- Email: [email protected]
Data upcycling
Accepted Manuscript Data Upcycling Julian Vearncombe, Angela Riganti, David Isles, Sian Bright PII: DOI: Reference:
S0169-1368(17)30524-3 http://dx.doi.org/10.1016/j.oregeorev.2017.07.009 OREGEO 2279
To appear in:
Ore Geology Reviews
Received Date: Accepted Date:
4 July 2017 10 July 2017
Please cite this article as: J. Vearncombe, A. Riganti, D. Isles, S. Bright, Data Upcycling, Ore Geology Reviews (2017), doi: http://dx.doi.org/10.1016/j.oregeorev.2017.07.009
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Data Upcycling
1 1
Julian Vearncombe, Angela Riganti, 3David Isles and 1Sian Bright
2
2
3 4
1
5
Australia
6
2
7
Australia
8
3
SJS Resource Management Pty Ltd, PO Box 1093, Canning Bridge 6153, Western Geological Survey of Western Australia, 100 Plain Street, East Perth 6004, Western TGT Consulting, PO Box 224, Palmyra 6957, Western Australia
9 10
Corresponding author
11
[email protected]
12 13
Keywords
14
Data Upcycling; Legacy Data; Fit-for-Purpose; JORC; Mineral Exploration; QA/QC
15 16
Abstract
17 18
Mineral exploration and mining are data-driven industries. Here, we emphasize
19
the role of “data upcycling” as a significant contributor to modern exploration.
20
Upcycling can take three basic forms, all aimed at enhancing the veracity of data: (1)
21
re-collection of data, (2) collection of complimentary data, and (3) assessment and
22
innovative portrayal/integration of data. Upcycling of previously collected (or legacy)
23
data allows information to be integrated into modern datasets. This paper offers
24
perspectives and examples on how data upcycling benefits mineral exploration, with
25
short case studies from Western Australia highlighting the role of government,
26
service providers and resource explorers.
27 28
1. INTRODUCTION
29 30
Mineral exploration has witnessed several critical paradigm shifts, starting at the
31
time when geological mapping began to drive exploration (1930‒1950s). From the
32
1960s to early 1980s, the rising use of geophysics and the development of effective
33
drilling techniques allowed deep and undercover deposits to be discovered (Bevan et
34
al., 2016). In the late 1990s and 2000s, digitization of information, computer
35
databases and 3D visualization enabled a more rapid and detailed understanding of
36
how mineralization may behave in three dimensions (Vollgger et al., 2015). Faster,
37
semi-automatic data acquisition together with machine-driven processing is reducing
38
the human input in data collection and assessment, and will change exploration in
39
the decades to come (Perrons and McAuley, 2015; Riganti and Vearncombe, in
40
press). However, to ensure current and future use, data must still fully integrate with
41
geological knowledge, and be both verified and verifiable for future use.
42
In the realm of exploration and mining, the speedy collection and delivery of data
43
is beneficial. However, because of the ever-increasing volumes of data collected, re-
44
collected, processed and interpreted, the need for greater control and better
45
understanding of data is recognized increasingly within the industry (Perrons and
46
McAuley, 2015; Wilde, 2015). The concern in mining and exploration is veracity.
47
Veracity is about data integrity, quality and precision, and the ability to validate those
48
data, not just for today but for use into the future. Even the fastest Resource
49
evaluations take place months after data collection and may be performed by
50
geologists who have had no association with the data collection. It follows that
51
veracity should be as important — if not more — than speed of data collection,
52
variety and data volume.
53
With large volumes of new data being routinely collected, it is easy to overlook
54
previously obtained datasets (Griffin, 2015), especially those acquired on paper prior
55
to digitization and often forgotten after passing through a myriad of agencies or
56
companies. Older geological data have a wealth of applications once “upcycled” to fit
57
into present day workflows. Data upcycling can be broadly divided into three main
58
categories:
59 60
1. Re-collection of data, for example the re-logging or re-assaying of remaining diamond core (Vearncombe et al., 2016).
61
2. Adding to or improving existing data. This may include the analysis of old
62
core with new methods, such as multi- or hyperspectral loggers, or more
63
accurately locating historical collars using a differential global positioning
64
system (DGPS).
65
3. Using data differently. ‘Old’ data may be re-processed using improved
66
algorithms
and/or
new
analytical
techniques
to
reveal
previously
67
unrecognized patterns and trends. This is common place in the geophysical
68
industry (Minty and McFadden, 1998; Isles and Cunneen, 2015), especially
69
with seismic data (Diviacco et al., 2015).
70
A critical aspect of the data upcycling process is the verification of legacy data.
71
With the increased ability to collect and process data at a rapid pace, verification of
72
data is a step often overlooked or taken for granted in the exploration and mining
73
industry (Riganti and Vearncombe, in press). Wisdom and knowledge are needed to
74
drive understanding that will determine how data, information, and knowledge relate
75
(Ronald, 2016). Upcycling should be driven by top-down questioning and model-
76
deductive, hypothesis-based geology (Vearncombe et al., 2016).
77 78 79
2. THE
ROLE
OF
GOVERNMENT
IN
DATA
UPCYCLING
—
EXAMPLES FROM WESTERN AUSTRALIA
80 81
Systematic mapping, interpretation and documentation of the geology and
82
mineral resources of Australia has been a key task of the States and Territories’
83
geological surveys from the very early days of exploration of the country. Since 1888,
84
the Geological Survey of Western Australia (GSWA), has been accumulating
85
information on the geology, mineral resources, and petroleum fields through its own
86
mapping programmes as well as from annual activity reports submitted by all mineral
87
exploration companies in compliance with State legislation. Given the extent,
88
geological complexity, and substantial mineral endowment of the State there is now
89
an enormous volume of legacy information. This requires effective management with
90
suitable documentation, archiving, standardisation and acquisition of metadata to
91
allow confident use in exploration. Three GSWA programmes that upcycle legacy
92
data through digital capture and distribution are detailed below. With the aim of
93
encouraging exploration, these legacy data are made available at no cost via a
94
number of systems designed to deliver these datasets online and in standalone
95
products (Riganti et al., 2015).
