Dunn Colin E. Biogeochemistry in Mineral Exploration

Подождите немного. Документ загружается.

involving 529 Douglas-ﬁr top stems from southern British Columbia the median

concentration was 0.012 ppm Y with a maximum of 0.083 ppm Y. In Africa, leaves of

Acacia have an order of magnitude more Y than stems, with concentrations locally

exceeding 2.5 ppm Y over lake sediments that once formed part of Lake Victoria in

the North Mara area of Tanzania. Plants from the Eden Project yielded from a low

of 0.007 ppm in leaves of the ‘serendipity berry’ (Thaumatococcus) that is native to

the rain forests of West Africa, to more than 1 ppm Y in foliage from a ﬁg tree

(Ficus).

Zinc (Zn) (Fig. 9-73)

Zinc is an essential element for plants, hence it is present at concentrations well

above its detection limit of 0.1 ppm Zn, commonly 100–1000 times greater,

and analytical precision is extremely good. Some species can ‘hyperaccumulate’ Zn

resulting in concentrations sometimes exceeding 1% Zn in dry tissue. However, these

are small plants (e.g., several species of Viola, Thlaspi and Minuartia) and therefore

unlikely to be of significant use in mineral exploration other than for the ﬁeld

geologist to recognize their presence as potential geobotanical indicators of Zn min-

eralization. These species are discussed in Brooks (1983) and Reeves and Brooks

(1983). Similarly, the so-called Zn moss (the bryophyte Pohlia wahlenbergii)isa

geobotanical indicator that stands out as an iridescent green covering of mineralized

outcrops at Howard’s Pass in the Yukon.

Zinc is essential for carbohydrate and protein metabolism, therefore some var-

iability of Zn data obtained from a biogeochemical survey is related to the health of

the tree rather than subtl e changes in substrate chemistry. Consequently, in order to

delineate areas of Zn mineralization it is necessary to look for quite substantial

changes in Zn concentrations. Cadmium closely follows Zn in nature, but it is not

known to be an essential element for plant metabolism. In plants it can, therefore, be

a better pathﬁnder elem ent for Zn mineralization than Zn itself. A considerable

Y ppm

V6 NIST 1575a

Target

Mean

Std. Dev.

RSD %

0.04

0.001

No data

0.023

0.003

0.1

0.2

0.3

0.4

0.5

0.6

V6 - 30 controls among 600 samples

Concentration

Detection

Fig. 9-72 . Yttrium – Precision and accuracy on dry tissues – V6 and NIST 1575a (see Fig. 9-1).

322 Biogeochemical Behaviour of the Elements

amount of research has been undertaken on the uptake of Zn by plants, and its

movement within plants. A good review, dealing mostly with crops, is given in

Kabata-Pendias (2001).

A survey over an epithermal Au/Ag system at the 3Ts property in central British

Columbia included analysis of needles from white spruce (Picea glauca) collected

along traverses over mineralized veins. One of these traverses crossed the Au-bearing

Ted vein that comprises quartz-calcite with ﬁnely disseminated sulphide minerals.

Pyrite is dominant, but also present are chalcopyrite, sphalerite, galena, Ag-sulphides,

tellurides and sulphosalts. Figure 9-74 shows the Zn signature in the dry needles,

demonstrating a strong positive signature over the Ted vein and the ‘Adrian West’

mineralized boulders. A similar proﬁle was obtained from analysis of outer bark from

lodgepole pine, but with lower concentrations and less geochemical relief than values

from the spruce needles. Data listings for this occurrence are part of the ‘Halogen

Zn ppm

V6 NIST 1575a

Target

Mean

Std. Dev.

RSD %

1.8

V6 - 30 controls among 600 samples

Concentration

Detection

Fig. 9-73. Zinc – Precision and accuracy on dry tissues – V6 and NIST 1575a (see Fig. 9-1).

Zn ppm

100

120

140

200 m

Adrian West

(boulders)

Ted

Vein

Fig. 9-74. Zinc in dry spruce needles (Picea glauca) along transect over the Ted epithermal

Au/Ag/base metal vein. 3Ts property (Silver Quest Ltd.), central British Columbia (Dunn

et al., 2006a,b). Additional details with ‘halogen study’ on the CD.

323Biogeochemistry in Mineral Exploration

Project’ for which digital data listings and a full account are contained on the CD that

accompanies this book.

Trees that are renowned for accumulating high levels of Zn are birch and willow.

Their bark, leaves and twigs commonly have background level s in excess of 100 ppm

Zn in dry tissue. By contrast spruce, pine and Douglas-ﬁr outer bark normally have

30–50 ppm Zn, with 100 ppm being an unusually high level. Outer bark from larch

has lower concentrations with 15–25 ppm as a common range of background values.

