Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An evaluation of regional stream sediment data by advanced statistical procedures Matysek, Paul Frank 1985

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1985_A6_7 M38.pdf [ 4.42MB ]
Metadata
JSON: 831-1.0052662.json
JSON-LD: 831-1.0052662-ld.json
RDF/XML (Pretty): 831-1.0052662-rdf.xml
RDF/JSON: 831-1.0052662-rdf.json
Turtle: 831-1.0052662-turtle.txt
N-Triples: 831-1.0052662-rdf-ntriples.txt
Original Record: 831-1.0052662-source.json
Full Text
831-1.0052662-fulltext.txt
Citation
831-1.0052662.ris

Full Text

EVALUATION OF REGIONAL STREAM SEDIMENT DATA BY ADVANCED STATISTICAL PROCEDURES by PAUL FRANK MATYSEK B. Sc., U n i v e r s i t y Of T o r o n t o , 1980. A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS' FOR THE DEGREE OF MASTER OF SCIENCE i n THE FACULTY OF GRADUATE STUDIES Department Of G e o l o g i c a l S c i e n c e s We a c c e p t t h i s t h e s i s as c o n f o r m i n g t o the ^ r e q u i r e d s t a n d a r d THE UNIVERSITY OF BRITISH COLUMBIA September 1985 © P a u l Frank Matysek, 1985 In present ing this thesis in partial fulf i lment of the requ i rements for an advanced deg ree at the University of British C o l u m b i a , I agree that the Library shall make it freely available for reference and study. I further agree that permiss ion for extens ive c o p y i n g of this thesis for scholarly pu rpo se s may be granted by the h e a d of my depar tment or by his or her representatives. It is u n d e r s t o o d that c o p y i n g o r publ icat ion of this thesis for financial gain shall not b e a l l owed wi thout my written permiss ion. Depar tment of T h e University of British C o l u m b i a 1956 M a i n Ma l l Vancouver , C a n a d a V 6 T 1Y3 Date O c r r ~ L O ~ &5 Abstract This study was directed towards the development of rigorous, systematic, computer-assisted s t a t i s t i c a l procedures for the interpretation of quantitative and q u a l i t a t i v e data commonly encountered in p r a c t i c a l exploration-oriented surveys. A suite of data analysis tools were developed to evaluate the q u a l i t y of geochemical data sets, to investigate the value and u t i l i z a t i o n of cat e g o r i c a l f i e l d data, and to recognize and rank anomalous samples. Data obtained from regional stream sediment surveys as undertaken by the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources in southern B r i t i s h Columbia were examined as a case h i s t o r y . A procedure based on a s t a t i s t i c a l analysis of f i e l d - s i t e duplicates was developed to evaluate the q u a l i t y of regional geochemical s i l t data. The technique determines: (1) whether differences in metal concentrations between sample s i t e s r e f l e c t a r e a l trend related to geological and geochemical features and not merely a consequence of sampling and a n a l y t i c a l error, and (2) absolute pr e c i s i o n estimates at any p a r t i c u l a r accumulation across a metal's concentration range. Results for metals Zn, Cu, Ni, Co, Fe and Mn indicated that combined v a r i a b i l i t y due to l o c a l and procedural error averged less than 5% of the t o t a l error and that p r e c i s i o n estimates at the 95th percentile concentration value averaged less than 6.0%. Results presented indicate duplicates are more in accord with s p l i t s of individual •^3_Mpbes^t-analytica 1 duplicates) rather than separate f i e l d - s i t e i i i d u p licates. This type of systematic approach provides a basis for i n t e r p r e t i n g geochemical trends within the survey area, while simultaneously allowing evaluation of the method of sampling and laboratory a n a l y s i s . A procedure u t i l i z i n g Duncan's Multiple Range Test examined the re l a t i o n s h i p s between metal concentrations and c l a s s - i n t e r v a l and c a t e g o r i c a l observations of the drainage catchment, sample s i t e and sediment sample. Results show that, many f i e l d observations can be systematically related to metal content of drainage sediments. Some elements are more susceptible than others to environmental factors and some factors influence few or many elements. For example, in sediments derived from granites there are s i g n i f i c a n t relationships between bank type and concentration of 8 elements (Zn, Cu, Ni, Pb, Co, Fe, Mn and Hg). In contrast, the texture of these sediments, using estimates of fines contents as an index, did not s i g n i f i c a n t l y a f f e c t the concentration of any of the elements studied. In general, r e s u l t s indicate that groups of environmental factors acting c o l l e c t i v e l y are more important than any single factor in determining background metal contents of drainage sediments. A procedure u t i l i z i n g both a graphical and mulitple regression approach was developed to i d e n t i f y and characterize anomalous samples. The procedure determines multivariate models based on background metal values which are used to describe very general . rgfiDcneTnical r e l a t i o n s of no interest for prospecting purposes. iv "These models are then applied to sample subsets selected on the basis of factor/s known to strongly influence geochemical r e s u l t s . Individual samples are characterized after comparisons with relevant determined threshold l e v e l s and background multielemenmodels. One hundred and f i f t e e n anomalous samples for zinc from seven provenance groups draining 1259 sample s i t e s were i d e n t i f i e d and characterized by t h i s procedure. Forty three of these samples had zinc concentrations greater than i t s calculated provenance threshold, while 72 of these anomalous samples were i d e n t i f i e d s o l e l y because t h e i r individual metal associations were s i g n i f i c a n t l y d i f f e r e n t than their provenance multivariate background model. The method provides a means to reduce the e f f e c t s of background variati o n s while simultaneously i d e n t i f y i n g and characterizing anomalous samples. The data analysis tools described here allow extraction of useful information from regional geochemical data, and as a result provide and e f f e c t i v e means of defining problems of geological interest that warrant further investigation. V T a b l e of C o n t e n t s A b s t r a c t i i L i s t of T a b l e s v i L i s t of F i g u r e s v i i Acknowledgement v i i i C h a p ter I INTRODUCTION 1 Chapter I I STATISTICAL EVALUATION OF DUPLICATE SAMPLES, REGIONAL SEDIMENT SURVEYS (MAP-AREA 92H, 921 AND 92J) . . 7 Chapter I I I STATISTICAL EVALUATION OF THE SIGNIFICANCE OF CATEGORICAL FIELD PARAMETERS IN INTERPRETATION OF REGIONAL GEOCHEMICAL SEDIMENT DATA (MAP-AREA 82F) 30 Chapter IV RECOGNITION AND RANKING PROCEDURE FOR MULTI-ELEMENT REGIONAL STREAM SEDIMENT SURVEYS (MAP-AREA 82F) 63 Chapter V CONCLUSIONS 88 BIBLIOGRAPHY 92 v i L i s t of Tables 2.1 Analysis of Variance 18 2.2 Test for systematic bias 18 2.3 Analysis of Variance Results for 132 duplicate pairs of stream sediments 22 2.4 Relative precision estimates, Thompson-Howarth method 23 3.1 Field-recorded categorical variables 39 3.2 Spatial density of stream sediment samples 39 3.3 5% l e v e l Duncan's Multipe Range Test of three unequally replicated means and c a l c u l a t i o n of D-ratio 45 3.4 D-Ratio calculated for categorical f i e l d variables from sediments draining g r a n i t i c provenances 55 3.5 D-Ratio calculated for categorical f i e l d variables from sediments draining q u a r t z i t i c provenances 55 3.6 S p e c i f i c categories of individual c a t e g o r i c a l f i e l d v ariables with highest means, Gr a n i t i c provenance ..57 3.7 S p e c i f i c categories of individual c a t e g o r i c a l f i e l d v ariables with highest means, Quartztic provenance .57 3.8 S p e c i f i c categories of individual c a t e g o r i c a l f i e l d variables with lowest means, Gr a n i t i c provenance ...58 3.9 S p e c i f i c categories of individual c a t e g o r i c a l f i e l d v ariables with lowest means, Quartztic provenance ..58 3.10 Nonparametric cor r e l a t i o n matrix, G r a n i t i c provenace 59 4.1 Examples of multivariate regression background models for zinc 77 4.2 Summary s t a t i s t i c s for multivariate background zinc models 78 4.3 Part of a table l i s t i n g anomalous samples in order of decreased ranking code 80 v i i L i s t of Figures 1.1 Location map of study area 82F, 92H, 921 amd 92J 4 2.1 Location map of study area 92H, 921 and 92J 13 2.2 Absolute differences of paired data versus corresponding mean values of pairs, copper in stream sediments 24 2.3 Linear model of average error as a function of concentration, copper in stream sediments 25 3.1 Location map of study area 82F 35 3.2 Simplified geology of the study area 82F 37 3.3 Duncan's Multiple Range Test for the influence of content of stream sediments associated with a g r a n i t i c G r a n i t i c provenance 46 3.4 Duncan's Multiple Range Test for the influence of sediment colour of stream sediments associated with a Gran i t i c provenance 47 4.1 Location map of study area 82F 68 4.2 Sample ranking in r e l a t i o n to f i e l d s on a plo t of observed value versus value calculated from a multivariate model 74 4.3 Plot of an area of anomalous samples from study area 82F 81 4.4 Observed versus calculated zinc values for provenance group A r g i l l i t e ; calculated values based on a model determined from a l l samples with background zinc values 83 4.5 Observed versus calculated zinc values for provenance group A r g i l l i t e ; calculated values based on a model determined from those samples with zinc background values that are also not anomalous in any other element 84 v i i i Ac knowledqement Special thanks are extended to Drs. A.J, S i n c l a i r and W.K. Fletcher for providing guidance and encouragement throughout th i s study. The author would l i k e to thank the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources for providing f i n a c i a l support. I am especially g r a t e f u l to Asger Bentzen for his valuable assistance with the computer work. A special mention goes to Kaarina T a l v i l a for her personal encouragement. 1 2 The completion of large multi-element geochemical surveys leads to the accumulation of enormous amounts of data. It i s probable that only the most obvious features of the data can be recognized by visu a l and manual int e r p r e t a t i o n of such complex and voluminous data and that subtle though s i g n i f i c a n t information has remained undetected. Consequently, a premium i s placed upon computer processing by mathematical and s t a t i s t i c a l procedures that provide a more d e t a i l e d interpretation of geochemical data, s p e c i f i c a l l y , those techniques which can evaluate q u a l i t y of geochemical data sets, i d e n t i f y regional trends in the data, and more importantly, d i s t i n g u i s h variations related to bedrock mineralization from secondary environment influences. However, i t must be stressed that results from these techniques must be reviewed c r i t i c a l l y in terms of t h e i r geochemical implications before acceptance and incorporation into an int e r p r e t a t i o n . The studies described here are concerned with the development of computer-based s t a t i s t i c a l procedures for extracting useful information from regional geochemical data, and as a r e s u l t , provide an e f f e c t i v e means of defining problems of geological interest that warrant further investigation. More formally, t h i s study was directed towards the development of rigorous, systematic, computer-assisted, interpretative 3 techniques for the evaluation both of quantitative and c a t e g o r i c a l data of the types commonly encountered in p r a c t i c a l exploration-oriented surveys. A suite of data analysis tools with appropriate presentation techniques were developed to: (a) evalate the qua l i t y of geochemical survey data (b) investigate the value and u t i l i z a t i o n of categorical f i e l d data (c) recognize and rank anomalous samples Data obtained from regional geochemical surveys undertaken by the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources (1981) and the Department of Energy, Mines and Resources, Geological Survey of Canada were (1977) were examined. These data consist of analyses of stream sediments and water samples c o l l e c t e d at an average density of one per 17 square kilometers over NTS mapsheets 82F, 92H, 921 and 92J (Figure 1.1). Twelve hundred samples were c o l l e c t e d on average per mapsheet. Semi-quantitative and q u a l i t a t i v e c l a s s interval and categorical observations on c h a r a c t e r i s t i c s of the drainage catchement, sample s i t e and sediment sample were also recorded. Samples were f i e l d dried and the minus 80 mesh (177 microns) fractio n was retained for subsequent analyses. The samples were analyzed for Zn, Cu, Pb, Ni, Co, Ag, Mn, Fe, Mo, As and Sb by atomic absorption. Tungsten was determined c o l o r i m e t r i c a l l y . F i e l d s i t e duplicates were reported to be ""talrem a t a density of one per block of 20 samples and could not R E G I O N A L G E O C H E M I C A L S U R V E Y S PRE 1980 RELEASES 1980 RELEASE RELEASE 1981 SURVEY F i g . 1,1 Location map of study area 82F, 92H, 921 and 92J. and index jap of regional geochemical surveys a v a i l a b l e i n B r i t i s h Columbia (McMillan, 1982) 5 "be distinguished from other s i l t samples by the contracted commercial laboratory. These data are a v a i l a b l e p u b l i c a l l y on magnetic tape from the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources, Geological Branch, Mineral Resources D i v i s i o n , V i c t o r i a . Due to the voluminous nature of thi s type of data, computer usage i s e s s e n t i a l i f the data are to be compiled in a timely and e f f i c i e n t manner. As a result, each procedure investigated was computer-based, and incorporated a variety of r e l a t i v e l y simple, commonly used univariate and multivariate techniques. This thesis i s organized in the form of a series of papers. Three i n t e r p r e t a t i v e procedures are examined and each one forms a subsequent chapter in t h i s thesis. Individual procedures have been reviewed previously and preliminary versions have been published (Matysek et a l , 1981; Matysek et a l , 1982; Matysek et a l , 1983; and Matysek and S i n c l a i r , 1984). A b r i e f description of each follows. An improved method of examining q u a l i t y of regional geochemical survey data i s presented in Chapter Two. The method i s based on f i e l d - s i t e duplicates and incorporates a bias test, followed by an analysis of variance technique and the Thompson and Howarth (1976) approach to quantifying p r e c i s i o n . This type of systematic procedure provides a basis for interpreting geochemical trends within the sampling area, while simultaneously allowing evaluation of the method of sampling and iaboTatory analysis. 