96
In the mid 1990s GSWA started a programme designed to capture historical field
97
observations (recorded in original notebooks and on air-photographs) and integrate
98
them with related petrographic and paleontological studies stored in hard copy
99
reports — all with the best possible spatial attribution. Upcycling of legacy field
100
observations into the WAROX (for Western Australia ROcks) database benefits the
101
design of mapping programmes, allows more efficient target resampling and provides
102
mappers with all existing information at the start of a fieldwork programme (Riganti et
103
al., 2012). Today, this information is available in the field on GPS-connected tablet
104
computers as an additional layer with existing mapping, remote sensing imagery and
105
geophysical data. Structural measurements from previous workers can be plotted on
106
published maps, particularly in areas of difficult access. Access to legacy field
107
observations thus provides direct cost-savings during mapping. The nearly 240,000
108
sites and related data in WAROX are also an important baseline for petrological and
109
structural investigations, for example in metamorphic studies that are critical to
110
prospectivity analysis (Occhipinti et al., 2016).
111
In 2005‒06, digital capture of hardcopy exploration records in combination with
112
current reports submitted by exploration companies online (now mandatory with
113
prescribed standard requirements) was also undertaken. The WAMEX (Western
114
Australia Mineral EXploration) database houses all exploration reports submitted
115
since 1957 (more than 86,000 open-file reports as of June 2017) as searchable PDF
116
files, with key-wording and abstracting at individual report level. These can be viewed
117
and downloaded through the WAMEX search tool in GeoVIEW.WA, a web
118
application designed in-house to view and query multiple geoscientific and related
119
datasets (Figure 1). Information in digitally-submitted WAMEX reports that has a
120
spatial attribution has been extracted into relational databases, and is grouped in
121
distinct thematic repositories for easy search and retrieval by explorers. These
122
include a ‘Company Mineral Drillhole Database’ and a surface geochemistry subset,
123
containing respectively 2,036,672 drill holes and 7,036,409 geochemistry samples
124
(as of June 2017). Drilling information includes collar coordinates, orientation, depth,
125
inclination, logs and surveys — key information that can be used to assess the
126
mineral potential of a region/area, and/or to avoid duplication of costly programmes
127
by subsequent explorers (see Mt Mulgine case study below).
128
The Abandoned Mine Sites Inventory conducted from 1999 to 2011 (Ormsby et
129
al., 2003; GSWA, 2012) is a registry of mining features within 10 km of major towns,
130
1 km of main roads and selected tourist routes, and within 5 km of smaller towns and
131
communities. It contains a total of 192,523 mine site features and 56,676 digital
132
photographs of individual underground and surface excavations, dumps, and
133
rehabilitation and infrastructure features. The inventory provides baseline data on
134
historical mining-related features in Western Australia, and can be used for future
135
independent assessments of hazards, heritage value, and environmental impact.
136
Data have been demonstrated to assist exploration targeting by contributing towards
137
the understanding of controls on mineralization. The selective extraction and
138
processing of bedrock gold mineralization features from the inventory has enabled
139
the generation of three-dimensional pseudo-colour drapes that highlight the gold
140
mineralization patterns within historic mine sites (Ormsby, 2010).
141
The value of information rescued, catalogued and upcycled through GSWA
142
legacy data capture programmes is illustrated in the following two case studies. In
143
both examples, authors accessed critical data collated by this government agency —
144
this proving to be a more reliable source than hard drives inherited from past
145
tenement holders.
146
147 148
3. UPCYCLING
DETAILED
GEOLOGICAL
MAPPING
AND
AEROMAGNETIC DATA, COMET VALE
149 150
Our second example of data upcycling is taken from Isles et al. (2016). The key
151
to this work is the integration of pre-existing ‘paper’, 1:25,000 scale geological
152
mapping with detailed aeromagnetic imagery/data. Digitising and georeferencing of
153
old geological maps has facilitated the compilation of a highly detailed solid geology,
154
resolving fold and fault geometry, as well as Archaean lithological relationships in an
155
area covered by Cenozoic regolith.
156
Mapping of the Yilgarn Craton by Jack Hallberg (2015) was a commercial
157
enterprise that became a staple for gold explorers of this region from the late 1970s
158
until around 2005. Over 300 map sheets at 1:25,000 scale (covering more than
159
55,000 km²) were mapped and compiled onto photogrammetrically-controlled base
160
sheets. The prevailing technology at the time dictated that the maps were provided
161
as uncoloured transparencies, thus leading to a decrease in their popularity when the
162
‘on screen’ digital revolution struck. It is worth reflecting on the value of that mapping:
163
precise, consistent field mapping by a single, highly motivated expert, covering a
164
high proportion of the Yilgarn Craton greenstone belts — many man-years of high
165
quality work — but not in digital format. Embarking on the enormous task of digitizing
166
this work was predicated on its inherent value. The ‘replacement cost’ would be
167
prohibitive in the context of today’s budgets, and the utility of the information is
168
greatly enhanced in digital form. The fact that parts of the mapping may have been or
169
may in future be revised and enhanced adds to the value of the digital product. It is
170
relatively straightforward to extend or amend the GIS data, and it can be readily
171
reconfigured for visualization and interrogation to suit each user.