Twigs of all of these conifers have similar Zn concentrations to their respective outer

bark. Needles of Douglas-ﬁr contain lower Zn concentrations than twigs (typically

20–30 ppm Zn compared to 50 ppm for twigs). Foliage from subalpine ﬁr typically

has 50–60 ppm Zn, whereas western redcedar has only 10–15 ppm Zn.

Samples of foliage from the Eden Project returned concentrations of more

than 100 ppm Zn in seven species, with a maximum of 455 ppm Zn in Impatiens

demonstrating its capacity to scavenge Zn from the substrate. This enrichment is

in accord with the ﬁndings of Tiagi and Aery (1982) who recognized Impatiens

balsamina as a geobotanical indicator of Zn-enriched mine tailings in India.

The lowest concentration of Zn in the Eden Project samples was 13 ppm Zn in

foliage of Acacia. This is in accord with foliage from ﬁeld samples of various Acacia

species from Africa and Australia. The Acacia branches, bark and roots that have

been tested have all contained lower concentration than the foliage. However, Acacia

seems to be sensitive to the presence of mineralization with concentrations in

Zn-enriched areas substantially higher than the values of 5–20 ppm common at

background sites.

In the vicinity of the former Sullivan Pb–Zn mine in British Columb ia alder

(Alnus sinuata) twigs provided the medium with the greatest range in concentration

between background sites (approximately 40 ppm Zn in dry twigs) to sites over

mineralization where there was over 1000 ppm Zn (Dunn, 2000).

Zirconium (Zr) (Fig. 9-75)

Precision for Zr is generally good and concentrations, although quite low, are

usually above the detection limit of 0.01 ppm Zr. The availability of Zr to plants is

low so concentrations in dry tissues are rarely above 1 ppm Zr and are usually below

0.1 ppm Zr. Any high values reported should be viewed with the suspicion that there

may well have been some airborne dust contamination in the ﬁeld, since contam-

ination in the laboratory or analytical artefacts are unlikely unless a high Zr ceramic

vessel was inadvertently used for grinding the tissues.

Eden Project foliage samples show a range from 0.02 to 0.7 ppm Zr. Approximately

70% of the samples yielded concentrations below 0.2 ppm Zr. Ferns appear to have a

propensity to be able to accumulate small amounts of Zr.

A study that included the analysis of 148 samples of dwarf birch leaves from the

tundra of Canada yielded a mean concentration of 0.03 ppm Zr with a maximum of

324

Biogeochemical Behaviour of the Elements

0.08 ppm Zr. Acacia from southern and central Africa has more Zr in foliage than

stems, with concentrations up to 4 ppm Zr in foliage and 0.6 ppm Zr in stems, although

most stem samples yielded less than 0.1 ppm Zr. Zirconium seems to play no significant

role in biogeochemical exploration for minerals other than to provide an indicator of

possible airborne dust contamination. Black spruce from a remote area of northern

Saskatchewan has demonstrated that twigs (up to 0.17 ppm Zr) have higher Zr con-

centrations that needles (0.03 ppm Zr), and that twigs and leaves of deciduous species

(e.g., alder, Alnus crispa) have similar concentrations of 0.04 ppm Zr.

Zr ppm

V6 NIST 1575a

Target

Mean

Std. Dev.

RSD %

0.05

0.01

No data

0.03

0.01

0.05

0.1

0.15

0.2

0.25

0.3

0.35

V6 - 30 controls among 600 samples

Concentration

Detection

Fig. 9-75. Zirconium– Precision and accuracy on dry tissues – V6 and NIST 1575a (see

Fig. 9-1).

325Biogeochemistry in Mineral Exploration

This page intentionally left blank

Chapter 10

DATA HANDLING AND ANALYSIS

FIRST ESTIMATION OF THE DATA

On receipt of a spreadsheet of data in digital form, it is advisable to immediately

make a ‘working copy’ and store the original in its exact format. This allows the user

to return to the original in case subsequent manipulations of the data inadvertently

corrupt the data (e.g., misalignment of cells, columns or rows during a so rting

procedure). The ‘working copy’ needs to have cross-reference numbers and ﬁeld

parameters inserted that subsequently comprise the base document that can be sorted

and manipulated in any manner that the user feels necessary.

After following through the rigours of collecting samples from a survey area and

awaiting results from the analytical laboratory, as soon as the data are received there

is a natural tendency to quickly scan the numbers to look for anomalously high

concentrations of the commodity metals of particular interest. Two immediate

scenarios may arise.



There are no obvious anomalies and so there is a natural sense of disappointment

and the data are given little further consideration.



There are anomalous element concentrations, and it is very tempting to jump to

conclusions as to their significance.

It is prudent to step back from the data and, before any conclusion is drawn, to

systematically investigate the entire data set, starting with the controls. To reiterate

what has been said earlier, no set of samples should be submitted without inserting

controls – to establish both analytical precision (duplicate ﬁeld and laboratory

samples) and analytical accuracy (using vegetation control s of known composition).