6 A procedure u t i l i z i n g Duncan's multiple range test to evaluate the s i g n i f i c a n c e of commonly c o l l e c t e d categorical f i e l d observations i s the subject of Chapter Three. The procedure examines the relationships between metal concentrations and ca t e g o r i c a l observations of the drainage catchment, sample s i t e and sediment sample. Chapter Four presents a multivariate interpretative technique that combines both a graphical and s t a t i s t i c a l approach to the i d e n t i f i c a t i o n , characterization and s i g n i f i c a n c e of anomalous samples. The procedure involves developing multivariate models based on background metal values and applying them to sample subsets selected on the basis of factors known to strongly influence geochemical r e s u l t s . General conclusions are summarized in Chapter Five. 7 I I . STATISTICAL EVALUATION OF DUPLICATE SAMPLES, REGIONAL  SEDIMENT SURVEYS (MAP-AREA 92H, 921 AND 9 2 J ) , BRITISH COLUMBIA 8 ABSTRACT A systematic computer-assited method of e v a l u a t i n g the q u a l i t y of geochemical survey data based on f i e l d s i t e d u p l i c a t e s i s d e s c r i b e d . The procedure in c o r p o r a t e s a bias t e s t , an a n a l y s i s of variance technique and the Thompson and Howarth (1976) approach to q u a n t i f y i n g p r e c i s i o n . The technique combines both a g r a p h i c a l and s t a t i s t i c a l approach to determine whether d i f f e r e n c e s i n metal concentrations between sample s i t e s r e f l e c t a r e a l trend r e l a t e d to g e o l o g i c a l and geochemical features and not merely a consequence of sampling and a n a l y t i c a l e r r o r , and e s t i m a t i o n of absolute p r e c i s i o n as a f u n c t i o n of metal c o n c e n t r a t i o n . D u p l i c a t e data obtained from r e g i o n a l sediment surveys undertaken by the B r i t i s h Columbia M i n i s t r y of Energy, Mines and Petroleum Resources i n southern B r i t i s h Columbia were evaluated by the proposed method. A n a l y s i s of 132 p a i r s of f i e l d - s i t e d u p l i c a t e s i l t samples i n d i c a t e d t h a t : (1) E r r o r s f o r metals Zn, Cu, N i , Co, Mn and Fe r e s u l t i n g from sampling, sample pr e p a r a t i o n and a n a l y s i s were small (<5% w i t h i n - s i t e variance component) compared to the r e g i o n a l geochemical v a r i a t i o n and, (2) P r e c i s i o n estimates c a l c u l a t e d by the Thompson and Howath method f o r the s i x metals examined averaged l e s s than 6.5% at the 90th concentration value p e r c e n t i l e , 6.0% at the 95th p e r c e n t i l e and 5.5% at the 99th p e r c e n t i l e . 9 Results presented here are more in accord with s p l i t s of in d i v i d u a l samples ( a n a l y t i c a l duplicates) rather than analyses of two separate f i e l d samples. If t h i s i s the case, the loss of information on the v a r i a b i l i t y associated with the sampling procedure and l o c a l environment s e r i o u s l y hampers the v a l i d i t y of these duplicate data in the recognition of s i g n i f i c a n t geochemical trends. Since there appears to be some question over the type of duplicate analyzed, every e f f o r t should be made to ensure that f i e l d s i t e duplicates are c o l l e c t e d in future surveys. This type of systematic procedure u t i l i z i n g f i e l d - s i t e duplicates provides a basis for inte r p r e t i n g geochemical trends within the sampling area, while simultaneously allowing evaluation of the adequacy of the method of sampling and laboratory analysis. 10 I N T R O D U C T I O N The a b i l i t y to discriminate real trends related to geological and geochemical causes from those that result from spurious factors such as sampling and a n a l y t i c a l errors i s of considerable importance in the success of geochemical data int e r p r e t a t i o n . An estimate of r e p r o d u c i b i l i t y (precision) allows us to quantify the amount of variation due to sampling and analysis, and i t i s an integral part of the evaluation of geochemical data. Quality control procedures including duplicate samples should be conducted prior to carrying out any detailed data i n t e r p r e t a t i o n . U t i l i z a t i o n of paired duplicates provide information on precision and aid in assessing: 1. Metal v a r i a b i l i t y within and between sample s i t e s . 2. A n a l y t i c a l precision over the range of values encountered. 3. D e f i n i t i o n of r e a l i s t i c detection l i m i t s . 4. Sample homogeneity by comparing p r e c i s i o n obtained on standards versus precision for paired duplicates at approximately the same concentration. As part of our continuing study of rapid thorough evaluation procedures for multielement stream sediment data (for example Matysek et a l . , 1981, 1982 and 1983) we have designed a systematic, computer-oriented method of evaluating the quality of geochemical data based on f i e l d - s i t e duplicates incorporating a bias t e s t , an analysis of variance technique and the Thompson and Howarth (1976) approach to quantifying p r e c i s i o n . This detailed procedure i s applicable to the type and 11 quality of data incorporated in various regional programs undertaken by the B r i t i s h Columia Ministry of Energy, Mines and Petroleum Resources but can be adapted e a s i l y for other data types. GENERAL METHODOLOGY In b r i e f , our general approach to evaluation of the qu a l i t y of geochemical data sets involves the following steps: 1. Extraction of at least 50 independent, f i e l d -s i t e duplicate pairs from a geochemical data set for use in subsequent data analysis. 2. Determination of the degree of systematic bias between duplicate p a i r s , based upon the number of cases in which the f i r s t observation i s greater than the dup l i c a t e . 3. Evaluation of the duplicated geochemical data set in terms of metal v a r i a b i l i t y at the regional (between-sample site) and at the l o c a l , sampling and a n a l y t i c a l levels (within-sample si t e ) by a two factor a n a l y s i s of variance technique. 4. Quantification of the within-sample s i t e v a r i a b i l i t y by estimating the prec i s i o n u t i l i z i n g the Thompson and Howarth (1976) method. DATA Data obtained from three recent regional geochemical surveys undertaken by the B r i t i s h Columbia Ministry of Mines, Energy and Petroleum Resources (1981) were evaluated by the proposed method. The base data consist of analyses of stream sediments and water samples c o l l e c t e d at an average density of one sample per 17.3 square kilometers over NTS mapsheet areas 12 92H, 921 and 92J (Figure 2.1). S i l t samples were f i e l d dried and the minus 80 mesh (177 microns) f r a c t i o n retained for subsequent analyses. The samples were analyzed for zinc, copper, lead, n i c k e l , c o b a l t , s i l v e r , manganese, iron, molybdenum, arsenic and antimony by atomic absorption. Tungsten was determined c o l o r i m e t r i c a l l y , uranium in water samples was determined by a fl u o r i m e t r i c method and f l u o r i d e in stream waters was determined using a s p e c i f i c ion electrode. F i e l d s i t e duplicates were also c o l l e c t e d to provide information on a n a l y t i c a l precision over a range of concentrations and to give some impression of sample representivity or geologic v a r i a t i o n . Altogether, 132 pairs of duplicate f i e l d s i t e samples were extracted from the data set. Individual duplicate samples were recorded as f i r s t of f i e l d s i t e duplicate and second of f i e l d s i t e duplicate in the p u b l i c a l l y a v a i l a b l e general information guide describing the data base ( B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources, 1981). Duplicates were reported to have been c o l l e c t e d at a density of one per block of 20 samples. Duplicate p a i r s within individual blocks were analyzed in batches and could not be distinguished from other s i l t samples by the analyst. These data are available p u b l i c a l l y on magnetic tape from the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources, Geological Branch, Mineral Resources D i v i s i o n , V i c t o r i a . ^^Sarnpling and a n a l y t i c a l error associated with zinc, copper, F i g . 2.1 L o c a t i o n map of study area 92H, 921, 92J, and index map o f r e g i o n a l geochemical surveys a v a i l a b l e i n B r i t i s h Columbia ( M c M i l l a n , 1982). 14 n i c k e l , cobalt, manganese and iron were investigated by th i s study. This p a r t i c u l a r suite of metals was selected on the following basis: 1. Their d i s t r i b u t i o n s approximated the Gaussian curve. 2. Majority of reported values were above th e i r published detection l i m i t s . ASSUMPTIONS The assumption that a n a l y t i c a l and sampling v a r i a t i o n c l o s e l y follows the Gaussian d i s t r i b u t i o n underlies most of t h i s discussion, and i t i s neccessary to examine the assumption and to be aware of situations where i t may not apply. The mathematical conception of the Gaussian curve i s that i t results from the combination of a large number of small independent errors. In practice, these small errors can be regarded as r e s u l t i n g from variations in manipulative operations such as sampling, sub-sampling, weighing, d i s s o l u t i o n , etc. Despite the fact that some of these contributions may not be small, they are usually independent and the r e s u l t i n g d i s t r i b u t i o n i s not usually distinguishable from a random sample drawn from a Gaussian population. Non-Gaussian forms are apparent in practice when two p a r t i c u l a r conditions are met: one or more contributing factors i s no longer r e l a t i v e l y small ( i . e . , i t makes a substantial contribution to the t o t a l v a r i a t i o n ) ; and i t s p a r t i c u l a r frequency d i s t r i b u t i o n i s non-Gaussian. The u n c r i t i c a l use of 15 s t a t i s t i c s calculated on metals characterized by non-Gaussian forms i s l i k e l y to lead to erroneous conclusions, e s p e c i a l l y when the v a r i a b i l i t y of measurement i s high r e l a t i v e to the actual metal abundance. Non-Gaussian metal d i s t r i b u t i o n s are recognized when the following conditions a r i s e : 1. Metal abundances are dependent on rare grains 2 . Precision of the method of analysis i s poor 3. Concentration l e v e l s are near the detection l i m i t 4. Data contains o u t l i e r s BIAS TEST A c h a r a c t e r i s t i c of numerical measurement i s inconsistency in repeated measurements of the same quantity. Two types of error contribute to the u n r e l i a b i l i t y of a measurement: random errors a r i s i n g from the variations inherent in any sampling or measurement process and non-random errors causing systematic negative or p o s i t i v e deviations from the true r e s u l t s . A bias test was u t i l i z e d to assess the degree of systematic non-random error between the determined metal contents of the f i r s t sample and i t s duplicate. The test i s based upon the number of instances in which the f i r s t observation i s greater than the duplicate. If there i s no bias t h i s number (m) should be close to one half the t o t a l number of pairs (n), and the frequency d i s t r i b u t i o n of possible results should correspond -Trvth ^ successive terms of the binomial expansion of: 16 [ ( 1/2 ) + ( 1/2 ) T The p r o b a b i l i t y of obtaining a p a r t i c u l a r deviation from n/2 or greater can be calculated exactly. If the number of duplicate p a i r s exceeds 50, i t ' s frequency d i s t r i b u t i o n approximates that of the normal d i s t r i b u t o n , having mean and a standard deviation as n/2 and ny/2 respectively. The observed incidence (m) i s converted into the standardized normal deviate. The p r o b a b i l i t y of obtaining a re s u l t of (m) or greater can be obtained from the usual tables for the areas in the t a i l s of the normal d i s t r i b u t i o n . ANALYSIS OF VARIANCE The second stage in our procedure i s to evaluate the geochemical data set in terms of v a r i a b i l i t y at regional (between-sample s i t e s ) and at l o c a l l e v e l s (within-sample s i t e ) . Only i f a s i g n i f i c a n t proportion of the data v a r i a b i l i t y i s at the regional l e v e l can one be confident that differences in metal concentrations between-sample s i t e s r e f l e c t a real trend related to geological and geochemical features and not merely a consequence of sampling or a n a l y t i c a l error. Significance of metal v a r i a b i l i t y within-sample s i t e s due to sampling and a n a l y t i c a l errors versus dispersion between-sample s i t e s can be determined by standard analysis of variance techniques. For our purposes, the si g n i f i c a n c e of the various "STttxree- of v a r i a t i o n can be determined from logarithmic values 17 of duplicate samples using a two factor analysis of variance. Theory and assumptions inherent in t h i s method are described by Krumbein and G r a y b i l l (1965) and Koch and Link (1971). The formal design for the analysis of variance i s given in Table 2.1 . The s i g n i f i c a n c e of metal v a r i a b i l i t y between-sample s i t e s (geochemical variation) and metal v a r i a b i l i t y within-sample s i t e s (sampling and a n a l y t i c a l errors) i s determined from the F-s t a t i s t i c . The value of F which i s s i g n i f i c a n t can be obtained from s t a t i s t i c a l tables and is a function of the number of s i t e s at which duplicates are col l e c t e d , the number of duplicates c o l l e c t e d at each s i t e and the sign i f i c a n c e l e v e l selected for the i n v e s t i g a t i o n . Relative variance components have also been calculated as described by Garrett(1969, 1973) and correspond to the average percentage of v a r i a b i l i t y explained by each source at a sample s i t e . THOMPSON AND HOWARTH PRECISION METHOD The f i n a l stage in our evaluation of the qua l i t y of geochemical data sets i s to quantify the amount of va r i a t i o n due to sampling and a n a l y t i c a l error. This v a r i a t i o n can be expressed in terms of precision which,in geochemical pr a c t i c e , i s s p e c i f i e d as the percent r e l a t i v e v a r i a t i o n at the two standard deviation (95%) confidence l e v e l : Pc = 2Sc/c * 100% [1] 18 Table 2 . 1 Analysis of Variance SOURCE OF VARIATION O E G R E E OF F R E E D O M SUM OF S Q U A R E S M E A N SQUARE (tampling, tub-sampling, analysts / - 1 i / - 1 / f, -S,'IS,' Geochemical variation m - 1 / ' 5 , " • — t l < 7 - ; i ' m — \j f, -S.'IS,' Residual V - UCm - 1) a V ' ( / - l l t m - 1) »•/' Total N - 1 rV - 1 i / ' - ; ) • Mjj - re*ult lor the rth replicate from the/th l i t e ; / - 1, 2 / th retult (J i< usually 21; /' " 1. 2 /nth reiult; Af - /m " total number of remits; j?• • mean o l nh replicate group; ?