172
A programme of integrating the Hallberg mapping with best available
173
aeromagnetic imagery was initiated in parallel with digitizing. For much of the
174
mapped area, data with flight line spacing of 100 m or smaller are freely available
175
(Isles and Cunneen, 2015), thus presenting the opportunity to compile solid geology
176
maps at scales of 1:50,000 or better. Such maps are not an end point. They are for
177
the most part not ‘validated’ by field observations on rocks. Indeed, for the Yilgarn
178
Craton, drilling is the only means of validation for well over 90% of the area. The
179
solid geology maps integrated with Hallberg’s surface geology provide explorers with
180
a robust and reliable working hypothesis. This can then drive targeting and field
181
exploration, after which review and consolidation of both surface geology and solid
182
geology can be readily effected, in much the same way as the periodic upgrade of
183
drilling campaign data.
184
The example in Figure 2 illustrates the ‘feast and famine’ nature of the surface
185
geology in the Yilgarn Craton and, by contrast, the consistent yet detailed view of the
186
subsurface geology provided by the detailed aeromagnetic imagery. The
187
aeromagnetic image combines recent, government-sponsored 100 m data, with
188
‘recycled’ 40 m and 25 m data, the latter two having been captured, cleaned up and
189
merged into a single grid by GSWA. Upcycling of the 1:25,000 scale Hallberg
190
mapping and its integration with best available (and in this case freely downloadable)
191
aeromagnetic images leads to a step change in the geological picture and a
192
pronounced sharpening of exploration focus.
193 194 195
4. UPCYCLING DRILLING DATA AT MT MULGINE TUNGSTEN PROJECT
196 197
The use of legacy data without undertaking new drilling at Mt Mulgine (Yilgarn
198
Craton, Western Australia) has allowed the definition of new Resources to JORC
199
(Joint Ore Reserves Committee) standards (Hazelwood Resources announcement to
200
ASX, 5 November 2014; JORC 2012). Two separate ore bodies were identified
201
within 2 km of each other, and are part of an Archaean tungsten–molybdenum
202
system, comprising endoskarn and exoskarn (at Mulgine Hill) and stockwork vein (at
203
Mulgine Trench) mineralisation (Migisha and Both, 1991; Conner et al., 2012;
204
Vearncombe et al., 2016). Tungsten exploration in the area dates back to the 1960s,
205
with most data collected throughout the 1970s and early 1980s. However,
206
exploration in the 1980s and 1990s focused primarily on gold, with the result that of
207
the 318 holes drilled over a number of campaigns not all were assayed for tungsten.
208
Most of the original, handwritten Mt Mulgine diamond core logs are available through
209
WAMEX, and contain the original assays. Handwritten logs are a valuable resource
210
as notes and drawings of individual geologists cannot be easily transferred to a
211
structured database format.
212
Data available for validation comprised legacy diamond drilling (drilled 1972–
213
1981) and RC drilling (drilled mostly in 2008). Stacked, racked, protected from the
214
weather and in excellent condition, about 84% of the legacy diamond core remains
215
(Vearncombe et al., 2016). This protected physical core is unambiguously superior to
216
digital data of uncertain legacy. As part of the validation process, core was
217
systematically re-marked, and the lithology, weathering, veining and mineralisation
218
re-logged after a conversion from feet and inches to metres. Representative
219
selections (about 6.5%) of the core were re-sampled to assess the reliability of
220
original assays, and for the first time specific gravity data were collected (Figure 3).
221
The latter is a prerequisite for present-day resource evaluation, and was easily
222
achieved using the legacy core ― this would have not been possible had the core
223
not been conserved.
224
After the completion of rigorous data validation exercises through re-logging and
225
sampling, spreadsheets of old and new data were amalgamated into a new
226
database, and the data imported into standard exploration software for geological
227
modeling. At both Mulgine Hill and Mulgine Trench, verified data were used to create
228
mineralisation wireframes and 3D geological models (Figure 4). Drill-hole data in
229
combination with georeferenced surface maps, shaft diagrams and cross-sections
230
with structural data enabled a confident interpretation of the geology.
231
Confidence in the veracity of the legacy data and geological knowledge were
232
instrumental in defining the two Resources at Mt Mulgine. Authors were able to verify
233
directly the historical information, e.g. with down-hole geology and location of drill
234
collars, and re-assaying of a representative selection of samples (taking into account
235
nugget effects). Re-logging diamond core from the 1960s to 1980s was not
236
significantly different than using cores from a recent programme and produced
237
similar results (although modern drilling would generate oriented core offering
238
detailed contact, vein and fabric data). Where legacy data did not exist (e.g. specific
239
gravity), new data were collected from the preserved core. This case study highlights
240
how resource definition was possible despite some inherent weaknesses, leading to
241
a classification of the Resource at Mulgine Trench as Inferred. This was achieved
242
without the cost of additional drilling, but simply by upcycling and assessing the
243
veracity of legacy data.
244 245
5. DATA UPCYCLING NOW AND IN THE FUTURE
246 247
The ability to assess older material and legacy datasets is a critical step in
248
unlocking their potential as part of the data upcycling process. The case studies in
249
this paper illustrate how previously collected data can successfully be integrated into
250
modern datasets, as well as the value and the cost-savings afforded by proper
251
conservation and data management in mineral exploration (Figure 5). But will the
252
data and information that explorers collect today always be suitable for upcycling
253
tomorrow?
