 2, pp. 716-720.
YAMAMoTo, K. (1973a): The trend surface analysis by computer and its application. Jour. Geol. Soc. Iapan, v. 79, n. 5, pp. 349-362 (inJapanese).
(1973b) : Computer program for displaying samples in a three-dimensional factor space by using an X-Y plotter. K2oto Univ. Data Prec. Center Bull., v. 6, n. 8, pp. 300-302 (in Japanese).
YAMAMoTo, K. (1974a): Computer simulation for a sedimentary process. Proc. Kansai Branch, Geol. Soc. JaPan, no. 76, pp. 6-7 (Abstract).
(l974b): An application of factor analysis to sedimentary environments. Jour. Geol. Soc. JaPan, v. 80, n. 4, pp. 165-177 (in Japanese).
(1974c): Explanation of BMDP programs converted for FACOM 230-60/75 (1). Kptoto Univ. Data Proc. Center Bntl., v. 7, n. 4, pp. 186-193 (in Japanese).
(1975a): Explanation of BMDP programs converted for FACOM 230-60/75 (2). Kloto Univ. Data Proc. Center Bull., v. 8, n. 1. (in Japanese).
(1976): Preparing a contour map with a line-printer from irregularly spaced data. Jour. Inform. Proc. Soc. JaPan, v. 7, n. 5, pp. 448-450. (in .Japanese).
, and NAKAGAwA, Y. (1974): Time trand analysis of the Plio-Pieistocene sequence in the central part of the Kinki District, Japan. Mem. Fac. Sti., Kpato Univ., Ser. Geol. MimraL, v. 40, n. 2, pp. 45-65. , and NtsHiwAKi, N. (1975a) : FORTRAN program of preparing eontour maps for geologic use. Mem. Fac. Sci., Kyoto Univ., Ser. Geol. Mineral., v. 41, n. 1, pp. 1-34.
, and (1975b): Computer proccssing of oceanographic data, by using SPSS "Statistical Package for the Social Sciences". Marine Scr'. Month., v. 7, n. 6, pp. 53-59. (inJapa-
nese).YAMAMoTo, K., and NrsHiwAir, N. (1975c): Computer prograrns of trend surface analysis by poly- nomial approximation and moving average, and of contouringwith an X-Y plotter. K7oto Univ. Data Proc. Center Butl., v. 8, n. 3, pp. 121-134 (in Japanese).
, and (1975d): Automatic analysis of the geologic structure from dip-strike data. Computers and Geosciences, v. 1, n. 4, pp. 309-323.