• - mean of / th tite group; and A*« overall mean. Table 2.2 Test for systematic bias P R O B A B I L I T Y OF O B T A I N I N G NO. O F POSIT IVE D I F F E R E N C E S per cent 36 30 22 19 17 <1 <1 N O P O S I T I V E N E G A T I V E M E T A L D I F F E R E N C E S D I F F E R E N C E S D I F F E R E N C E S No. Iper cent) Nickel 43 (32.6) 46 4 3 Arsenic 49 (37.1) 44 39 Cobalt 56 142.4) 42 34 Copper 31 (28.51 55 46 Manganese 14 (10.6) 64 54 Iron 31 (23.5) 64 37 Zinc 23 (17.4) 68 41 19 where "("PC) i s the p r e c i s i o n in percent at concentration (c) and (Sc) i s an estimate of the a n a l y t i c a l standard deviation (crc) at concentration ( c ) . Application of analysis of variance techniques can only determine an average precison value for a range of concentrations. In actual fact, i t has been shown (Thompson and Howarth, 1976) that where there i s a wide range of concentrations in a set of samples both the absolute and r e l a t i v e errors in a n a l y t i c a l determinations can vary across the range. To deal with t h i s complexity, a l t e r n a t i v e ways of estimating precision using randomly selected duplicates, have been considered in d e t a i l by Thompson and Howarth (1973,1976,1978). B r i e f l y , their preferred method involves d i v i d i n g 50 or more duplicate samples into groups with narrow concentration ranges, and employing the median of absolute differences between pairs of duplicate analyses ( x, , x L ) as an estimator of the standard d e v i a t i o n ^ . The group mean value of a l l the mean average values [ (x, + x^ ) / 2 ] i s used as an estimator of the average concentration. If this procedure i s repeated for a number of successive narrow concentration ranges a set of corresponding mean concentration and standard deviation estimates i s obtained. The rel a t i o n s h i p between them can be found by simple l i n e a r regression. From the expression: S c = S0 + K c [2] 20 through substitution of [1] above, Pc i s given by: Pc = 200 ( So / c + K ) [3] where (S© ) i s the standard d e v i a t i o n at zero concentration and (K) i s a constant. This linear function has been determined in many p r a c t i c a l cases to be a s a t i s f a c t o r y model for the expression of the v a r i a t i o n . The following rapid procedure i s suggessted for estimation of precision from a minimum of 50 p a i r s of duplicate samples (Thompson and Howarth, 1976): 1. From the duplicate analyses obtain a l i s t of the means and absolute d i f f e r e n c e s . 2. Arrange the l i s t in increasing order of concentration means. 3. From the f i r s t 11- r e s u l t s obtain the mean concentration and median d i f f e r e n c e from that group. 4. Repeat step (3) for each successive group of 11 r e s u l t s , ignoring any remainder le s s than 11. 5. Calculate or obtain g r a p h i c a l l y , the l i n e a r regression of the median d i f f e r e n c e s on the means and multiply the intercept and c o e f f i c i e n t by 1.048 (factor derived from the properties of the half-normal d i s t r i b u t i o n ) to obtain (So) and (K) respectively. In order to assess the s i g n i f i c a n c e of the precision parameters obtained from the linear regression, the calculated slope and intercept were evaluated i n d i v i d u a l l y by a t - t e s t . Significance of the r e s u l t i n g regression was also determined by an analysis of variance. 21 RESULTS Table 2.2 i l l u s t r a t e s the results obtained from the bias t e s t . From t h i s table we observe: 1. Incidence of p o s i t i v e differences in metal content between duplicate pairs i s greater than the number of negative differences for each of the metals analyzed. 2. Metals n i c k e l , cobalt, copper and manganese exhibit s t a t i s t i c a l l y s i g n i f i c a n t , but minor bias. 3. A s i g n i f i c a n t bias for zinc and iron i s indicated by the extremely low p r o b a b i l i t y ( < 1% ) of obtaining such a high proportion of positive differences in metal contents between duplicate pa i r s . 4. The percentage of duplicate p a i r s exhibiting i d e n t i c a l metal concentrations i s about 25 percent. These r e s u l t s suggest that for the majority of metals, no major systematic biases are present and application of techniques such as analysis of variance and Howarth and Thompson precision procedure should provide meaningful r e s u l t s . The r e s u l t s of the analysis of variance are presented in Table 2.3 As expected, the between-stream v a r i a b i l i t y i s decidely higher than the within-stream dispersion for a l l of the metals studied. This feature i s encouraging because the purpose of these regional surveys i s to define a regional trend related to geological and geochemical phenomena; the greater the v a r i a b i l i t y in metal concentrations between-sample s i t e s the greater the ease of defining possible trends. Results obtained by other investigations (Bolviken and Sinding-Larsen, 1973 and Chork, 1977 using s i m i l i a r type data 22 Table 2.3 Analysis of Variance Results For 132 duplicate pairs of stream sediments B E T W E E N W I T H I N . V A R I A N C E C O M P O N E N T S C R A N O M E A N S I T E S S I T E S B E T W E E N W ITM IN M E T A L ( . G E O M E T R I C * * *\ f. S I T E S S I T E S X X Zinc so IS658' 852" 89 t l Copper 26 11S.4S* 1.6t nt 97 3 Nickel IS 5951 ' 2Si ~ 91 9 Cobalt 8 40.70" 020 99 1 Mangenete 292 276J68* OX nt 99 1 Iron 1.68 313.30* 1356' 91 9 The l«it d«pen<f« on the following null hypothet't: (a) Between ump l i titct — mean metal content! are equal at each lilt. Oegreet ol freedom are (/n — 1. m — I J. where m - number of titet - 132. <b| Within (ample <ile< — mean of templet Ml. it equal to mean of ouplicatct *2. Oegreet of freedom are (I. m — It. ' All valuet ant in ppm exoept for iron which it in per cent. Probability range* for accepting the nutt hypothetic • -0.01 >P;* - O j 0 S > f » > 0 j 0 t : e < « - l * > O X > S Table 2.4 Relative p r e c i s i o n estimates, Thompson-Howarth method B r i t i s h Columbia regional survey data (92H, 921 and 92J). PRECISION ESTIMATES AT SELECTED OETECTION MAXIMUM REGRESSION PERCENTILES METAL LIMIT* V A L U E SLOPE INTERCEPT F—VALUE 90th «Sth 99th X X X Zinc 2 210 OJ027* 0.892 nt 7.19* 7.0 6.8 6.5 Copper 2 720 0J018* 0-94S" 16.60' 6.7 S3 5.1 Nickel 2 1300 0.012' 0.629* 55.44' 4J0 3.2 2.5 Cobalt 2 96 0,029* 0.212 nt 3.14* 8.0 1A 6.7 Mangenete 5 3700 0.036' 2.78Snt 40.72' 7.8 7.7 7.6 Iron 0.02 S.70 0.011* 0.051* 2.S7* S.0 4.7 4.1 ' All valuet are in ppm except iron which it in per cent. Probability ranget for accepting t-tetu on the tlope and intercept and F-tett on the regrettion: • -0.01 > P; * -0.0S > P > 0.01;* -0.10 > P > 0.05; nt - P > 0.10 23 and t e s t s found that the v a r i a t i o n within-sample s i t e s was about 10-25% and the v a r i a t i o n between-sample s i t e s was about 75-90%. Such r e s u l t s are t y p i c a l f o r low d e n s i t y stream sediment surveys and c o n t r a s t markedly w i t h r e s u l t obtained from t h i s study (Table 2.3). The B.C. D u p l i c a t e data e x h i b i t s u r p r i s i n g l y lower within-sample s i t e v a r i a n c e , averaging l e s s than 4% of the t o t a l v a r i a b i l i t y . The s i g n i f i c a n c e of these anomalous r e s u l t s are elaborated upon i n a l a t e r s e c t i o n . R e s u l t s obtained from the Thompson and Howarth method are presented i n Table 2.4 and i l l u s t r a t e d g r a p h i c a l l y f o r copper on Figures 2.2 and 2.3. Although, the r e g r e s s i o n of the median of absolute d i f f e r e n c e s on the c o n c e n t r a t i o n means was only based on 12 s i n g u l a r p o i n t s , a n a l y s i s of variance on the regression proved s i g n i f i c a n t at the 99% confidence i n t e r v a l f o r a l l metals except c o b a l t . A regression p l o t of copper i l l u s t r a t e d i n Figure 2.3 shows c o n c l u s i v e l y t h a t the simple l i n e a r r e g r e s s i o n more than adequately accounts f o r the r e l a t i o n s h i p between the median of a b s o l u t e d i f f e r e n c e s and mean c o n c e n t r a t i o n s ; thus i t provides an e x c e l l e n t i n d i c a t o r of p r e c i s i o n over the c o n c e n t r a t i o n range. I t would appear that the slope i s the major i n f l u e n c e on the p r e c i s i o n estimate of a given metal, and second, i t s r e l a t i v e magnitude q u a n t i f i e s the p r e c i s i o n estimate. Examination of Table 2.4 reveals that n i c k e l posseses the r e l a t i v e l y lowest slope (0.012) wherease manganese has the highest (0.036). The i n t e r c e p t for most of the elements i s not ' ^ x g h i f x c a n t l y d i f f e r e n t than zero, and the magnitude of the 24 Figure 2.2 Absolute differences of paired data versus corresponding mean values of pairs, copper i n stream sediments, B r i t i s h Columbia Regional Geochemical Survey Data, (92H, 921 and 92J). t n -tu-tC Ul »- o < -o -o. D o «« ui ~ ft. Ui o o U i " or ui ui „ ' t-o «> -» < no j*.o —r— 4«0 —T • 0.0 •••0 — I — lOt.O 110.0 MEAN OF OUPLICATE RESULTS COPPER (ppm) L i n e a r model of average e r r o r as a f u n c t i o n of c o n c e n t r a t i o n , copper i n stream sediments. B r i t i s h Columbia R e g i o n a l Geochemical Survey.Data (92H, 92J and 921). I 26 slopes i s extremely small, therefore the precision i s high. For example, Figures 2.2 and 2.3 i l l u s t r a t e for the metal copper (1) the plot of absolute differences between duplicate pairs against means of duplicate p a i r s , with the concentration range divided into equal frequency i n t e r v a l s ; and (2) a plot of the regression of the median absolute difference against mean concentration of copper. To determine the precision as an absolute value obtain graphically the absolute difference corresponding to a selected concentration value on the regression l i n e and multiply by 2. Thus, for copper at the : 50th percentile (26 ppm) absolute precision = 2.8 ppm 95th percentile (59 ppm) absolute precision = 3.9 ppm 99th percentile (80 ppm) absolute precision =4.7 ppm DISCUSSION OF RESULTS In general, a l l data c o l l e c t e d in stream sediment surveys contain errors that are acquired through sampling, laboratory analysis and data handling. Taking the existence of sampling errors into account a precision of 10 - 15% at the 95% confidence l e v e l i s generally regarded as acceptable for laboratory v a r i a b i l i t y in most exploration programmes (Fletcher, 1981). Studies t a i l o r e d to the evaluation of error in stream sediment surveys such as Plant(1971), Howarth and Lowenstein (1971), Bolviken and Sinding-Larsen (1973), Plant et a l . (1975) and Chork (1977) generally concluded: 27 1. V a r i a b l e b i a s and v a r i a b l e p r e c i s i o n i n t r o d u c e d by secondary environment e f f e c t s obscure t h e p r i m a r y r e g i o n a l g e o c h e m i c a l v a r i a t i o n . The . f a c t o r s i n v o l v e d a r e complex and r e l a t e d t o s e v e r a l v a r i a b l e s i n the p r i m a r y and secondary environment of a r e a s i n v e s t i g a t e d . 2. M e t a l d i s p e r s i o n w i t h i n sample s i t e s depend on such f a c t o r s such a s : (a) c o n c e n t r a t i o n of t h e m e t a l under i n v e s t i g a t i o n (b) c o n c e n t r a t i o n of o t h e r m e t a l s ( e g . i r o n and manganese) (c) homogeneity of sediment c o m p o s i t i o n (d) catchment s i z e and d r a i n a g e d e n s i t y of s a m p l i n g s i t e 3. The combined v a r i a b i l i t y due t o l o c a l v a r i a t i o n and p r o c e d u r a l e r r o r ranged from 10 - 25% of t h e t o t a l e r r o r . 4. Sampling e r r o r s tended t o exceed a n a l y t i c a l e r r o r s when p r e c i s e a n a l y t i c a l t e c h n i q u e s such as a t o m i c a b s o r p t i o n s p e c t r o m e t r y a r e used. P r e c i s i o n e s t i m a t e s c a l c u l a t e d f o r t h i s s tudy a r e s i g n i f i c a n t l y lower than those o b t a i n e d from the o t h e r i n v e s t i g a t i o n s c i t e d . I t seems h i g h l y u n l i k e l y t h a t t h e magnitude of v a r i a b i l i t y a s s o c i a t e d w i t h t h e l o c a l e n v i r o n m e n t , s a m p l i n g and a n a l y t i c a l t e c h n i q u e s a r e a d e q u a t e l y a c c o u n t e d f o r , e s p e c i a l l y , i f we c o n s i d e r t h a t : 1. A h i g h p e r c e n t a g e of i n d i v i d u a l d u p l i c a t e p a i r s ( T a b l e 2.1, column 2) a r e c h a r a c t e r i z e d by i d e n t i c a l m e t a l c o n t e n t s o c c u r i n g o v e r the e n t i r e c o n c e n t r a t i o n range f o r a l l elements s t u d i e d . 2. R e l a t i v e v a r i a n c e components of w i t h i n - s a m p l e s i t e a v e r a g e d l e s s than f i v e p e r c e n t . 3. P r e c i s i o n e s t i m a t e s c a l c u l a t e d by the Thompson-Howarth method f o r s i x d i f f e r e n t m e t a l s a t d i f f e r e n t c o n c e n t r a t i o n l e v e l s a v e r a g e d l e s s than 6.5% a t the 9 0 t h p e r c e n t i l e c o n c e n t r a t i o n v a l u e , l e s s than 6.0% a t t h e 95th p e r c e n t i l e and l e s s t h a n 5.5% a t the ; 3 9 t h p e r c e n t i l e . 28 Results presented are more in accord with s p l i t s of an individual sample ( a n a l y t i c a l duplicate) rather than analyses of two separate f i e l d samples. If t h i s i s the case, the loss of information on the v a r i a b i l i t y associated with the sampling procedure and l o c a l environment may s e r i o u s l y hamper the interpretation of s i g n i f i c a n t geochemical trends. Furthermore, the whole reason for these regional surveys i s to f a c i l i t a t e provincal comparison, and i f the data are not of si m i l a r precision for a pa r t i c u l a r element and consistent r e l a t i v e to some baseline l e v e l in terms of accuracy, such comparisons w i l l be both d i f f i c u l t and unreliable. Since there appears to be some question over the type of duplicate analyzed, every e f f o r t should be made to ensure that proper f i e l d - s i t e duplicates are co l l e c t e d in future surveys. CONCLUSIONS To make the most e f f i c i e n t use of geochemical analyses the i r r e l i a b i l i t y must be known. Quality control throughout sample c o l l e c t i o n , sample preparation and a n a l y s i s should be regarded as an essential aspect of any geochemical prospecting method. The procedure outlined here combines both a graphical and s t a t i s t i c a l approach to the estimation of p r e c i s i o n . The technique allows q u a n t i f i c a t i o n of: 1. S i g n i f i c a n t systematic biases; 2. Relative within-sample s i t e and between-sample s i t e variance components; 3. Absolute precision at any p a r t i c u l a r 29 accumulation across i t s concentration range. As a r e s u l t , the method provides s a l i e n t information for decisions on threshold values, the significance of anomalies and contour i n t e r v a l s . This type of systematic procedure u t i l i z i n g f i e l d duplicates provides a basis for interpreting geochemical trends within the study area, while simultaneously allowing evaluation of the adequacy of the method of sampling and laboratory analysis. This information can also be used to design more e f f i c i e n t sampling schemes to improve contrast in geochemical r e s u l t s . ACKNOWLEDGEMENTS The author would l i k e to thank the B r i t i s h Columbia Ministry of Energy, Mines and Petrolem Resources for providing f i n a n c i a l support. 