254
In the pre-digital era, companies maintained well-documented datasets (albeit in
255
smaller volumes) and associated records. Sketches and drawings showed inter-
256
relationships, and everything was documented in reports signed by geologists who
257
took responsibility for the veracity of their work. In legacy hard copy documents, even
258
the author’s state of mind can be assessed from the quality of handwriting. Following
259
the technical and digital revolutions of the 21st century, this appears to become
260
much more difficult, especially during downturn times when staff levels are reduced.
261
New technologies generate large volumes of data. Examples include GPS
262
location data, airborne magnetics, handheld XRF, down-hole geophysical logging,
263
and spectral mineralogical assessments of drill core. They also offer highly
264
automated acquisition and capture, and almost wholly computerized input allows
265
previously
266
understanding and knowledge of a prospect or mine area. So, whereas in the past
267
geologists gathered only and exactly what was deemed essential (as it was time-
268
consuming and too expensive to collect peripheral data), today’s explorers collect
269
more data than what is specifically required, simply because they can be generated
270
rapidly, in a cost effective manner, and in great volumes.
unavailable
visualization
of
data,
enabling
new
perspectives,
271
Because it is easier to collect and store more data, now more than ever it is also
272
necessary to understand and record exactly how and why data were collected. This
273
is important not only for QA/QC purposes but to maximize their application without
274
exceeding the data limitations, now and in the future. QA/QC is essential to
275
determine the quality of data collected. With large volumes of digital data it is often
276
already impossible for a geologist to assess if the data are reliable. Procedures need
277
to be in place to ensure continued integrity of data, starting well before the point of
278
collection. For example, when using a handheld XRF, decisions need to be made on
279
choice of sample, how and how often to calibrate the tool and how calibrations are
280
documented, what mode to use and how to ensure fairness between samples before
281
data are collected (Arne et al., 2014; Brand and Brand, 2014). All this is in addition to
282
an assessment of whether the samples analysed answer the question(s) being
283
asked. As well as using QA/QC procedures to ensure raw data are reliable, data
284
must also be fit-for-purpose. Data collected for one specific reason may not be
285
suitable for a different purpose, a key consideration in the upcycling process.
286
Large datasets should always be accompanied by a comparable amount of
287
metadata, i.e. sets of fields and values that describe the original purpose of the data
288
collected, and categorise content and managed objects (in other words, data and
289
information about data). Metadata accompanying datasets may describe features
290
such as what coordinate system was used, what settings were used on a handheld
291
XRF, and what filters or processing parameters were applied to geophysical data.
292
The Greek “meta” means “after”, as in its use in metamorphism. Looking back to the
293
collection and defining with hindsight is better than nothing. But this is not the same
294
as active documentation of information about the data environment, before, during
295
and after collection. Undoubtedly the phrase metadata is today used in a broad
296
context, but the unfortunate name highlights the low-levels of attention given to data
297
about data. To preserve the integrity of data for optimal usage at any time (and
298
specifically during upcycling), it is also critical that the links between metadata and
299
their related datasets are maintained.
300
A possible solution to ensure future trust in today’s datasets may lie in a
301
confidence analysis system that assigns data a confidence score that remains with
302
the raw data, thus enabling future generations to know how much confidence they
303
can have in using or upcycling data. For example, if magnetic images are acquired
304
with no metadata to detail how the image was produced it would get a low score,
305
whereas the same image with details of the processing in the image title may get a
306
moderate score, and a complete dataset with raw data, processed images and
307
metadata explaining processing steps taken would be given a high score. This
308
method documents data quality but does not in itself encourage quality.
309
The digital revolution has massive benefits. But the future cannot be simply more
310
and more data collected by machines or unqualified technicians. We suggest that all
311
geological data be a geologist’s personal direct responsibility. Specifically, we
312
suggest that the geologist be required to formally report on data quality. A full written
313
report with professional signature accepting responsibility should include details on:
314
(1) project environment and geological purpose, (2) original data request and set-up,
315
(3) data widget calibration(s), (4) widget use, especially at what material it was
316
directed, how, when and where, (5) data collection and data upcycling, (6) that all
317
data collected have been entered digitally with specified quality control, (7) that data
318
fall within established and documented QA/QC constraints (e.g. 2 sigma and similar
319
tests), (8) that all data have been plotted in the context of local geology and make
320
sense scientifically, and (9) the authors name and qualifications. We believe that
321
(formal or informal) requirements for personal responsibility will enhance data
322
veracity, in much the same way that the JORC code requirements have improved the
323
veracity of Resource and Reserve estimations and reporting.
324 325 326
6. CONCLUSION
327
To preserve future confidence in the large datasets assembled today, geologists
328
should take full responsibility for the design, implementation, collection, upcycling,
329
compilation, analysis and reporting of geological data, information and knowledge.
330
As part of the re-instatement of geologists to the heart of geoscience data we
331
emphasize the contributing role of data upcycling in the mineral exploration industry.
332
New discoveries and mine developments, and advances in geological understanding
333
resulting from the upcycling process are increasingly common. Data collected now
334
should have the veracity to maintain its value for future generations.
335 336
ACKNOWLEDGEMENTS
337
We thank Jun Cowan, Jack Hallberg, Dave Lawie and Stephen Sugden for
338
discussions that have significantly impacted our views and the content of this paper.
339
Anonymous referees made significant comment to this paper, which we have taken
340
into account. This article is published with the permission of the Director of the
341
Geological Survey of Western Australia.
342 343 344
REFERENCES
345
Arne, D., Jeffress, G., Segeev, N. and Margereson, A., 2014, What reporting
346
Portable XRF Data to 2012 JORC Code Guidelines Means. Australian Institute of
347
Geoscientists (AIG) News, v. 115, p. 1–6.