Whereas these controls should be inserted ‘blind’ so that a completely objective

assessment of precision and accuracy can be made, many laboratories also provide

the data from their own analytical duplicates and control samples. It is useful to

interrogate all of these data in assessing the data quality.

Microsoft Excel is now a universally available and co mprehensive software

programme that can be used to meet all of the init ial requirements of assessing

quality control and sorting data. Most laboratories now report analytical data in

Excel format or in an Excel-compatible format. Since it is recommended that ‘blind’

controls be inserted within a sequence of samples, the ﬁrst step is to identify these

controls and extract them into separate spreadsheets.

A visual appraisal of data structure is an important aspect of exploratory data

analysis. By plotting data as histograms or on X–Y plots, the eye is quick to focus on

associations, outliers and trends. An examination of the dataset in the format that it

is received from the laboratory should constitute the ﬁrst assessment. Construction of

a histogram (bar chart) of each individual element is useful for identifying trends

in the analytical data (perhaps attributable to analytical instrument drift) and any

‘blockiness’ that might be attributed to changes in instrument calibration (or con-

tamination) after a block of samples is analyzed, such as an individual rack of test

tubes. Figure 10-1 gives an example of set of samples that has a suspicious-looking

blocky appearance. This proved to be an artefact of the analytical procedure, since it

was later learned that each ‘block’ of 33 samples (notably that identiﬁed by the

arrow) represented a single rack of test tubes, and that the laboratory checked

instrument calibrations after each rack and made appropriate adjustments. Most of the

single point peaks (4 ppm Cr) represent a blind control sample. These could later be

extracted and their precision and accuracy assessed separately. However, the control

sample proved to have concentrations that were more than double the average values

of the ﬁeld samples, and by looking at only the controls the changes by blocks of

samples would probably have been missed. In the case of Cr the analytical data

obtained by ICP-MS on an aqua regia digestion require considerable correction (by

the laboratory) for high inter-element interferences. Chromium data obtained by this

method are, therefore, of lower quality than for many other elements, and due con-

sideration should be given to this lack of quality during data interpretation.

Another feat ure of the data that should be looked for is any periodicity to the

element concentrations. On occasion a trend has been noted such as that shown in

Fig. 10-2.

Cr ppm

1 32 63 94 125 156 187 218 249 280 311 342 373 404 435 466 497 528 559 590 621

Sequence Number

Fig. 10-1. Analysis of a set of 630 Douglas-ﬁr twig samples showing a blocky appearance in

the data for Cr. Individual peaks are mostly the blind control samples.

328 Data Handling and Analysis

In the situation shown in Fig. 10-2 plant ash samples were submitted for analysis

with standards inserted as blind controls after the 10th, 30th, 50th and 70th samples in

the data sequence. However, the standard contained a similar As concentration

(1 ppm) to most of the ﬁeld samples and so on the ﬁrst pass no analytical abnormalities

were found. In this situation data were plotted without thorough preliminary inves-

tigation of data quality, and the locations of the anomalies made no geological sense.

On re-examination of the data, as the histogram shown in Fig. 10-2, it soon became

evident that the peaks in As enrichments occurred on a regular cycle of every 18th

sample, and that values decreased over the following few samples. The laboratory had

not reported their own quality control data, and so they were contacted to discuss

these trends. It turned out that they had inserted their control at every 18th site and the

control they used was material that originated from As-rich tailings! Consequently,

because their analyst did not thoroughly purge the pipette used to extract the solution

from each test tube after each analysis, there was some contamination transferred to

the subsequent sample which, after a few samples, was ﬁnally ﬂushed from the system.

This particular data set was received a good many years ago, and nowadays any

reputable laboratory is unlikely to follow such practices. However, although mistakes

are few they can still be made and the geochemist receiving the data has to be alert to

potential problems and errors. In the situation described, there were also some Cu

enrichments and clusters of samples with elevated Cu and As could have led to some

wasteful follow-up studies if the analytical data had been taken at face value.

The need for quality control has been emphasized in Chapter 6 with an extensive dis-

cussion of Standard Reference Materials. In a discussion of data quality, (Thompson,

1989) stated

all analytical measurements are wrong; it is just a question of how large the errors are

and whether they are acceptable

As ppm

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 7

uence Number

Fig. 10-2. Periodicity in data for As from a sequence of 74 samples of plant ash analysed by

ICP-ES.