30 I I I . STATISTICAL EVALUATION OF THE SIGNIFICANCE OF CATEGORICAL  FIELD PARAMATERS IN INTERPRETATION OF REGIONAL GEOCHEMICAL  SEDIMENT DATA (MAP-AREA 82F), BRITISH COLUMBIA 31 ABSTRACT In an attempt to study the value and u t i l i z a t i o n of ca t e g o r i c a l data c o l l e c t e d during regional stream sediment surveys, we have taken data from the Canadian Uranium Reconnaissance Programme in S.E. B r i t i s h Columbia. After i n i t i a l c l a s s i f i c a t i o n of the data into six subsets on the basis of catchment geology, p r o b a b i l i t y plots were constructed for each of 11 elements (Zn, Cu, Pb, Ni, Co, Fe, Mn, Mo, W, Hg and U) and used to select thresholds to reject anomalous samples. The remaining background populations then were successively subdivided into groups according to their c l a s s i f i c a t i o n with respect to four sediment c h a r a c t e r i s t i c s (abundance of fines, sand, organic matter and sediment colour) and six environmental parameters (physiography, water flow rate, stream cl a s s , drainage pattern, bank type and contamination). After c a l c u l a t i o n of log means for each group and a pooled common standard error, differences between group means were tested for s i g n i f i c a n c e using Duncan's multiple range t e s t . The r e l a t i v e degrees of confidence in the s i g n i f i c a n c e of diff e r e n c e between cat e g o r i c a l means for any single f i e l d parameter were determined using a r a t i o method. Results of Duncan's multiple range test, show that, many f i e l d observations can be related systematically to metal content of drainage sediments. Some elements are more •..susceptible than others to environmental factors and some factors influence few or many elements. For example, in 32 sediments derived from granites there are s i g n i f i c a n t relationships between bank type and concentration of 8 elements (Zn, Cu, Ni, Pb, Co, Fe, Mn and Hg). In contrast, the texture of these sediments, using estimates of fines content as an index, did not s i g n i f i c a n t l y affect the concentration of any of the elements studied. In general, r e s u l t s indicate that groups of environmental factors acting c o l l e c t i v e l y are more important than any single factor in determining background metal content of drainage sediments. 33 INTRODUCTION During c o l l e c t i o n of drainage sediments for regional geochemical surveys, i t has become customary to record c l a s s -i n t e r v a l and categorical observations on c h a r a c t e r i s t i c s of the drainage catchment, sample s i t e and sediment sample. However, both the subjective character, and q u a l i t a t i v e and interdependent nature of these categorical observations has made i t d i f f i c u l t to make quantitative evaluations of their s i g n i f i c a n c e with respect to elemental concentrations. Consequently, apart from c l a s s i f y i n g data in terms of geology and screening anomalous samples ( S i n c l a i r and Fletcher, 1980; Brummer et a l . , 1978), u t i l z a t i o n of categorical information i s often so limi t e d or non-existent that the e f f o r t and cost involved in i t s a c q u i s i t i o n become d i f f i c u l t to j u s t i f y . This led Meyer et a l . , (1979) to suggest that serious consideration be given to reducing the number of observations to those of proven s i g n i f i c a n c e . In an attempt to study the value and u t i l i z a t i o n of cat e g o r i c a l f i e l d data c o l l e c t e d during regional stream sediment surveys, data from the Joint Federal-Provincial Uranium Reconnaissance Program in south-eastern B r i t i s h Columbia (Geological Survey of Canada Open F i l e Report OF514) have been evaluated (Matysek et a l . , 1981). Here we describe a systematic, computer-assisted method of investigating the sig n i f i c a n c e of a non-parametric data u t i l i z i n g Duncan's multiple range test (Duncan, 1955, 1957). 34 The more extreme ranges of the small populations of anomalous values complicate the probability density functions of geochemical variables and confuse the p o t e n t i a l l y d i f f e r e n t r e l a t i o n between categorical variables on the one hand and metal content on the other. Our philosophy has been to extract recognizable anomalous information from the data sets we are studying in order to better understand controls on the great bulk of background a n a l y t i c a l data normally acquired during large-scale regional stream sediment surveys. DESCRIPTION OF STUDY AREA The study area (Fig. 3.1), including approximately 14,500 km2 of the Columbia Mountains in south-eastern B r i t i s h Columbia, i s bounded by the 49th and 50th p a r a l l e l s of north latitude and by 116th and 118th meridians of west longitude (NTS map-sheet 82F). Topographically, the region i s characterized by northerly trending mountain ranges r i s i n g in a l t i t u d e from wooded h i l l s (1800m to 2100m asl) in the south to rugged mountainous te r r a i n s (2400m to 2800m asl) in the north. P r e v a i l i n g westerly winds result in an average annual p r e c i p i t a t i o n of about 750 mm and there i s a well developed active drainage system over the entire region. Major r i v e r s include the Columbia, Kootenay, Slocan, Moyie and South Mary. 't3Ehe mountain ranges, p a r t i c u l a r l y to the west of Kootenay lake L o c a t i o n map o f study area 82F, and index map o f r e g i o n a l geochemical surveys a v a i l a b l e i n B r i t i s h Columbia (McMillan, 1982). 36 are dissected by the major east-west or northwest-southeast flowing t r i b u t a r i e s of the main river systems. The area i s underlain by severely folded and faulted sedimentary and volcanic rocks, ranging from P u r c e l l (late Precambrian) to Cretaceous age, intruded by Mesozoic plutons (Fig. 3.2). These plutons, which include granodiorites, granites, syenites and d i o r i t e s , occupy much of the te r r a i n in the Selk i r k Mountains west of Kootenay Lake and enclose large gneissic roof pendents. East of Kootenay Lake the p r i n c i p a l l i t h o l o g i e s are Proterozoic quartzites of the Creston and Aldridge Formations. Less extensive, but nonetheless s i g n i f i c a n t rock units, include s l a t e s , a r g i l l i t e s and andesitic volcanic rocks a l l of which occur at several l e v e l s in the st r a t i g r a p h i c succession, (Rice (1941); L i t t l e (i960)). Over much of the region, to an elevation of 2400m, bedrock units are covered by a thin mantle of g l a c i a l t i l l with reworked f l u v i o - g l a c i a l deposits in the v a l l e y s . Above 2400m t i l l i s replaced by talus and felsenmeer. The region has a long mineral exploration h i s t o r y with four important producing mining camps; Rossland (Au-Cu), Slocan (Ag-Pb-Zn), Ainsworth (Zn-Pb-Ag) and Sheep Creek (Au) (Fig. 3.2). The Slocan and Ainsworth camps alone account for 350 vein deposits found in the area ( S i n c l a i r , 1979). METHODOLOGY The data base for the study comprises f i e l d observations 1 18*00' so'ocr 1 17*30' 117*00' 1 16*30' 49»45' 49"30 49*15 49°00' 1 16*00" ' 50*00' 1 18*00 117*30' 117*00' 116*30' GRANITE GNEISS SLATE ANDESITE t=^--EJ ARGILLITE QUARTZITE S C A L E 0 5 10 15 2 0 25 Kilometres 49°0CT 116*00" F i g u r e 3.2 S i m p l i f i e d geology of the study a r e a , m o d i f i e d from G e o l o g i c a l Survey of Canada Open F i l e Report 514, 38 (Table 3.1) and a n a l y t i c a l results (minus 80-mesh, Zn, Cu, Pb, Ni, Co, Mn, Fe, Mo, W, Hg, U) for drainage sediments c o l l e c t e d at 1313 s i t e s as part of the Joint Federal-Provincial Uranium Reconnaissance Program. D e t a i l s of sampling and a n a l y t i c a l procedures are to be found in Geological Survey of Canada Open F i l e Report OF514. These data are a v a i l a b l e p u b l i c a l l y on magnetic tape from the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources, Geological Branch, Mineral Resources D i v i s i o n , V i c t o r i a , and the Federal Department of Energy, Mines and Resources, Ottawa. For the purposes of t h i s study, geology has been s i m p l i f i e d by combining s i m i l i a r l i t h o l o g i e s , i r r e s p e c t i v e of their age, into the six rocktypes - granite, q u a r t z i t e , s l a t e , a r g i l l i t e , andesite and gneiss - that characterize the catchments of the sediment samples (Fig. 3.2). Spatial density and number of samples for these units are summarized in Table 3.2. Evaluation of the f i e l d observations then included the following steps: 1. Sorting of data into provenance groups, i . e . , predominant rock type in drainage basin upstream from the sample. 2. Evaluation of simple s t a t i s t i c s and p r o b a b i l i t y graphs for each element. 3. Threshold selection using the method of S i n c l a i r (1976) to i s o l a t e anomalous samples from background samples. 4. Subdivision of background populations into groups according to their c l a s s i f i c a t i o n with respect to sediment c h a r a c t e r i s t i c s and environmental v a r i a b l e s . 5. Calculation of log means and pooled standard error for each group: testing of 39 Table 3.1 Field-recorded categorical variables, regional «tream-<ediment survey. Geological Survey of Canada OF514 (NCR 25-1977) Sample variables (4): Categories (1) Fines Content Absent, Minor (0—33%), Medium (34—66%), Major (> 67%) (2) Sand Content Absent, Minor (0—33%). Medium (34—66%), Major (> 67%) (3) Organic Content Absent, Minor (0—33%), Medium (34—66%), Major (> 67%) (4) Sediment Colour Red, White, Black, Yellow, Green Environmental variables (6): Categories (1) Banktype Alluvial, Colluvial, Glacial Till, Other (2) Waterflow Rate Slow, Moderate, Fast (3) Stream Class Primary, Secondary, Tertiary, Quaternary (4 j Physiography Hilly, Mountainous Mature, Mountainous Youthful (5) Drain Pattern Poor, Dendritic, Herringbone, Other (6) Contamination None, Possible, Probable, Definite Table 3.2 Spatial density of stream sediment samples: Nelson map-area Lithology Map unit Area (km1) No. of samples Samples/km* Granite 20 6 825 485 12.0 Quartzite 1 3 775 318 11.9 7 430 32 13.4 Combined 4 205.' 350 12.0 Slate 6 585 61 9.6 9 420 51 8.2 14 390 33 11.8 Combined 1 395 145 9.6 Argillite 2 630 58 10.9 17 420 31 13.5 Combined 1 050 89 11.8 Andesite 5 40 5 8.0 15 50 6 8.3 18 495 62 8.0 Combined 585 73 8.0 Gne'ss 13 660 73 9.0 Totals 13 720. 1 215 11.3 40 difference between means using Duncan's Multiple Range Test. 6. Determination of r e l a t i v e degree of confidence and/or consistency in the s i g n i f i c a n c e of the differences between group means using a ra t i o method. 7. Results of (5) v i s u a l l y presented in Venn diagram form. This procedure has been computerized to allow rapid systematic evaluation of f i e l d parameters, and u t i l i z e s programs by Le (1970) and the senior author. Analysis of histograms and pr o b a b i l i t y plots indicates that in d i v i d u a l metal frequency d i s t r i b u t i o n s for each rock type are generally both lognormal and bimodal in form. Data were therefore p a r t i t i o n e d using log-probability p l o t s into low (probably background) and high (probably anomalous) populations ( S i n c l a i r , 1976). Anomalous samples were then rejected leaving only background samples for c l a s s i f i c a t i o n into groups based on the f i e l d observations. These were then further subdivided into groups according to their c l a s s i f i c a t i o n with respect to four sediment c h a r a c t e r i s t i c s (fines, sand, organic content and colour) and six environmental variables (physiography, waterflow rate, stream c l a s s , bank types, drainage pattern and contamination) as described in Table 3.1. Log means and a pooled standard error from an analysis of variance on these categorical means for each f i e l d v a r i a b l e were calcu l a t e d . The significance of differences between the means of a l l groups were then tested with Duncan's multiple range test (Duncan, 1955, 1957). This test determines i f dif f e r e n c e s among 41 means of s e v e r a l groups a r e s i g n i f i c a n t a t a g i v e n l e v e l of c o n f i d e n c e . The t e s t has been a p p l i e d p r e v i o u s l y t o r e g i o n a l g e o c h e m i c a l d a t a by M i e s c h (1976) and D o y l e and F l e t c h e r (1979). The t e s t assumes t h a t the means x,, x 2 , . . . , x a r e drawn i n d e p e n d e n t l y from 'n' normal p o p u l a t i o n s h a v i n g t r u e means m,, m2,... , m r e s p e c t i v e l y . The c o n v e n t i o n a l s t a t i s t i c f o r the co m p a r i s o n of two means i s : t = ( x , - x 2 ) / S (1/n, + 1/n 2) (1 ) where x, and x 2 a r e the e s t i m a t e s of the two means from the p o p u l a t i o n s w i t h e q u a l v a r i a n c e , e s t i m a t e d by S 2 x and a r e based on n, and n 2 independent measurements of v a r i a b l e s x, and x 2 , r e s p e c t i v e l y . For a t e s t a t the 95 per c e n t c o n f i d e n c e l e v e l (a=0.05) the computed t i s compared w i t h the t a b l e d v a l u e s of S t u d e n t ' s t d i s t r i b u t i o n (1-a/2,v) where v i s the degrees of freedom on which t h e e s t i m a t e of S i s based ( e q u a l t o n*+n|-2). I f n5=nt?=n, and t h e c r i t i c a l v a l u e of T ( l - a / 2 ) i s used f o r t , e q u a t i o n 1 then becomes: X, - x 2 = Sr = 2 */ t S p (2) Sr i s the s h o r t e s t s i g n i f i c a n t s t u d e n t i z e d range f o r the two means, and Sp i s t h e p o o l e d s t a n d a r d e r r o r common t o x, and x 2 . The q u a n t i t y 2 * J t S p i s r e f e r r e d t o as t h e s h o r t e s t s i g n i f i c a n t s t u d e n t i z e d range range by Duncan ( 1 9 5 5 ) . For a=0.05 and say v=60, t h e s i g n i f i c a n t s t u d e n t i z e d range i s • -2:,33. Thus, d i f f e r e n c e s of a t l e a s t 2.83 t i m e s t h e s t a n d a r d 42 error of the means are required in order to declare the means as s i g n i f i c a n t l y d i f f e r e n t . Duncan (1957) extended th i s method by c a l c u l a t i n g s i g n i f i c a n t studentized ranges (zp) when more than two means (p>2) are compared and variances are not equal. The means are f i r s t ordered by magnitude then i f two adjacent means are being compared, p=2; i f the two means compared are separated by one mean, p=3, and so on. The s i g n i f i c a n t studentized range increases with an increase in p. For example where p=4 with v=60 the value i s 3.08, compared with 2.83 where p=2 (a=0.05). If a l l the means are equally replicated, the s i g n i f i c a n t studentized ranges (tsp) are multi p l i e d by the pooled standard error, and the means are ranked and divided into homogeneous groups. More commonly, i f the means are unequally r e p l i c a t e d , shortest s i g n i f i c a n t studentized ranges are calculated in the same manner. These must be compared with an adjusted difference (AD) between the p means which takes into consideration the number of observations each mean i s based upon (n, and n 2 r e s p e c t i v e l y ) : AD = (x, - x 2 ) * J(2n,*n 2/(n 1+n 2)) (3) Any subset of p means i s homogeneous i f the largest adjusted difference in the subset f a i l s to exceed the c r i t i c a l value Sr. Two means not in the same homogeneous subset are s i g n i f i c a n t l y d i f f e r e n t . The r e l a t i v e degree of confidence in the s i g n i f i c a n c e of differences between c a t e g o r i c a l means for any single f i e l d 43 parameter can be determined by comparing the adjusted difference between the highest c a t e g o r i c a l mean (xh) and the lowest categorical mean (xl) with t h e i r shortest s i g n i f i c a n t studentized range (Sr). D = (xh - xl ) *JT2n,*n 2/(n,+n 2)) / Sr (4) A r a t i o less than 1 indicates that the differences between means are just s i g n i f i c a n t at the 95% confidence l e v e l . Greater confidence in the sig n i f i c a n c e of the difference between means is r e f l e c t e d by an increased r a t i o . Comparison of the D-ratio thus provides a guide to the r e l a t i v e strength or consistency of the influence of the various f i e l d parameters on trace element composition. For example, considering zinc in sediments associated with granites, 43 samples (from a t o t a l of 485) are rejected as anomalous a f t e r interpretation of the zinc probabality graph. If sediment colour i s the f i e l d observation of interest, the remaining 442 sediments can be subdivided into f i v e (red, white, yellow, black, and