348 349
Bevan, J., Elias, M. and Vearncombe, J.R. 2016. The Yilgarn Retrospective 1950 -
350
1999 Symposium. Applied Earth Sciences (Transactions Institute Mining, Mineral
351
and Metallurgy B), v. 125: 190–201. DOI:10.1080/03717453.2016.1258101
352 353
Brand, N.W. and Brand, C.I., 2014. Performance comparison of portable XRF
354
instruments,
355
DOI:10.1144/geochem2012-172.
Geochemistry:
Exploration,
Environment,
Analysis
14:125‒138,
356 357
Conner, G., Bright, S., Vearncombe, J.R. and Butler-Blaxell, T., 2012, Structural
358
Control and Exploration for Tungsten Ore, in Vearncombe, J.R., eds., Structural
359
Geology and Resources (ed.) Australian Institute of Geoscientists (AIG) Bulletin v.
360
56, p. 28–42.
361
362
Diviacco, P., Wardell, N., Forlin, E., Sauli, C., Burca, M., Busato, A., Centonze, J.
363
and Pelos, C., 2015, Data rescue to extend the value of vintage seismic data: The
364
OGS-SNAP experience, GeoResJ, v. 6, p. 44–52, DOI: 10.1016/j.grj.2015.01.006.
365 366
Griffin, E.R. 2015, When are Old Data New Data?, GeoResJ, v. 6, p. 92–97, DOI:
367
10.1016/j.grj.2015.02.004.
368 369
GSWA, 2012, Inventory of abandoned mine sites: progress 1999–2002: Geological
370
Survey of Western Australia.
371 372
Hallberg, J.A., 2015. A Different Perspective on the WA Mining Boom. Australian
373
Institute of Geoscientists (AIG) Bulletin v. 60, p. 32–33.
374 375
Hazelwood Resources Ltd., 2014, Hazelwood Continues to Increase Tungsten
376
Resource. Announcement to Australian Stock Exchange, ASX, 5 November 2014.
377 378
Isles, D. and Cunneen, P., 2015, From a Mile to a Chain: The Story of
379
Aeromagnetics in the Yilgarn, in Bevan, J., Elias, M. and Vearncombe, J.R., eds.,
380
Yilgarn Retrospective, Extended Abstracts, Australian Institute of Geoscientists (AIG)
381
Bulletin v. 60, p. 34–38.
382 383
Isles, D., Hallberg, J., Rankin, L. and Hitch, B., 2016, Integration of Detailed
384
Geological Mapping and Aeromagnetics in Mature Exploration Environments. in
385
Vearncombe, J.R. ed., Brownfields Exploration: Deep and Meaningful, Australian
386
Institute of Geoscientists (AIG) Bulletin v. 6, p. 29-31.
387 388
JORC, 2012, The JORC Code 2012 Edition. Australasian Code for Reporting of
389
Exploration Results, Mineral Resources and Ore Reserves. 44pp.
390 391
Migisha, C.J.R. and Both, R.A., 1991, Mineralogy and genesis of the Trench
392
tungsten-molybdenum deposit, Mount Mulgine, Western Australia. Mineralium
393
Deposita, v. 26, p. 247–256.
394 395
Minty, B. and McFadden, P., 1998, Improved NASVD smoothing of airborne gamma-
396
ray spectra, Exploration Geophysics, v. 29, p. 516–523. DOI:10.1071/EG998516
397
398
Occhipinti, S.A., Metelka, V., Lindsay, M.D., Hollis, J.A., Aitken, A.R., Sheppard,
399
S., Orth, K., Tyler, I.M., Beardsmore, T., Hutchinson, M. and Miller, J.M., 2016,
400
Prospectivity analysis of the Halls Creek Orogen, northern Western Australia —
401
using a Mineral Systems Approach. Geological Survey of Western Australia, Report
402
159,
403
analysis-of-the-halls-creek-orogen-western-australia-using-a-mineral-systems-
404
approach.do
51pp.
http://dmpbookshop.eruditetechnologies.com.au/product/prospectivity-
405 406
Ormsby, W.R., Howard, H.M. and Eaton, N.W., 2003, Inventory of abandoned mine
407
sites: progress 1999–2002: Geological Survey of Western Australia, Record 2003/9,
408
76pp.
409
abandoned-mine-sites-progress-1999-2002-geographical-product-n03a.do.
http://dmpbookshop.eruditetechnologies.com.au/product/inventory-of-
410 411
Ormsby, W.R., 2010, The application of abandoned mine site data to the
412
interpretation of regional geology and gold mineralization in the Kalgoorlie Terrane,
413
Western Australia: Geological Survey of Western Australia, Record 2010/9, 112p.
414
http://geodocs.dmp.wa.gov.au/document/launchViewer.do?documentId=325700&cab
415
inetId=1101&forward=documentSearch&from=topNav.
416 417
Perrons, R.K. and McAuley, D., 2015, The case for “n<
418
revoluton will probably happen differently in the mining sector. Resources Policy
419
v.46: p. 234-238
420 421
Riganti, A., Farrell, T., Johnson, S. and Hocking, R., 2012, Managing the geological
422
data behind the maps — digital fieldnotes in WAROX, in Proceedings of the 34th
423
International Geological Congress 2012, 5–10 August 2012, Brisbane, Australia, p.
424
3296.
425 426
Riganti, A., Farrell, T.R., Ellis, MJ, Irimies, F., Strickland, C.D., Martin, S.K. and
427
Wallace, D.J. 2015, 123 years of Legacy Data at the Geological Survey of Western
428
Australia:
429
10.1016/j.grj.2015.02.015.