329Biogeochemistry in Mineral Exploration

With this in mind the concept of ‘Fit for Purpose’ was introduced in Chapter 7,

and in Chapter 9 discussion continues by demonstrating from the charts at the

beginning of each element description, the realistic precision that can be expected

from low-cost multi-element analysis by ICP-MS. For those looking for an in-depth

evaluation of quality control in geochemical exploration, there are good accounts

given by Thompson and Howarth (1978), Long (1999), Sketchley (1999), Smee (1998,

1999), Steger (1999), Taylor (1987) and Vallee and Sinclair (1999).

COMPUTER SOFTWARE TOOLS

There are many excellen t competing software products for examining, evaluating

and plotting geochemical data. It often comes down to personal preference or an

employer’s company policy as to which system to use. Excel is an excellent pro-

gramme for the raw database platform, and many manipulations of the data can be

undertaken within Excel. Programmes for more advanced statistical interpretations

are, among others, SPSS, SAS, S-plus and Geosoft.

Data analysis systems are available that require a modest ﬁnancial outlay for a

stand-alone product, and/or perhaps an annual license fee. Among these products

there is DataShed, provided by Maxwell Geoscience (Fremantle, Western Australia).

This is a desktop software product, which is a data management system, designed to

allow storage, management, analysis, reporting and integration of data in a func-

tional and intuitive environment.

New software designed specifically for the geochemist is ioGA S (Geochemical

Analysis System) by ioGlobal of Perth, Australia. The components of this software

package have been designed by geochemists to make complex geochemical data

analysis intuitive and easy to use for all levels of geoscientific knowledge. As such, it

is not a generic statistics package, but focuses on the use of statistical tools that are of

speciﬁc use for the geochemist and biogeochemi st. Among its features are intuitive

ways of rapidly plotting data by such tools as X–Y plots, dynamic histograms,

probability plots and box and whisker plots. Menus permit the selection of different

colours and symbols to represent differing classes of attributes, such as readily

comparing the composition of leaves and stems and their spatial distributions

with respect to underlying geological structures and lithology. There is a direct link

with Goog le Earth upon which spatial data distribution plots can be draped to assess

relationships to topography and cultural features. The website for this product

(http://www.ioglobal.net/ioGas.aspx) includes coloured screenshots and a compre-

hensive range of free videos that explain how to navigate the software. As with many

such programmes, there is a wide range of features that sufﬁce to ﬁll almost any

aspect of data interpretation and analysis, but of relevance to the exploration geo-

chemist is that the content is designed by geochemists with meaningful examples to

which the geochemist can readily relate.

330

Data Handling and Analysis

Geographical Information Systems (GIS) are used extensively in plotting the

spatial relationships of data. MapInfo and the ERSI products (including ArcView,

ArcGIS and ArcInfo) are some of the most widely recognised and widely used

systems. Low-cost applications include Manifold System and Surfer (Golden

Software, Colorado). The latter two systems are among the best for the individual

with limited budget, because for a few hundred dollars they can be purchased

outright, and require no annual licence fee. With a modest amount of effort in

learning these latter systems, the non-computer specialist can generate clear and

informative pictorial output, and the programmes allow the user to quickly interact

with many parameters to obtain a perspective that is tailored to explore for a

particular commodity. An up-to-date list of GIS options can be found at http://

en.wikipedia.org/wiki/List_of_GIS_softw are.

DATA ANALYSIS

Biogeochemical data from samples that have been carefully and consistently

collected, pr epared and analyzed, r equire statistical evaluation by the same pro-

cedures that would be applied to any set of geochemical exploration data. Stand-

ard statisti cal treatment b y univariate and bivariate methods should be applied,

and it is usually advantageous to undertake further a nalysis by a multivariate

method.

There are many textbooks on statistical analysis of data. Two old but good texts

are by Moroney (1951), and for non-param etric statistics Siegel (1956). Among the

more recent publications that provide accounts of statistics and computer applica-

tions with how and why to use a wide range of techniques, there are those by Rock

(1988), Davis (2002), Grimm and Yarnold (1995, 2000), Hels el and Hirsch (2000),

Taylor and Cihon (2004) and Tabachnick and Fiddell (2006). In Europe, Eurachem

is a ‘network of organisations y having the objective of establishing a system for the

international traceability of chemical measurements and the promotion of good

quality practices’. On their website (http://www.eurachem.ul.pt), they include a

number of documen ts that can be downloaded, each addressing a particular aspect of

quality control and data analysis.

A summary of statistical techniques used in biogeochemical data analysis is

given by Brooks (1982, 1995). He provides succinct explanations, each within a few

pages, of populations and distributions, the differences between parametric and

non-parametric methods, tests of significance, bivariate analysis and multivariate

analysis.

The following brief accounts of a number of statistical options outline the essence

of methods commonly used in geochemical data analysis in general. There are many

others that an individual worker may wish to invoke or experiment with, in order to

understand the structure of a data set. The reader is referred to the publications listed

above for a more in-depth understanding of statistical methods.

331

Biogeochemistry in Mineral Exploration