grey) groups, and log means and an analysis of variance on the means ca l c u l a t e d . However, the only colours recorded with reasonable frequency (n>lO) are red (n=31l), white (n=105) and black (n=26) with corresponding means of 63, 49 and 70 ppm. Table 3 i l l u s t r a t e s the steps involved in application of Duncan's multiple range test and D-ratio method to these three c a t e g o r i c a l means. CTJ,j3,this case the test establishes that concentrations of zinc in black and red sediments are indistinguishable but those 44 in white sediments are s i g n i f i c a n t l y lower. The res u l t s can be presented as in Table 3.3 and below, where any two means not appearing within the same parentheses are s i g n i f i c a n t l y d i f f e r e n t , or as Venn diagrams in which overlapping parts of c i r c l e s indicate groups whose means are not s i g n i f i c a n t l y d i f f e r e n t (Fig. 3.3 and 3.4). RESULTS The s i g n i f i c a n c e of differences of means among categorical f i e l d observations were tested and presented as Venn diagrams as shown in Figs. 3.3 and 3.4. Although, the procedure was tested on six d i f f e r e n t drainage basin l i t h o l o g i e s , only results from sediments associated with streams draining granite and quartzite terrains w i l l be presented in d e t a i l . Bulk composition - Fines content Categories: 2=minor(0-33%) 3=medium(34-66%) 4=major(>67%) Granite: (no s i g n i f i c a n t differences) Quartzite: Zn,Cu,Ni,Co,Fe,Mn (23)(31) Variation in the content of fines in sediments derived from granites has no apparent influence on any of the eleven metal concentrations studied, whereas for quartzites concentrations of zinc, copper, n i c k e l , cobalt, iron and manganese are s i g n i f i c a n t l y a f fected. However, there i s a considerable degree 45 T a b l e 3.3 5% level Duncan's Multiple Range Test of three unequally replicated means and calcula-tion of D-ratio (1) Ranked means an d replication numbers Sediment colour White Red Black Log mean 1.69 1.80 1.845 Geometric mean 49 63 70 Number (105) (311) (26) (2) Analysis of variance Source Degrees of Freedom Mean Square F-ratio Between means 2 0.50781 13.815 Error 439 0.03676 Pooled standard error (S p) = V0.03676 - 0.1917, F-crit - 4.567 (3) Critical values Sr = Sn zn r P P p: (2) (3) zp: 2.7853 2.9290 Sr: 0.5340 0.5616 (4) Calculated adjusted differences (AD) between p means Black-White - (1 .845 — 1.69) >/2(26)(105)/(26 + 105) - 1.000 Black-Red = (1 .845 — 1.80) >/2(26)(311)/(26 + 311) - 0 . 3 1 2 Red-White = (1 .80 — 1.69) >/2(105)(311)/(105 + 311>» 1.378 (5) Test sequence: comparing Sr to calculated adjusted difference Black-White = 1.000 > ST = (0.5616) Black-Red - 0 .312 < S r = (0.5340) Red-White = 1.378 > S r = (0.5340) (6) Result (White) (Red, Black) Any two means not appearing together within the same parentheses are significantly different. Any two means appearing together within the same parenthesis are not different. (7) D-ratio: ratio comparing the adjusted difference (i4D) between the highest and lowest means with their critical value 5, means significantly different means not significantly different means significantly different D •* 1.000/0.5616 = 1.78 X - 57 68 77 N- 283 135 30 ( 1 ) ( 2 3 ) X - 11 16 20 N- 288 129 31 0 X - 9 9 9 N- 270 130 31 ' 0 X - 7 8 8 N- 142 294 36 0 X - 5 5 5 N- 137 271 36 X - 17 23 30 N-295 142 33 0 X - 1.4 1.4 1.5 N- 131 287 35 M " 0 X - 350 386 397 N- 297 129 31 0 X - 11 13 14 N- 298 137 25 M D 0 X - 1 1 2 N-265 129 33 0 X - 2 2 2 N- 128 265 38 V A R I A B L E S 1 - ABSENT 2 - MINOR (0-34%) 3 - MEDIUM (34-67%) ALL ELEMENTS ARE IN ppm. EXCEPT: Hg ppb. Fe % F i g u r e 3,3 Duncan's m u l t i p l e range t e s t f o r t h e i n f l u e n c e o f o r g a n i c c o n t e n t o f stream sediments a s s o c i a t e d ; w i t h a g r a n i t i c p r o v e n a n c e . Common o r o v e r l a p p i n g c i r c l e s i n d i c a t e t h a t group means a r e not s i g n i f i c a n t l y d i f f e r e n t a t the 0.05 c o n f i d e n c e l e v e l . X- 50 64 70 N- 105 311 26 X*- 9 12 18 N- 102 302 15 c o o 5C- 8 9 12 N- 95 313 23 " O O X - 8 8 13 N- 336 109 23 ° 0 X - 5 5 5 N- 99 318 27 " C 0 0 X- 14 21 31 N- 108 336 22 F G 0 X - 1.4 1.5 1.5 N- 105 320 28 M " 0 D X~- 317 339 382 N- 109 320 28 U 0 D 5?- 9 13 14 N- 108 23 325 Mo O X - 1 1 1 N- 96 28 303 W . P 2 7 X - 2 2 2 N- 27 96 305 V A R I A B L E S 1 - RED 2 - WHITE 3 - BLACK ALL ELEMENTS ARE IN ppm. EXCEPT: Hg ppb. Fe% F i g u r e 3.4 Duncan's m u l t i p l e range t e s t f o r the i n f l u e n c e o f sediment c o l o u r o f stream sediments a s s o c i a t e d w i t h a g r a n i t i c provenance. Common o r o v e r l a p p i n g c i r c l e s i n d i c a t e t h a t group means are not s i g n i f i c a n t l y d i f f e r e n t a t the 0.05 conf idence l e v e l . 48 of overlap between categories ( i . e . , no single categorical group forms an independent s t a t i s t i c a l l y s i g n i f i c a n t population) and although r e l a t i v e l y higher concentrations of metals are found in sediments containing major amounts of fines, concentrations are not s i g n i f i c a n t l y d i f f e r e n t to those in sediments lacking f i n e s . Bulk composition - Sand content Categories: 1=absent 2=minor(0-33%) 3=medium(34-66%) 4=major(>67%) Granite: Zn(324)(24l) Pb(3)(24)(41) Mn(324)(4l) W(412)(3) Mo(342)(421), Hg(324)(41), U(34)(421) Quartzite: Zn(234)(341) Texture, as re f l e c t e d by sand content s i g n i f i c a n t l y a f f e c t s concentrations of seven elements in sediments derived from granites but only zinc for q u a r t z i t e s . As with fines, there i s considerable overlap between categories and the greatest concentrations of zinc, lead, manganese, molybdenum and mercury appear to occur in sediments containing either major quantities of sand (category 4) or no sand at a l l (category 1). Bulk composition ^ Organic content (Fig. 3) Categories: 1=absent 2=minor(0-33%) 3=medium(34-66%) Granite: Zn(l)(23) Pb(l)(23) Hg(l)(2)(3) MO(12)(3) Quartzite: Pb(l3)(32) Hgd3)(32) Varia t i o n in the content of organic matter s i g n i f i c a n t l y inlluences z i n c , lead, mercury and molybdenum concentrations in 49 sediments derived from g r a n i t i c terrane, and lead and mercury in those derived from q u a r t z i t e s . In general, although sediments from both rock types display increasing metal contents with increase in organic matter, those derived from granites have more s t a t i s t i c a l l y s i g n i f i c a n t independent ca t e g o r i c a l groups (Fig. 3.3). In a l l cases, lowest metal concentrations are found in sediments judged to contain no organic matter with t h i s category forming a s t a t i s t i c a l l y independent group for lead, mercury and zinc for sediments derived from granites. Sediment colour (Fig. 3.4) Categories: 1=red 2=white 3=black Granite: Zn(2)(13), C U(21)(3), Pb(2)(1)(3), N i ( l 2 ) ( 3 ) , Mn(23)(31), U(23)(31), H g ( 2 ) ( D ( 3 ) Quartzite: Zn(23)(31), Pb(23)(3l), W(13)(32), Hg(2l)(3) Variation in sediment colour i s related s i g n i f i c a n t l y to concentrations of six elements in g r a n i t i c catchments and four elements in q u a r t z i t i c catchments. Zinc, lead and mercury concentrations are s i g n i f i c a n t l y affected in both groups and the general trend throughout i s for r e l a t i v e l y high metal content to be associated with red or black sediments. For granites, greater average concentrations of copper, lead, n i c k e l , mercury and zinc are associated with black sediments and except for zinc these form s t a t i s t i c a l l y s i g n i f i c a n t independent groups. In the case of mercury there i s ^ppxxtximately a twofold difference between average concentrations in white (14) and black (31) sediments 50 (Fig. 3 . 4 ) . In contrast, red sediments contain somewhat higher average concentrations of manganese and uranium. Banktype Categories: 1=alluvial 2=colluvial 3=glacial t i l l Granite: Z n ( D ( 3 2 ) , C u d ) ( 2 3 ) , P b ( l ) ( 3 2 ) , N i ( 1 2 ) ( 2 3 ) , C o ( D ( 2 3 ) , F e ( l ) ( 3 2 ) , M n ( l ) ( 2 ) ( 3 ) , H g ( l ) ( 2 3 ) Quartzite: U ( 3 ) ( 2 1 ) , Mo ( 3 D ( 2 ) , Hg ( l 3 ) ( 3 2 ) Variation in bank type s i g n i f i c a n t l y affects concentrations of eight elements associated with predominantly gr a n i t i c catchment basins and three elements in samples derived from qu a r t z i t e s . Mercury i s the only element influenced in both rocktypes. For g r a n i t i c sediments, r e l a t i v e l y high average concentrations of zinc, copper, lead, n i c k e l , cobalt, iron, manganese and mecury are associated with g l a c i a l and c o l l u v i a l banktypes, and for a l l of these elements, except n i c k e l , s i g n i f i c a n t l y lower average concentrations are associated with a l l u v i a l banks. In contrast, for q u a r t z i t i c sediments, r e l a t i v e l y higher average concentrations of uranium, molybdenum and mercury are associated with c o l l u v i a l banktypes and r e l a t i v e l y lower contents with samples obtained from g l a c i a l l y derived banktypes. However, only uranium forms a s t a t i s t i c a l l y independent group for the l a t t e r category. Wa t er f 1 o w ,xa t e t 51 Categories: 1=slow 2=moderate 3=fast Granite: Z n ( 3 ) ( 2 ) d ) , P b ( 3 ) ( 2 ) ( 1 ) , N i ( 2 ) ( 3 ) ( 1 ) , C U ( 3 2 ) ( 1 ), Hg ( 3 2 ) ( D Quartzite: C U ( 2 3 ) ( 1 ) , N i ( 2 3 ) ( D , M n ( 2 D ( 1 3 ) , Hg (23) (31> V a r i a t i o n in stream flow s i g n i f i c a n t l y relates to metal contents for f i v e elements in g r a n i t i c terranes and for four elements at sample s i t e s with a quartzite provenance. Both groups exhibit s i m i l i a r trends with sediments associated with the slower flow rates having the highest metal contents. This i s most apparent with g r a n i t i c sediments in which z i n c , lead, n i c k e l , copper and mercury a l l have the i r greatest concentrations, forming a s t a t i s t i c a l l y s i g n i f i c a n t d i f f e r e n t category, in slow flowing streams. Nickel and copper exhibit a s i m i l i a r phenomenom in sediments derived from quartzites whereas for granites the lowest and s t a t i s t i c a l l y independent concentrations of zinc, lead, copper and mercury are associated with fast flowing streams. Stream c l a s s Categories: 1=secondary 2=tertiary 3=quaternary Granite: Z n ( 2 ) ( 3 l ) , C u ( 3 2 ) ( D , P b ( 2 ) ( 3 l ) , N i ( 3 2 ) ( l ) , C o ( 2 3 ) ( l ) , Hg ( 3 2 ) ( 2 l ) Quartzite: Z n ( 2 ) ( 3 ) , Cu ( 2 ) ( 3 ) , P b ( 2 ) ( 3 ) , N i ( 2 ) ( 3 ) C o ( 2 ) ( 3 ) , Mn (2) (3) Differences in stream class are s i g n i f i c a n t l y related to xonc-ejitrations of zinc, copper, lead, nickel and cobalt in sediments from both granites and quartzites. In addition 52 mercury and manganese concentrations are affected in granite and quartzite derived sediments, respectively. In a l l cases for quartzites, together with zinc and lead in sediments derived from granites, the lowest metal concentrations are found in samples from t e r t i a r y streams. In sediments on granites the secondary drainage class (category 1) forms a s i g n i f i c a n t l y d i s t i n c t i v e category characterized by above average copper, nickel and cobalt concentrations. Drainage pattern Categories: 1=poor 2=dendritic 3=herringbone Granite: Zn(13)(32), Cu(D(32), Pb(31)(2), Co(13)(32), Mn(13)(32), U(12)(3), Hg(3l)(l2) Quartzite: Cu(3)(2D U(12)(3) Variation in drainage pattern type are s i g n i f i c a n t l y related to concentrations of seven and two elements from gran i t i c and q u a r t z i t i c t e r r a i n s , respectively. Only uranium i s influenced for both rock types. Although there i s considerable overlap between drainage pattern categories and metal concentrations, there i s a general tendancy for r e l a t i v e l y high metal concentrations to occur in sediments of streams characterized by either d e n d r i t i c or herringbone drainage patterns. Contamination Categories: 1=absent 2=possible 3=probable 4=definite 53 Granite: U ( 4 3 2 ) ( 3 2 l ) Quartzite: M n ( 3 2 ) ( 2 4 l ) H g ( 3 2 4 ) ( 4 1 ) Of the ten f i e l d parameters examined, contamination appears to influence the least number of metals. However, i t should be noted, that anomalous concentrations of the metals res u l t i n g from contamination have for the most part been eliminated from the data set. For example, for g r a n i t i c sediments 2 4 , 2 5 and 34 samples have been removed from the test as a r e s u l t of their anomalous content of zinc, copper and lead, respectively. Physiography Categories: 1=hilly 2=mountainous mature 3=mountainous youthful Granite: Z n ( 3 ) d 2 ) , P b ( 3 ) ( 1 2 ) , M n ( 3 ) ( l 2 ) , U ( 3 1 ) ( 1 2 ) , H g ( l 3 ) ( 3 2 ) Quartzite: Z n ( 2 3 ) ( l ) , C u ( 3 ) ( 2 1 ) , N i ( 3 2 ) ( l ) , F e ( 3 2 ) ( 1 ) Variation in physiographic c l a s s i f i c a t i o n of stream catchments are s i g n i f i c a n t l y related to five elements for sample s i t e s on granites and two on q u a r t z i t e s . In both cases, higher metal concentrations are associated with sediments obtained from h i l l y and mountainous mature t e r r a i n s . Concentrations of zinc, n i c k e l and iron form a s i g n i f i c a n t l y unique category of above average concentrations in catchments c l a s s i f i e d as h i l l y . In contrast, on granites a s i g n i f i c a n t l y d i s t i n c t i v e category with below average concentrations of zinc, lead and manganese i s ^associated with the mountainous young physiographic c l a s s i f i c a t i o n . 5 4 From the foregoing i t i s apparent that despite their q u a l i t a t i v e , subjective character f i e l d observations can be related to variations in the trace element content of sediments associated with a single rock unit. These findings are summarized conveniently in Tables 3 . 4 and 3 . 5 , for granites and quartzites respectively, where i t i s apparent which parameters influence the most elements or which elements are most susceptible to p a r t i c u l a r environmental factors. As an example, for streams draining granites (Table 3 . 4 ) we note that the bank type s i g n i f i c a n t l y influences 8 of 1 1 elements, whereas the content of fines apparently influences none of the elements. Conversely, the s u s c e p t i b i l i t y of an element for i t s concentration to influence to the f i e l d parameters decreases from zinc, lead and mercury abundances of which are affected s i g n i f i c a n t l y by eight factors, to those elements whose abundances are affected by fewer recorded factors, eg , copper, manganese and uranium ( 5 f a c t o r s ) , n i c k e l ( 4 f a c t o r s ) , cobalt ( 3 factors), molybdenum ( 2 f a c t o r s ) , iron and tungsten ( 1 f a c t o r ) . Furthermore, comparisons of D ratios indicate, for example, that of the eight parameters s i g n i f i c a n t l y a f f e c t i n g concentrations of mercury, sediment colour and organic content have the most consistent or strongest influence. These results provide an objective c r i t e r i a for assessing s u s c e p t i b i l i t y of trace element dispersion to environmental factors and determining which are most relevant to geochemical exploration programs. DISCUSSION 55 T a b l e 3 .4 D-Ratio calculated for categorical field variables. Granitic provenance, with group means shown to be significantly different by Duncan's Multiple Range Test (95% confidence) Zn Pb Hg Cu Ni Co Fe Mn U W Mo Tot Fines content _ _ _ — — — — ^QJ Sand content 1.79 2.37 1.28 — — — — 1.39 1.13 i n 1 3 7 (7) Organic content 1.63 2.43 2.32 — — — — — — — 1 4 3 (4) Sediment colour 1.78 1.79 2.42 1.51 1.34 — — 1.90 2.17 — — (7) Water flow rate 1.99 1.94 1.99 1.68 1.74 — — — — — — (5) Physiography 2.59 2.80 1.46 — — — — 1.93 1.30 — — (5) Stream class 1.07 1.16 1.98 1.84 1.35 1.26 - - - - - (6) Drain pattern 1.47 2.57 1.98 1.81 — 1.15 — 1.14 1.26 — — (7) Bank type 1.34 1.95 2.06 1.62 1.73 1.48 1.42 3.23 - - - (8) Contamination — — — — — — I.55 — — — — (1) Totals (8) (8) (8) (5) (4) (3) (1) (5) (5) (1) (2) (50) Dashes indicate that significant differences among groups were not identified by Duncan's Multiple Range Test. T a b l e 3 . 