Capture
and
Delivery,
GeoResJ
v.
6:
p.
175–194,
DOI:
430 431
Riganti, A. and Vearncombe, J.R., in press, Managing data for mineral exploration, in
432
Australian Institute of Mining and Metallurgy, Australian Ore Deposits Monograph,
433
434
Ronald, E. 2016, How to Train Your Geologist,
435
Geoscientists
436
content/uploads/2016/03/aig_news_126_01_v17_MR_FINAL_spreads.pdf
(AIG)
News,
126,
19–25.
Australian Institute of https://www.aig.org.au/wp-
437 438
Vearncombe, J.R., Conner, G. and Bright, S., 2016, Value from Legacy Data.
439
Applied Earth Sciences (Transactions Institute Mining, Mineral and Metallurgy B), v.
440
125, p. 231–246.
441 442
Vollgger S.A., Cruden A.R., Ailleres, L. and Cowan, E.J., 2015, Regional dome
443
evolution and its control on ore-grade distribution: insights from 3D implicit modelling
444
of the Navachab gold deposit, Namibia. Ore Geology Reviews, v. 69, p. 268–284.
445
DOI:10.1016/j.oregeorev.2015.02.020
446 447
Wilde, A., 2015, Data Collection and Management in the Uranium Industry; A Paladin
448
Perspective. In Big Data, Inspiring Information and Strategic Knowledge, Australian
449
Institute of Geoscientists, Bulletin 61, p. 20–21.
450
451
Figure Captions
452 453
Figure 1. Example of data layers upcycled by GSWA and made available through
454
GeoVIEW.WA, a custom web application for viewing and querying multiple datasets
455
simultaneously. Highlight windows are return queries for the WAMEX, Company
456
Mineral Drillhole, and Surface Geochemistry datasets. Other government agencies in
457
Australia have developed similar applications.
458 459
Figure 2. Integrated 1:50,000 scale solid geological interpretation of the Comet Vale
460
area (near Goongarrie, 100 km north of Kalgoorlie) based on Hallberg 1:25,000 scale
461
mapping and GSWA aeromagnetic imagery. The pink and red colours in the geology
462
maps are granitic rocks, purple and mauve are ultramafic rocks, greens mafic rocks,
463
and the grey and yellow tones show sediments.
464 465
Figure 3. Photographs of data upcycling at Mt Mulgine. (a) Half core from Mt Mulgine
466
in good condition stored in covered racks at the core yard; (b) core being washed
467
down; (c) remarking core trays (d) Specific gravity measurement at Mt Mulgine, using
468
legacy core samples.
469 470
Figure 4. Mulgine Trench 3D model of tungsten mineralisation wireframes (WO3%>
471
0.10) and the relationship to geology.
472 473
Figure 5. Motivated by the journal requirement to provide a pictorial abstract, we
474
summarize in schematic form the difference between legacy data and information,
475
and how they may be upcycled with re-collected data, extra data and novel analysis,
476
plus QA/QC and fit-for-purpose tests into New Data and Knowledge.
477
478 479
480 481
482 483
484 485
486 487
488
Data Upcycling
489
Julian Vearncombe, Angela Riganti, David Isles and Sian Bright
490 491
Highlights:
492 493 494
This paper offers perspectives and examples on how “data upcycling” benefits mineral exploration, with short case studies from Western Australia highlighting the role of government, service providers and resource explorers.
495 496
497
S0169-1368(17)30524-3 http://dx.doi.org/10.1016/j.oregeorev.2017.07.009 OREGEO 2279
To appear in:
Ore Geology Reviews
Received Date: Accepted Date:
4 July 2017 10 July 2017
Please cite this article as: J. Vearncombe, A. Riganti, D. Isles, S. Bright, Data Upcycling, Ore Geology Reviews (2017), doi: http://dx.doi.org/10.1016/j.oregeorev.2017.07.009
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Data Upcycling
1 1
Julian Vearncombe, Angela Riganti, 3David Isles and 1Sian Bright
2
2
3 4
1
5
Australia
6
2
7
Australia
8
3
SJS Resource Management Pty Ltd, PO Box 1093, Canning Bridge 6153, Western Geological Survey of Western Australia, 100 Plain Street, East Perth 6004, Western TGT Consulting, PO Box 224, Palmyra 6957, Western Australia
9 10
Corresponding author
11
[email protected]
12 13
Keywords
14
Data Upcycling; Legacy Data; Fit-for-Purpose; JORC; Mineral Exploration; QA/QC
15 16
Abstract
17 18
Mineral exploration and mining are data-driven industries. Here, we emphasize
19
the role of “data upcycling” as a significant contributor to modern exploration.
20
Upcycling can take three basic forms, all aimed at enhancing the veracity of data: (1)
21
re-collection of data, (2) collection of complimentary data, and (3) assessment and
22
innovative portrayal/integration of data. Upcycling of previously collected (or legacy)
23
data allows information to be integrated into modern datasets. This paper offers
24
perspectives and examples on how data upcycling benefits mineral exploration, with
25
short case studies from Western Australia highlighting the role of government,
26
service providers and resource explorers.
27 28
1. INTRODUCTION
29 30
Mineral exploration has witnessed several critical paradigm shifts, starting at the
31
time when geological mapping began to drive exploration (1930‒1950s). From the
32
1960s to early 1980s, the rising use of geophysics and the development of effective
33
drilling techniques allowed deep and undercover deposits to be discovered (Bevan et
34
al., 2016). In the late 1990s and 2000s, digitization of information, computer
35
databases and 3D visualization enabled a more rapid and detailed understanding of
36
how mineralization may behave in three dimensions (Vollgger et al., 2015). Faster,
37
semi-automatic data acquisition together with machine-driven processing is reducing
38
the human input in data collection and assessment, and will change exploration in
39
the decades to come (Perrons and McAuley, 2015; Riganti and Vearncombe, in
40
press). However, to ensure current and future use, data must still fully integrate with
41
geological knowledge, and be both verified and verifiable for future use.
42
In the realm of exploration and mining, the speedy collection and delivery of data
43
is beneficial. However, because of the ever-increasing volumes of data collected, re-
44
collected, processed and interpreted, the need for greater control and better
45
understanding of data is recognized increasingly within the industry (Perrons and
46
McAuley, 2015; Wilde, 2015). The concern in mining and exploration is veracity.
47
Veracity is about data integrity, quality and precision, and the ability to validate those
48
data, not just for today but for use into the future. Even the fastest Resource
49
evaluations take place months after data collection and may be performed by
50
geologists who have had no association with the data collection. It follows that
51
veracity should be as important — if not more — than speed of data collection,
52
variety and data volume.
53
With large volumes of new data being routinely collected, it is easy to overlook
54
previously obtained datasets (Griffin, 2015), especially those acquired on paper prior
55
to digitization and often forgotten after passing through a myriad of agencies or
56
companies. Older geological data have a wealth of applications once “upcycled” to fit
57
into present day workflows. Data upcycling can be broadly divided into three main
58
categories:
59 60
1. Re-collection of data, for example the re-logging or re-assaying of remaining diamond core (Vearncombe et al., 2016).
61
2. Adding to or improving existing data. This may include the analysis of old
62
core with new methods, such as multi- or hyperspectral loggers, or more
63
accurately locating historical collars using a differential global positioning
64
system (DGPS).
65
3. Using data differently. ‘Old’ data may be re-processed using improved
66
algorithms
and/or
new
analytical
techniques
to
reveal
previously
67
unrecognized patterns and trends. This is common place in the geophysical
68
industry (Minty and McFadden, 1998; Isles and Cunneen, 2015), especially
69
with seismic data (Diviacco et al., 2015).
70
A critical aspect of the data upcycling process is the verification of legacy data.
71
With the increased ability to collect and process data at a rapid pace, verification of
72
data is a step often overlooked or taken for granted in the exploration and mining
73
industry (Riganti and Vearncombe, in press). Wisdom and knowledge are needed to
74
drive understanding that will determine how data, information, and knowledge relate
75
(Ronald, 2016). Upcycling should be driven by top-down questioning and model-
76
deductive, hypothesis-based geology (Vearncombe et al., 2016).
77 78 79
2. THE
ROLE
OF
GOVERNMENT
IN
DATA
UPCYCLING
—
EXAMPLES FROM WESTERN AUSTRALIA
80 81
Systematic mapping, interpretation and documentation of the geology and
82
mineral resources of Australia has been a key task of the States and Territories’
83
geological surveys from the very early days of exploration of the country. Since 1888,
84
the Geological Survey of Western Australia (GSWA), has been accumulating
85
information on the geology, mineral resources, and petroleum fields through its own
86
mapping programmes as well as from annual activity reports submitted by all mineral
87
exploration companies in compliance with State legislation. Given the extent,
88
geological complexity, and substantial mineral endowment of the State there is now
89
an enormous volume of legacy information. This requires effective management with
90
suitable documentation, archiving, standardisation and acquisition of metadata to
91
allow confident use in exploration. Three GSWA programmes that upcycle legacy
92
data through digital capture and distribution are detailed below. With the aim of
93
encouraging exploration, these legacy data are made available at no cost via a
94
number of systems designed to deliver these datasets online and in standalone
95
products (Riganti et al., 2015).
96
In the mid 1990s GSWA started a programme designed to capture historical field
97
observations (recorded in original notebooks and on air-photographs) and integrate
98
them with related petrographic and paleontological studies stored in hard copy
99
reports — all with the best possible spatial attribution. Upcycling of legacy field
100
observations into the WAROX (for Western Australia ROcks) database benefits the
101
design of mapping programmes, allows more efficient target resampling and provides
102
mappers with all existing information at the start of a fieldwork programme (Riganti et
103
al., 2012). Today, this information is available in the field on GPS-connected tablet
104
computers as an additional layer with existing mapping, remote sensing imagery and
105
geophysical data. Structural measurements from previous workers can be plotted on
106
published maps, particularly in areas of difficult access. Access to legacy field
107
observations thus provides direct cost-savings during mapping. The nearly 240,000
108
sites and related data in WAROX are also an important baseline for petrological and
109
structural investigations, for example in metamorphic studies that are critical to
110
prospectivity analysis (Occhipinti et al., 2016).
111
In 2005‒06, digital capture of hardcopy exploration records in combination with
112
current reports submitted by exploration companies online (now mandatory with
113
prescribed standard requirements) was also undertaken. The WAMEX (Western
114
Australia Mineral EXploration) database houses all exploration reports submitted
115
since 1957 (more than 86,000 open-file reports as of June 2017) as searchable PDF
116
files, with key-wording and abstracting at individual report level. These can be viewed
117
and downloaded through the WAMEX search tool in GeoVIEW.WA, a web
118
application designed in-house to view and query multiple geoscientific and related
119
datasets (Figure 1). Information in digitally-submitted WAMEX reports that has a
120
spatial attribution has been extracted into relational databases, and is grouped in
121
distinct thematic repositories for easy search and retrieval by explorers. These
122
include a ‘Company Mineral Drillhole Database’ and a surface geochemistry subset,
123
containing respectively 2,036,672 drill holes and 7,036,409 geochemistry samples
124
(as of June 2017). Drilling information includes collar coordinates, orientation, depth,
125
inclination, logs and surveys — key information that can be used to assess the
126
mineral potential of a region/area, and/or to avoid duplication of costly programmes
127
by subsequent explorers (see Mt Mulgine case study below).