5 D-Ratio calculated for categorical field variables. Quartzitic provenance, with group means shown to be significantly different by Duncan's Multiple Range Test (95% con-fidence) Zn Pb Hg Cu Ni Co Fe Mn U W Mo Tot Fines content 1.28 — — 1.24 1.80 1.11 1.31 1.23 — — — (6) Sand content 1.14 — — — — — — — — — — (1) Organic content — 1.56 1.33 — — — — — — — — (2) Sediment colour 1.18 1.22 1.29 — — — — — — 1.03 — (4) Water flow rate — — 1.24 1.18 1.41 — — 1.37 — — — (4) Physiography 1.41 — — 1.91 1.29 — 1.16 — — — — (4) Stream class 1.22 1.57 — 1.59 1.83 2.59 — 1.80 1.72 — — (7) Drain pattern — — — 1.37 — — — — 1.50 — — (2) Bank type _ _ _ _ _ _ _ _ 1 - 4 4 _ 1 > 2 6 (2) Contamination — — 1.09 — — — — 1.32 — — — (2) Totals (5) ( 3 ) (4) (5) (4) (2) (2) (4) (3) (1) (1) ( 3 4 ) Dashes indicate that significant differences among groups were not identified by Dun ca « ' S Multiple Range Teat. 56 Results c l e a r l y indicate that, although there are differences both between rock units and elements, the various ca t e g o r i c a l and non-parametric f i e l d observations can be related to s i g n i f i c a n t v a r i a t i o n s in trace metal concentrations. Furthermore, i t i s encouraging to note that, despite their subjective character, results are reasonably consistent with well known controls on the behaviour of the trace elements. For example, association of enhanced concentrations of z i n c , lead and mercury with organic-rich sediments on granites i s consistent with scavenging of these elements by organic matter. S i m i l a r i l y sediment colour exhibits a consistent r e l a t i o n s h i p with above average metal concentrations of zinc, copper, lead, n i c k e l , manganese, uranium and mercury on granites being associated with strongly coloured red or black sediments presumably a result of scavenging by iron sesquioxides and organic matter, respectively. Consistency between the results and anticipated geochemical trends i s further emphasized in Tables 3.6 and 3.7 (3.8 and 3.9) in which those factors favoring above (or below) average concentrations of the metals are summarized. Thus, on granites, i t i s apparent that r e l a t i v e l y high organic content, black, fine grained sediment and a slow water flow rate a l l favor above average metal contents. In contrast, low organic content in r e l a t i v e l y coarse sediments from fast flowing streams are l i k e l y to be characterized by low metal concentrations. Physiographic features, stream c l a s s and drainage pattern are also implicated as important factors. However, i t i s not so clear what t h e i r T a b l e 3.6 Specific categories of individual categorical field variables with highest means, Granitic provenance Zn Pb Hg Cu Ni Mn U Fines content (Major) (Major) (Major) (Major) (Major) (Major) (Major) Sand content Absent Absent Absent (Absent) (Major) Absent Absent Organic content Medium Medium Medium (Medium) (Medium) (Medium) (Medium) Sediment colour Black Black Black Black Black Red Red Water flow rate Slow Slow Slow Slow Slow (Slow) (Moderate) Physiography Mt. Mature Mt. Mature Mt. Mature Mt. Mature (Mt. Mature) Mt. Mature Mt. Mature Stream class Secondary Secondary Secondary Secondary Secondary (Quaternary) (Quaternary) Drain pattern Dendritic Dendritic Dendritic Dendritic (Herringbone) Dendritic Herringbone Bank type Colluvial Colluvial Glacial Glacial Glacial Colluvial (Alluvial) Unbracketed categories are significantly different (95% confidence level) from at least one other category for each variable. Cate-gories within brackets were not found to be significantly different from other categories. Table 3.7 Specific categories of individual categorical field variables with highest means, Quartzitic provenance Zn Pb Hg Cu Ni Mn U Fines content Minor (Minor) (Minor) Minor Minor (Major) (Minor) Sand content Absent (Absent) (Major) (Absent) (Major) (Major) (Minor) Organic content (Medium) Minor Minor (Medium) (Absent) (Minor) (Minor) Sediment colour Red Red Red (Red) (Red) (Red) (Red) Water flow rate (Slow) (Slow) Slow Slow Slow (Fast) (Fast) Physiography Hitly (Hilly) Mt. Mature Hilly Hilly (Mt. Youth) Mt. Youth Stream class Quaternary Quaternary Quaternary Quaternary Quaternary Quaternary (Tertiary) Drain pattern (Poor) (Poor) (Herringbone) Poor (Poor) (Poor) Poor Bank type (Alluvial) (Alluvial) (Colluvial) (Colluvial) (Alluvial) (Alluvial) Alluvial Unbracketed categories are significantly different (95% confidence level) from at least one other category for each variable. Cate-gories within brackets were not found to be significantly different from other categories. Table 3.8 Specific categories of individual categorical Held variables with lowest means, Granitic provenance Z n Pb H g C u N i M n U Finet content (Minor) ( M i n o r ) (Medium) (Minor) ( M i n o r ) ( M i n o r ) ( M i n o r ) Sand content M e d i u m M e d i u m M e d i u m (Minor) ( M i n o r ) M e d i u m M e d i u m Organic content Absent Absent Absent (Absent) ( M i n o r ) (Absent) (Absent) Sediment colour White White White White R e d White White Water flow rate Fast Fast Fast Moderate Moderate (Fast) (Fast) Physiography Mt. Y o u t h M t . Y o u t h M t . Y o u t h M t . Y o u t h (Mt. Y o u t h ) Mt. Y o u t h Hil ly Stream class Tertiary Tertiary Tertiary Quaternary Quaternary Tertiary (Tertiary) Drain pattern Poor Poor P o o r Herringbone (Poor) Poor Poor Bank type Altuvial Alluvial Alluvial Alluvial Alluvial Alluvial (Glacial) Unbracketed categories are significantly different (95% confidence level) from at least one other category for each variable. Cate-gories within brackets were not found to be significantly different from other categories. Table 3.9 Specific categories of individual categorical field variables with lowest means, Quartzitic provenance Z n Pb H g C u N i M n U Fines content M e d i u m (Major) (Major) M e d i u m M e d i u m M e d i u m ( M e d i u m ) Sand content M i n o r ( M e d i u m ) ( M e d i u m ) (Minor) ( M i n o r ) ( M e d i u m ) (Minor) Organic content (Absent) Absent Absent (Absent) ( M e d i u m ) ( M e d i u m ) (Medium) Sediment colour White White White (Black) (White) (White) (White) Water flow rate (Moderate) (Moderate) Moderate Moderate Moderate Moderate (Moderate) Physiography Mt. Y o u t h M t . Mature (Hil ly) (Mt. Mature) Mt. Y o u t h (Hil ly) (Mt. Mature) Stream class Tertiary Tertiary (Tertiary) Tertiary Tertiary Tertiary (Quaternary) Drain pattern (Herringbone) (Herringbone) (Poor) Dendritic (Herringbone) (Herringbone) Poor Bank type (Glacial) (Colluvial) (Alluvial) (Alluvial) (Colluvial) (Glacial) Glacial Unbracketed categories are significantly different (95% confidence level) from at least one other category for each variable. Cate* gories within brackets were not found to be significantly different from other categories. 59 Table 3.10 Nonparametric correlation matrix. Granitic provenance (italicized r-values are significant at 95% confidence level, N = 485) Fines Sand Organic Water Physiography Stream content content content flow rate class Fines content — Sand content —0.79 — Organic content —0.09 —0.95 — Water flow rate —0.05 0.22 —0.41 — Physiography 0.24 0.19 0.11 0.11 — Stream class 0.4 7 —0.29 0.13 0.11 0.08 60 role i s . It i s apparent that composition of a p a r t i c u l a r sample w i l l r e f l e c t the interactions and r e l a t i v e strengths of many factors which, by rei n f o r c i n g or counteracting each other, influence metal content associated with a p a r t i c u l a r bedrock u n i t . However, many of the environmental factors are interdependent; for example, a non-parametric correlation (Table 3.10) c l e a r l y indicates the i n t e r r e l a t i o n s h i p s between stream c l a s s , f i n e s , sand and organic content, and waterflow rate for sediments derived from granites. Consequently, groups of factors, acting c o l l e c t i v e l y , become more important than any single factor in both characterizing the sediment and sample s i t e , and in determining i t s metal content. CONCLUSIONS As far as we are aware, this is the f i r s t systematic attempt to evaluate the sig n i f i c a n c e of f i e l d observations in rela t i o n to background v a r i a t i o n s in metal contents of drainage sediments. We conclude: ( 1 ) Despite their subjective character f i e l d observations can related to s i g n i f i c a n t variations in metal content of drainage sediments associated with a single rock u n i t . (2) Variations in metal content of drainage sediments observed for most ca t e g o r i c a l f i e l d parameters are consis with well known controls on the behaviour of the trace 61 elements. (3) Results indicate that groups of environmental factors, acting c o l l e c t i v e l y are more important than any single factor characterizing the sediment and sample s i t e , and in determining metal content. (4) Results provide an objective c r i t e r i a for assessing s u s c e p t i b i l i t y of trace element dispersion to environment factors and deciding which factor(s) warrant recording in future surveys. (5) The procedure presented, enables the significance of f i e l d observations to be systematically related to var i a t i o n s in background metal content of sediments and provides a guide to the re l a t i v e strength or consistency of the influence of the various f i e l d parameters on trac element composition. (6) The procedure also provides a means of characterizing those s u r f i c i a l environments favouring abnormally high or low metal contents. ACKNOWLEDGEMENTS The authors would l i k e to thank the B r i t i s h Columbia 62 Ministry of Energy, Mines and Petroleum Resources for providing f i n a n c i a l support. We are e s p e c i a l l y grateful to A. Bentzen for his valuable assistance with the computer work. IV. RECOGNITION AND RANKING PROCEDURE FOR MULTI-ELEMENT REGIONAL STREAM SEDIMENT SURVEYS (MAP-AREA 82F), BRITISH COLUMBIA 64 ABSTRACT A systematic, computer-oriented method of recognizing and ranking anomalous samples from geochemical data sets i s described. The technique employs a multivariate approach (multiple regression) to identify and reduce the e f f e c t s of background variations by re-expressing background multi-element data in metal associations (background regression models). These models are subsequently applied to sample subsets selected on factor/s (environmental controls) known to strongly influence geochemical r e s u l t s . B r i e f l y the method involves: 1. Sorting of data into environmental control groups (usually rocktype) 2. Threshold se l e c t i o n using the method of S i n c l a i r (1976) to i s o l a t e anomalous samples from background samples. 3. Selection of one or more elements to serve as focus of the study. 4. Backward stepwise regression of each environemtal group to develop background models for selected metal in terms of other elements. 5. Ranking i n d i v i d u a l samples in terms of the regression model and threshold. Multielement data obtained from a regional stream sediment survey undetertaken by the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources in S.E. B r i t i s h Columbia (map-65 area B2F) were evaluated. Samples were grouped i n i t i a l l y as to dominant rocktype in the provenance region. Background multivariate models for zinc in terms of other log-transformed elements (Cu, Pb, Ni, Co, Fe, Mn, Hg, Mo and W) were obtained for each of the seven selected provenance groups. Individual samples were subsequently characterized and ranked after comparison with relevant threshold l e v e l s and background multi-element models. From a survey draining 1259 sample s i t e s , 115 samples anomalous for zinc were i d e n t i f i e d and characterized. 43 of these samples contained zinc concentrations greater than i t s determined threshold value, while 72 of these anomalous samples were only i d e n t i f i e d because their i n d i v i d u a l metal associations were s i g n i f i c a n t l y d i f f e r e n t than t h e i r relevant provenance background model. The method provides a means to reduce the effects of background variations while simultaneously i d e n t i f y i n g and characterizing anomalous samples. The technique promotes recovery of geological information and i s applicable where geochemical data are complexly inter e l a t e d and the geological features within the survey area can be characterized by diagnostic metal assoc i a t ions. 66 INTRODUCTION Exploration geochemistry u t i l i z e s geochemical patterns, that i s natural v a r i a t i o n , of elements, as a guide to the presence or absence of ore mineralization. T r a d i t i o n a l l y , interpretation of the bulk of exploration geochemical data has been based on the evaluation of single element d i s t r i b u t i o n s . This approach, however, frequently proves inadequate since metal d i s t r i b u t i o n s are often characterized by multiple populations of which presence ore or p o t e n t i a l ore i s only one of a host of genetic factors that may play a part in the development of the o v e r a l l geochemical pattern. Consequently, a premium i s placed upon interpretative techniques which can i d e n t i f y and reduce the influence of l o c a l geochemical patterns and secondly sort out a l l samples of background character. A number of previous investigations (Rose, Dalberg and Keith, 1970; Rose and Suhr, 1971; Closs and Nichol, 1975; and Austria and Chork, 1976) have indicated the p o t e n t i a l of various i n t e r p r e t a t i o n a l procedures as aids in i d e n t i f y i n g background and anomalous populations in geochemical data. These studies commonly employed a multivariate approach to i d e n t i f y and reduce background variations by re-expressing the entire multi-element geochemical data set in metal associations (multi-element background models) which characterized general geologic, geochemical and environmental processes. Unfortunatley, many of these studies included obviously anomalous samples into t h e i r regression analyses, tending to compromise the v a l i d i t y of their multi-67 element background models. In order to overcome t h i s l i m i t a t i o n , a systematic computer-oriented method of recognizing and ranking anomalous samples was developed. Our detailed procedure u t i l i z e s the type and q u a l i t y of data incorporated in various regional programs undertaken by the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources (Fig. 4.1) but can be adapted e a s i l y for data for other programs. Regional multi-element stream sediment surveys of the type c a r r i e d out in B r i t i s h Columbia under terms of the Uranium Reconnaissance Program contain coded information on the p r i n c i p a l rock unit forming the provenance region of each sample. Rock-type coding for t h i s purpose i s never perfect: some basins may be miscoded, perhaps because of the scale of geological base maps a v a i l a b l e . In any case i t i s apparent that some apparently anomalous metal concentrations a r i s e from incorrect assignment of the dominant rock-type or from mixing of sediments derived from several rock types. Consequently, the following procedure for determining multi-element background models i s intended to be applied to individual samples so that i t s anomalous nature can be assessed for a variety of l i t h o l o g i e s hosting the drainage basin. GENERAL METHODOLOGY In b r i e f , our general approach to the recognition and ranking of anomalous samples involves the following steps: F i g u r e ^4 L o c a t i o n map o f s t u d y . a r e a 82F, and i n d e x map o f r e g i o n a l g e o c h e m i c a l s u r v e y s a v a i l a b l e i n B r i t i s h Columbia ( M c M i l l a n , 1982). 69 1. Sorting of data into provenance groups, that i s , predominant rock type in drainage basin above the sample s i t e 2. Evaluation of simple s t a t i s t i c s and pro b a b i l i t y graphs for each element in each provenance group. 3. Threshold selection using the method of S i n c l a i r (1976) to i s o l a t e anomalous samples from background samples. 4. Selection of one or more elements to serve as the focus of the study (for example, z i n c ) . 5. Backward stepwise regression of each provenance group to develop background models for zinc in terms of other elements. 6. Ranking individual samples in terms of (a) their contamination code and (b) the regression model and threshold. 7. Output of sample information in a manner convenient for p r a c t i c a l use in follow-up examination. SORTING INTO PROVENANCE GROUPS The basic assumption here is that for most samples the abundant rock type in the provenance area w i l l exert a dominant control on the chemical character of derivative stream sediments. As a r e s u l t data for each provenance group was dealt with separately. Means and standard deviations of a l l raw and log-transformed metal abundances provide insight into l e v e l s of abundance, dispersion, and general aspect of population densities (histogram). Correlation c o e f f i c i e n t s indicate metal associations of geological importance (for example, S i n c l a i r and ^.:i"K,T^es9ari-,. "1980). If only background values are considered, these 70 associations commonly r e f l e c t differences in background environments and are not related to anomalous samples. THRESHOLD SELECTION Separation of background and anomalous samples i s ess e n t i a l to our method because i t leads d i r e c t l y to s t a t i s t i c a l models for background metal abundances. Samples with unusual values in any metal w i l l tend to have an unfortunately strong influence on correlations and other multivariate s t a t i s t i c s . Consequently, the method of threshold recognition i s important. We have adopted the p r o b a b i l i t y graph approach of S i n c l a i r (1976) because t h i s procedure i s systematic and has been shown by numerous examples to provide e f f e c t i v e thresholds for many types of geochemical data. ELEMENT SELECTION We must decide which element or elements are of d i r e c t concern to our search problem. Are we interested in s i l v e r -lead-zinc, copper-molybdenum, tungsten-uranium, or others? Of course, we may want to investigate many associations of the sort l i s t e d , but in our approach each association would be dealt with separately. Within a p a r t i c u l a r metal association i t may not be necessary to deal thoroughly with a l l elements because some may be r-edundant, others may not show adequate geochemical contrast, 71 and s t i l l others may present l i m i t a t i o n s r e s u l t i n g from a n a l y t i c a l problems. In our case we w i l l use zinc data as a basis for evaluating regional s i l t samples in terms of s i l v e r -lead-zinc associations t y p i c a l of our study area (map-area 82F). MULTIVARIATE MODELLING OF BACKGROUND VALUES Multiple regression has been shown by many to be an eff e c t i v e method of demonstrating empirical relationships between a p a r t i c u l a r element (dependent variable) and a group of other elements (independent v a r i a b l e s ) . In many cases a high proportion of the v a r i a b i l i t y of the dependent variable i s explained in terms of variations in the independent variables ( S i n c l a i r and Fletcher, 1980). Where such methods are applied to background samples only, the abundance of a dependent variable (for example, zinc) can be expressed as a linear combination of the abundances (or logarithms of abundances) of many other elements to provide a multivariate background model. It should however, be emphasized that in order for the independent variables or predictor elements to be e f f i c i e n t predictors of the expected background l e v e l s of the dependent variable (element), the i n i t i a l data should preferably be chosen only to describe the normal range of geochemical background var i a t i o n . If numerous anomalous samples are included the regression l i n e w i l l be pulled away from the trend of the \*v.v->^5^e*gr4>und samples into a compromise pos i t i o n , and some anomalous samples w i l l not be recognizable in terms of their 72 deviations from t h i s trend. We have experimented with two approaches to the se l e c t i o n of samples used to e s t a b l i s h a multiple regression model. In our f i r s t attempts sample selection was based on the dependent variable for a single provenance group with only those values below the threshold (based on p r o b a b i l i t y graphs) being selected. In a later refinement we edited the data base for a Single provenance group by omitting samples that were also obviously anomalous with respect to any of the independent variables. The s p e c i f i c method we use for multivariate background modelling i s backward, stepwise regression which starts with a l l independent variables in the data base and sequentially drops those that make no s t a t i s t i c a l l y s i g n i f i c a n t contribution to explaining the v a r i a b i l i t y of the dependent v a r i a b l e . Eventually a point i s reached where a l l remaining variables are s t a t i s t i c a l l y s i g n i f i c a n t (at the 0.05 l e v e l , for example) and an equation i s obtained of the form: Log (Zn) = B 0 + B, lo g ( x 3 ) + B» l o g ( x B ) + B s log(x 5) where B's are constants and x's are abundances of metal i . RECOGNITION AND RANKING OF ANOMALOUS SAMPLES For each sample we determine a series of ranks from 0 to 3 by comparing the observed value of the dependent variable with „^':^t^r.^3r^lues calculated by each of the provenance group 73 multivariate models. Significance of the rank numbers i s shown on Figure 4.2. We then c a l c u l a t e a 4 - d i g i t ranking code for each sample where the f i r s t d i g i t i s the number of rock types for which rank 3 was obtained, the second d i g i t i s the number of rock types for which rank 2 was obtained, and so on. If there are seven rock types a l l with very high zinc values (rank 3) the ranking code would be 7000; in another case rank might be (3) for two rock types, (2) for three rock types, (1) for two rock types, and (0) for one rock type to give a ranking code of 2331. The main advantage of t h i s procedure i s as a refinemnet in the selection of anomalous values r e l a t i n g to the p r o b a b i l i t y graph procedure and the assigning of r e l a t i v e p r i o r i t i e s to anomalous samples. Values above t i (Figure 4.2) are recognized as being anomalous without the aid of multiple regression. In addition, however, values below t1 that depart substanially from the expectation according to a multiple regression model (1 and 2 on Figure 4.2) are also out of the ordinary and warrant examination. In p a r t i c u l a r , we are interested in those values below t1 that are much higher than the corresponding calculated values. Such samples are anomalous in one element, r e l a t i v e to a l i n e a r combination of other elements. On Figure 4.2 the suggestion i s made graphically that samples are anomalous i f observed values are more than two standard errors greater than values calculated according to the multiple regression model. OUTPUT PROCEDURES 74 METAL (log ppm calc.) F i g u r e 4.2 Sample ra n k i n g i n r e l a t i o n t o f i e l d s bn a p l o t o f observed v a l u e versus a v a l u e c a l c u l a t e d from a m u l t i v a r i a t e model. 75 We have designed an output system by which samples can be ordered in terms of decreasing p r i o r i t y for follow-up exploration. A l l anomalous samples recognized by the foregoing procedures are ranked according to the estimated l i k e l i h o o d of sample contamination from such factors as known mines, man-made metallic features, or f e r t i l i z e r , on a scale of 0 to 3. Our f i r s t rank of anomalous samples i s based on t h i s coded parameter, zero contamination being of most inte r e s t . Within this group we code a sample for each background model as 3, 2, 1, or 0 as described previously and a 4-digit ranking code i s used to l i s t samples within each contamination group in order of decreasing ranking code. Locations for each sample are l i s t e d as i s the observed abundance of the dependent variable and the sample number. These items are arranged in a manner as to promote e f f i c i e n c y of evaluation of each sample. In addition, we use plot locations of anomalous samples with th e i r i d e n t i f i c a t i o n number and ranking code. CASE HISTORY (MAP AREA 82F) The data base for the study comprises f i e l d observations and a n a l y t i c a l r e s u l t s (minus 80-mesh, Zn, Cu, Pb, Ni, Co, Mn, Fe, Mo, Hg and U) for drainage sediments c o l l e c t e d at 1313 s i t e s in S.E. B r i t i s h Columbia (Figure 4.1) as part of the Joint Federal-Provincal Uranium Reconnaissance Program. Details of sampling and a n a l y t i c a l procedures are to be found in Open F i l e 514 of the Department of Energy, Mines and Resources, Canada and 76 B r i t i s h -Columbia Ministry of Energy, Mines and Petrolem Resources (1977). These data are available p u b l i c a l l y on magnetic tape from the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources, Geological Branch, Mineral Resources D i v i s i o n , V i c t o r i a . Samples were grouped i n i t i a l l y on the basis of coding as to dominant rock type in the provenance region. Data for each element in each provenance group were examined as a p r o b a b i l i t y graph and a threshold selected separating two populations (presumably anomalous and background) using the method of S i n c l a i r (1976). We chose to examine zinc as the dependent variable described here because of the association s i l v e r - l e a d -zinc in known vein deposits in the area. Background multivariate models for zinc in terms of other elements were obtained for each of the seven provenance groups for which we have adequate samples. Three of these models are summarized in Table 4.1 to i l l u s t r a t e the type of re s u l t s obtained. S t a t i s t i c s for a l l seven provenance group models for zinc are given in Table 4.2 to i l l u s t r a t e the s t a t i s t i c a l q u a l i t y of the background models. A l l samples coded in one the seven provenance groups for which we could calculate background models were treated by each of the background equations separately. The calculated zinc background according to a given model was then compared with expectations for that model so that for each background model a sample received a ranking from 0 to 3 i n c l u s i v e (compare Figure m4-~2K . In our case each sample was ranked seven times, once for 77 T a b l e 4.1 Examples of m u l t i v a r i a t e r e g r e s s i o n background models f o r z i n c , Map a r e a 82F. GRANITE QUARTZITE log (Zn) -SCHIST lo g (Zn) 0.4726 + 0.0713 log CCu) + 0.3994 log (Mn) R 2 cs 0.62 S e EE 0.1277 n m 393 1.1020 0.2721 log (Pb) + 0.1412 log (Mo) R 2 as 0.74 S e cs . 0.0915 n ** 287 0.8392 + 0.4100 log (Pb) + 0.2412 log (W) R 2 CK- 0.76 S e CC 0.1008 n CX 27 78 PROVENANCE GROUP GRNT QRTZ SLTE ANDS ARGL GNSS SCST n 393 287 100 57 56 53 27 R .79 .86 .84 .84 .92 .85 .87 R 2 .62 .74 .70 .70 .85 .71 .76 .1277 .0915 .1184 .1031 .0812 .0940 .1008 Table 4.2 Summary s t a t i s t i c s f o r m u l t i v a r i a t e background z i n c models f o r seven provenance groups, Map area 82F. 79 each model. These r a n k i n g s were accumulated i n t o a s i n g l e r a n k i n g code. Samples r e c o g n i z e d as anomalous or p o t e n t i a l l y anomalous were d i v i d e d i n t o t h r e e c o n t a m i n a t i o n c l a s s e s w i t h p r i o r i t y d e c r e a s i n g as c e r t a i n t y of c o n t a m i n a t i o n i n c r e a s e s . F o r each c o n t a m i n a t i o n c a t e g o r y samples a r e r a n k e d a c c o r d i n g t o d e c r e a s i n g numeric v a l u e of the r a n k i n g code. An example i s shown i n T a b l e 4.3, where a s m a l l p a r t of t h e O - c o n t a m i n a t i o n c a t e g o r y i s l i s t e d . From a t o t a l of 1259 samples, t h i s p r o c e d u r e produced 115 anomalous samples i n the O - c o n t a m i n a t i o n c a t e g o r y . F o r t y t h r e e of t h e s e samples had z i n c c o n c e n t r a t i o n s g r e a t e r than i t s s i m p l e t h r e s h o l d , c a l c u l a t e d f o r i t s o b s e r v e d provenance group, w h i l e t h e r e m a i n i n g 72 samples were i d e n t i f i e d s o l e l y on the b a s i s o f h a v i n g s i g n i f i c a n t l y d i f f e r e n t m e t a l a s s o c i a t i o n s than t h e i r p a r t i c u l a r provenance m u l t i v a r i a t e background model. Our p r o c e d u r e i s t o l i s t t h e s e samples i n t a b u l a r form i n Table 4.3 and t o produce computer-drawn p l o t s of anomalous sample l o c a t i o n s as i l l u s t r a t e d on F i g u r e 4.3. In a d d i t i o n t o r a n k i n g i n f o r m a t i o n , o r i g i n a l raw d a t a , and c o o r d i n a t e s , t h e output t a b l e c o n t a i n s a s i m p l e c o n s e c u t i v e numeric i d e n t i f i e r used f o r c l a r i t y on t h e map output and p e r m i t t i n g easy combined use of the t a b u l a t e d d a t a and t h e o u t p u t map. The o u t p u t map i s of p a r t i c u l a r use because i t i d e n t i f i e s t h e most o b v i o u s anomalous samples ( f o r example 7000) from those t h a t might escape d e t e c t i o n ( f o r example, 0520). The s c a l e of t h e l o c a t i o n p l o t s h o u l d be i d e n t i c a l t o g e o l o g i c a l base maps of t h e a r e a so t h e two can s t u d i e d t o g e t h e r w i t h o u t Table 4.3 Part of a table l i s t i n g anomalous samples i n order of decreased ranking code UTME UTMN GNSS ORTZ GRNT SLTE ANOS SCST ARGL TOTALS (3210) ZN (PPM) ROCK TYPE CONTAM URPS (8: 10 0 54 558659 5532787 3 3 3 2 2 2 2 3400 181 GRNT 0 775229 55 538165 5479004 3 3 2 2 0 2 0 2302 177 GRNT 0 7731 16 56 548189 5536835 3 3 2 2 0 0 2 2302 167 ORTZ 0 775248 57 518892 5514798 3 3 0 2 2 0 0 2203 177 ORTZ 0 775391 58 525483 5477274 3 3 0 2 0 0 2 2203 164 ORTZ 0 779153 59 444220 5512900 3 3 0 0 0 0 0 2005 167 GNSS 0 771294 eo 483724 5474162 0 2 2 2 2 2 2 601 123 ANOS 0 779063 61 489795 5497195 2 2 2 2 0 2 2 601 89 SLTE 0 7772 15 62 447 145 5447 184 2 2 2 2 0. 2 2 601 79 GNSS 0 775011 63 51152e 5479314 2 2 2 2 0 2 2 601 77 GRNT 0 779204 64 479680 5464101 0 2 0 2 2 2 2 502 132 ARGL 0 771091 65 501640 5430728 0 2 0 2 2 2 2 502 69 ANOS 0 777078 66 510994 5485659 2 2 0 0 0 2 2 403 125 ORTZ 0 773211 67 485042 5493023 2 2 2 0 0 2 0 403 127 GRNT 0 777234 68 466702 55102 12 2 2 2 0 0 2 0 403 123 GNSS O 771328 69 512650 5440330 0 2 0 2 0 2 2 403 46 ORTZ 0 775264 70 568389 5428516 0 2 0 2 0 2 2 403 47 ORTZ 0 773096 71 537227 5485766 2 2 0 0 0 0 2 304 141 ORTZ 0 775385 72 511127 548724 1 2 2 0 0 0 2 0 304 121 ORTZ 0 773214 81 Figure 4.3 Plot of an area of anomalous samples frcm map-area 82F 6100^ (49) * _0403 •(61) 0403 ><67) .6100 (44) 7000 (38) • 7000 .7000 .(36) 14300 (52) „0106 (83) 117°0o' W 49°30' N 0403• (66) .0204 (72) .0601 (63) 0601 * (60) 0502 (64) 0106 (84) 0106 •(86) 4300 • (51) (7000 (30) 7000 * (26) 7000 (37) *7000 7,°°?« 7000 (40£ (23) (34) ,7000 (4) 7000» 7000 7 ( (21) (19) n •7000 .7000 (18) #(15) 0 5 10 Kilometres 82 ambiguity. We tested s e n s i t i v i t y of the regression procedure for determining a multivariate background for zinc by establishing such models based on two t r a i n i n g sets: (1) a l l samples indicated as having background zinc values, and (2) the same data set minus any samples that appeared to be anomalous in any element other than z i n c . Tables 4.1 and 4.2 are based e n t i r e l y on the second t r a i n i n g set. Figures 4.4 and 4.5 i l l u s t r a t e the contrasting results obtained in background d e f i n i t i o n . It is clear that the 'cleaner' data set (number 2 previously) leads to a better multiple regression r e l a t i o n s h i p , that i s , with less scatter of calculated and observed values. The problem with using the second t r a i n i n g set i s that more work i s required to set i t up and more samples w i l l be included in the anomalous category. DISCUSSION The methodology described here would appear to have a wide range of applications to geochemical data evaluation, perhaps with minor modifications to s u i t p a r t i c u l a r data sets. For example, many geochemical surveys may not record the l i k e l i h o o d that a sample i s contaminated, and t h i s l e v e l of ranking might have to be omitted. The precise l i m i t s to the coding regions i l l u s t r a t e d on Figure 4.2 can be changed to suit a p a r t i c u l a r bias to anomaly sel e c t i o n , r e s u l t i n g in a s l i g h t l y d i f f e r e n t l i s t i n g of anomalous samples. One of the serious problems i s the question of i n t i a l 83 Log Zn (calc.) Figure 4.4 Observed versus calculated zinc values for provenance group 'ARGL' ( a r g i l l i t e ) , map-area 82F; calculated values based on a model determined from a l l samples with background zinc values. 84 co — O CD O Log Zn (calc.) Figure 4.5 Observed versus calculated zinc values f o r provenance group *ARGL' ( a r g i l l i t e ) , map-area 82F; calculated values based on a model determined from those samples wit h zinc background values that also are not anomalous i n any other element (that i s , a 'cleaner' subset of the data used i n Figure 4.4). 85 grouping of data on the basis of dominant rock type that underlies the drainage basin of each sample, a c l a s s i f i c a t i o n which i s fundamental to our procedures. A substantial amount of e f f o r t i s required to code t h i s rock-type information even i f the data are av a i l a b l e . If rock type has not been coded i t may be necessary to use some less s a t i s f a c t o r y method of approximation of background geology for each sample. In some environments, of course, some other parameter may be more useful than rock type for grouping data. CONCLUSIONS A method of anomaly selection and ranking for multi-element regional stream sediment data has been described. The procedure of f e r s the following advantages: ( 1 ) The method i s rigorous in making use of established s t a t i s t i c a l method for treating geochemical data such as a p r o b a b i l i t y graph analysis and backward stepwise regression. (2) The procedure i s computer based and i s rapid and thorough. (3) The methodology ensures that some anomalous values which are not obvious (that i s , are not higher than a simple threshold) w i l l be recognized. (4) A novel ranking procedure i s described that assigns r e l a t i v e p r i o r i t i e s to samples for further i n v e s t i g a t i o n . Details of the ranking procedure are subjective but a system for ranking codes c l e a r l y describes the manner in which a sample i s anomalous. (5) Because samples are tested against every rock type, the procedure incorporates an evaluation as to whether other rock types might be contributing to the provenance area of a p a r t i c u l a r sample. Possible additional rock types are i d e n t i f i e d and can be compared with available geological maps. (6) The method provides a means to reduce the e f f e c t s of background v a r i a t i o n s , while simultaneously i d e n t i f y i n g and characterizing anomalous samples. Consideration should be given to i t s implementation in a l l geochemical surveys where the data are complexly i n t e r r e l a t e d and the geological features within the survey area can be characterized by diagnostic metal associations. ACKNOWLEDEMENTS This project has been funded by the B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources. 88 89 A suite of systematic, computer-assisted, s t a t i s t i c a l and graphical procedures have been developed and applied to data from published government stream sediment surveys in south-eastern B r i t i s h Columbia and the geochemical interpretation of the data analysis had been presented. Results show that s t a t i s t i c a l analysis of f i e l d - s i t e duplicate samples provide a rapid an e f f e c t i v e means of interpreting geochemical trends, while simultaneously allowing evaluation of the adequacy of sampling and laboratory analysis. Government s i l t data appear to be of very high q u a l i t y and are characterized by low precision estimates (6%) at the 95th percentile concentration value for metals Zn, Cu, Ni, Co, Fe and Mn, and a combined v a r i a b i l i t y due to l o c a l and procedural error of less than 5% of the t o t a l e r r o r . Metal concentrations between drainage sample s i t e s r e f l e c t real trends related to geological and geochemical features. Results presented are more in accord with s p l i t s of an i n d i v i d u a l sample (a n a l y t i c a l duplicate) rather than analyses of two separate f i e l d samples. Relationships between background metal concentrations and f i e l d observations of the drainage catchment, sample s i t e and sediment sample commonly recorded in geochemical surveys were examined systematically for the f i r s t time. We conclude: 1. Despite t h e i r subjective character f i e l d observations can be related systematically to 90 s i g n i f i c a n t v a r i a t i o n s in metal content of drainage sediments associated with a s i n g l e rock unit. 2. Variations in metal content of drainage sediments observed for most categorical f i e l d parameters are consistent with well known controls on the behaviour of the trace elements 3. Results indicate that groups of environmental factors, acting c o l l e c t i v e l y are more important than any single factor characterizing the sediment and sample s i t e , and in determining metal content. 4. Results provide an objective c r i t e r i a for assessing s u s c e p t a b i l i t y of trace element dispersion to environmental factors and deciding which factor(s) warrant recording in future surveys. A method of anomaly selection and ranking for multi-element regional stream sediment data has been described. The procedure allows the geochemical character of individual s i l t samples draining various provenance regions to be screened against relevant threshold l e v e l s and background multi-element models, so as to insure that some anomalous samples which are not obvious (that i s , are not higher than a simple threshold) are recognized. Because samples are tested against every rock type the procedure incorporates an evaluation as to whether other rock types might be contributing to the provenance area of a p a r t i c u l a r sample. Possible a d d i t i o n a l rock types are i d e n t i f i e d and can be compared with a v a i l a b l e geological maps. The method provides a means to reduce the effects of background v a r i a t i o n s while simultaneously i d e n t i f y i n g and characterizing anomalous samples. Consideration should be given to i t s implementation in a l l geochemical surveys where the data are •scoiEpI«xiy i n t e r r e l a t e d and the geological features within the 91 search area can be characterized by diagnostic metal associations. In conclusion, i t i s believed that the adoption of such procedures as outlined here w i l l provide . rapid and promote maximum recovery of geochemical information from geochemical data sets. 92 BIBLIOGRAPHY Austria, V. and Chork, C.Y., 1976. A study of the application of regression analysis for trace element data from stream sediments in New Brunswick. J . Geochem Explor., 6: 211-232. Bolviken, B. and Sinding-Larsen, R., 1973. Total error in the interpretation of stream sediment data. In M.J. Jones (Editor) Geochemical Exploration 1972. I n s t i t u t e of Mining and Metallurgy, London, pp. 285-293. B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources., 1981. Regional Stream Sediment and Water Geochemical Reconnaissance Data, Map-sheet 92H, 921 and 92J. Brummer, J.J., Wallis, R.H., and Gleeson, C.F., 1978. Geological implications of regional stream sediment geochemical data from south-central B r i t i s h Columbia. Can. Inst. Min. Metall., 68: 68-83. Chork, C.Y., 1977. Seasonal, sampling and a n a l y t i c a l variations in stream sediment surveys. J . Geochem. Explor., 7: 31-47 Closs, L.G. and Nichol, I., 1975. The role of factor and regression analysis in the inte r p r e t a t i o n of geochemical reconnaissance data. Can. Jour. Earth S c i . , 12: 1316-1 330. Department of Energy Mines and Resources and B.C. Ministry of Energy, Mines and Petroleum Resources.,1977. Regional Stream Sediment and Water Geochemical Reconnaisance Data, Southeastern B r i t i s h Columbia, Geological Survey of Canada Open F i l e Report OF514. Duncan, D.B., 1955. Multiple Range and Multiple F Test. Biometrics, 11: 1-42. Duncan, D.B., 1957. Multiple Range Tests for correlated and heteroscedastic means. Biometrics, 13: 164-176. Doyle, P.J. and Fletcher, W.K., 1979. Regional geochemical mapping in areas lacking surface drainages: Cu, Fe, Mn, and Zn content of overburden and s o i l in south-central Saskatchewan. Can. Jour. Earth S c i . , 16: 1086-1093. Fletcher, W.K., 1981. A n a l y t i c a l methods in geochemical prospecting. Handbook of exploration geochemistry. Volume 1. E l s i e v e r , 255 pp. Garrett, R.G., 1969. The determination of sampling and 93 a n a l y t i c a l errors in exploration geochemistry. Econ. Geol., 64: 568-569. Garrett, R.G., 1973. The determination of sampling and a n a l y t i c a l errors in exploration geochemistry - a reply. Econ. Geol., 68: 282-283. Howarth, R.J. and Lowenstein, P.L., 1971. Sampling v a r i a b i l i t y of stream sediments in broad-scale regional geochemical reconnaissance. Trans. Inst. Min. Metall., 80: B363-B372. Koch, G.S. and Link, R.F., 1971. S t a t i s t i c a l analysis of Geological Data. John Wiley and Sons, New York, N.Y,. Vol. 1, 375 pp., Vol 2, 417 pp. Krumbein, W.C. and G r a y b i l l , F.A., 1965. An Introduction of S t a t i s t i c a l Models in Geology. McGraw-Hill, New York, N. Y., 413 pp. Le, Chinh., 1980. UBC Duncan. Computing Centre documentation, University of B r i t i s h Columbia Computing Centre. L i t t l e , H.W., 1960. Nelson map-area, west-half, B r i t i s h Columbia (82FW). Geol. Surv. Can., Memoir 308, 205 pp. Matysek, P.F., Fletcher, W.K., S i n c l a i r , A.J. and Bentzen, A., 1981. A preliminary evaluation of categorical f i e l d observations for regional stream sediment samples (82F,K), B.C. Ministry of Energy, Mines and Pet. Res., Geological Fieldwork, 1980, Paper 1981-1: 149-158. Matysek, P.F., S i n c l a i r , A.J., and Fletcher, W.K., 1982. Rapid Anomaly Recognition for Multi-element Regional stream sediment surveys. B.C. Ministry of Energy, Mines and Pet. Res., Geological Fieldwork, 1981, Paper 1982-1: 176-186. Matysek, P.F., Fletcher, W.K., and S i n c l a i r , A.J., 1983. S t a t i s t i c a l evaluation of the s i g n i f i c a n c e of f i e l d parameters in the interpretation of regional geocchemical sediment data. J. Geochem. Explor. 19: 393-402. Matysek, P.F. and S i n c l a i r , A.J., 1984. S t a t i s t i c a l evaluation of duplicate samples, regional sediment surveys (Map-area 92H, 921 and 92J), B r i t i s h Columbia. B.C. Ministry of Enegy, Mines and Petroeleum Resources., Geological Fieldwork, 1983, Paper 1984-1: 186-196. McMillan, W.J., 1982. Regional geochemical surveys, B r i t i s h Columbia, B.C. Ministry of Energy, Mines and Pet. Res., Geological Fieldwork, 1981, Paper 1982-1: 279-271. -.•4leye-t^ .-4J-;T., Theobald, P.K. and Bloom, H., 1979. Stream sediment geochemistry in Geophysics and Geochemistry in the 94 Search for Metalic Ores. P.J. Hood, editor, Geol. Surv. Canada, Econ. Geol., Rept. 31: 411-434. Miesch, A.T., 1967. Theory of error in geochemical data. U.S. Geol. Surv., Prof. Paper 574-A, 17 pp. Miesch, A.T., 1976. Geochemical survey of Missouri - methods of sampling, laboratory analysis and s t a t i s t i c a l reduction of data. U.S. Geol. Surv., Prof. Paper 954-A, 39 pp. Plant, J . , 1971. Orientation studies on stream sediment sampling for a regional geochemical survey in northern Scotland. Trans. Inst. Min. Metall., 80: B324-B345. Plant, J ., Jeffery, K., G i l l , E. and Fage, C. 1975. The systematic determination of accuracy and precision in geochemical exploration data. J . Geochem. Explor., 4: 467-486. Rice, H.M., 1941. Nelson map area, east-half, B r i t i s h Columbia. Geol. Surv. Can., Memoir 228, 85 pp. Rose, A.W., Dahlberg, E.C., and Kieth, M.L., 1970. A multiple regression technique for adjusting background values in stream sediment geochemistry. Econ. Geol., 65: 156-165. Rose, A.W. and Suhr, N.H., 1971. Major element content as a means of allowing for background va r i a t i o n in stream sediment geochemical exploration. Toronto International Geochemical Exploration Symposium Paper: 587-593. S i n c l a i r , A.J., 1976. Applications of Probabi l i t y Graphs in Mineral Exploration. Assoc. Explor. Geochem. Spec. Vol. 4, 95 pp. S i n c l a i r , A.J., 1979. Preliminary evaluation of summary production s t a t i s t i c s and location data for vein deposits, Slocan, Ainsworth and Slocan City camps, southern B r i t i s h Columbia; in Current Research, PART B, Geol. Surv. Can., Paper 79-1B: 173-178. S i n c l a i r , A.J. and Fletcher, W.K., 1980. Evaluation procedure for geochemical data uranium reconnaissance program. B r i t i s h Columbia Ministry of Energy, Mines and Petroleum Resources, Geol. Fieldwork, 1979, Paper 1980-1: 130-141. S i n c l a i r , A.J. and Tessari, O.J., 1981. Vein geochemistry, an exploration t o o l in Keno H i l l camp, Yukon T e r r i t o r y , Canada. J. Geochem Explor., 14: 1-24. Thompson, M. and Howarth, R.J., 1973. The rapid estimation and control of precision by duplicate determinations. Analyst, S3: 153-166. Thompson, M. and Howarth, R.J., 1976. D u p l i c a t e a n a l y s i s i n geochemical p r a c t i c e (2 p a r t s ) . A n a l y s t , 101: 690-709. Thompson, M. and Howarth, R.J., 1978. A new approach to the e s t i m a t i o n of a n a l y t i c a l p r e c i s i o n . J . Geochem. Explor 9: 23-30. \ I 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0052662/manifest

Comment

Related Items