128
The Abandoned Mine Sites Inventory conducted from 1999 to 2011 (Ormsby et
129
al., 2003; GSWA, 2012) is a registry of mining features within 10 km of major towns,
130
1 km of main roads and selected tourist routes, and within 5 km of smaller towns and
131
communities. It contains a total of 192,523 mine site features and 56,676 digital
132
photographs of individual underground and surface excavations, dumps, and
133
rehabilitation and infrastructure features. The inventory provides baseline data on
134
historical mining-related features in Western Australia, and can be used for future
135
independent assessments of hazards, heritage value, and environmental impact.
136
Data have been demonstrated to assist exploration targeting by contributing towards
137
the understanding of controls on mineralization. The selective extraction and
138
processing of bedrock gold mineralization features from the inventory has enabled
139
the generation of three-dimensional pseudo-colour drapes that highlight the gold
140
mineralization patterns within historic mine sites (Ormsby, 2010).
141
The value of information rescued, catalogued and upcycled through GSWA
142
legacy data capture programmes is illustrated in the following two case studies. In
143
both examples, authors accessed critical data collated by this government agency —
144
this proving to be a more reliable source than hard drives inherited from past
145
tenement holders.
146
147 148
3. UPCYCLING
DETAILED
GEOLOGICAL
MAPPING
AND
AEROMAGNETIC DATA, COMET VALE
149 150
Our second example of data upcycling is taken from Isles et al. (2016). The key
151
to this work is the integration of pre-existing ‘paper’, 1:25,000 scale geological
152
mapping with detailed aeromagnetic imagery/data. Digitising and georeferencing of
153
old geological maps has facilitated the compilation of a highly detailed solid geology,
154
resolving fold and fault geometry, as well as Archaean lithological relationships in an
155
area covered by Cenozoic regolith.
156
Mapping of the Yilgarn Craton by Jack Hallberg (2015) was a commercial
157
enterprise that became a staple for gold explorers of this region from the late 1970s
158
until around 2005. Over 300 map sheets at 1:25,000 scale (covering more than
159
55,000 km²) were mapped and compiled onto photogrammetrically-controlled base
160
sheets. The prevailing technology at the time dictated that the maps were provided
161
as uncoloured transparencies, thus leading to a decrease in their popularity when the
162
‘on screen’ digital revolution struck. It is worth reflecting on the value of that mapping:
163
precise, consistent field mapping by a single, highly motivated expert, covering a
164
high proportion of the Yilgarn Craton greenstone belts — many man-years of high
165
quality work — but not in digital format. Embarking on the enormous task of digitizing
166
this work was predicated on its inherent value. The ‘replacement cost’ would be
167
prohibitive in the context of today’s budgets, and the utility of the information is
168
greatly enhanced in digital form. The fact that parts of the mapping may have been or
169
may in future be revised and enhanced adds to the value of the digital product. It is
170
relatively straightforward to extend or amend the GIS data, and it can be readily
171
reconfigured for visualization and interrogation to suit each user.
172
A programme of integrating the Hallberg mapping with best available
173
aeromagnetic imagery was initiated in parallel with digitizing. For much of the
174
mapped area, data with flight line spacing of 100 m or smaller are freely available
175
(Isles and Cunneen, 2015), thus presenting the opportunity to compile solid geology
176
maps at scales of 1:50,000 or better. Such maps are not an end point. They are for
177
the most part not ‘validated’ by field observations on rocks. Indeed, for the Yilgarn
178
Craton, drilling is the only means of validation for well over 90% of the area. The
179
solid geology maps integrated with Hallberg’s surface geology provide explorers with
180
a robust and reliable working hypothesis. This can then drive targeting and field
181
exploration, after which review and consolidation of both surface geology and solid
182
geology can be readily effected, in much the same way as the periodic upgrade of
183
drilling campaign data.
184
The example in Figure 2 illustrates the ‘feast and famine’ nature of the surface
185
geology in the Yilgarn Craton and, by contrast, the consistent yet detailed view of the
186
subsurface geology provided by the detailed aeromagnetic imagery. The
187
aeromagnetic image combines recent, government-sponsored 100 m data, with
188
‘recycled’ 40 m and 25 m data, the latter two having been captured, cleaned up and
189
merged into a single grid by GSWA. Upcycling of the 1:25,000 scale Hallberg
190
mapping and its integration with best available (and in this case freely downloadable)
191
aeromagnetic images leads to a step change in the geological picture and a
192
pronounced sharpening of exploration focus.
193 194 195
4. UPCYCLING DRILLING DATA AT MT MULGINE TUNGSTEN PROJECT
196 197
The use of legacy data without undertaking new drilling at Mt Mulgine (Yilgarn
198
Craton, Western Australia) has allowed the definition of new Resources to JORC
199
(Joint Ore Reserves Committee) standards (Hazelwood Resources announcement to
200
ASX, 5 November 2014; JORC 2012). Two separate ore bodies were identified
201
within 2 km of each other, and are part of an Archaean tungsten–molybdenum
202
system, comprising endoskarn and exoskarn (at Mulgine Hill) and stockwork vein (at
203
Mulgine Trench) mineralisation (Migisha and Both, 1991; Conner et al., 2012;
204
Vearncombe et al., 2016). Tungsten exploration in the area dates back to the 1960s,
205
with most data collected throughout the 1970s and early 1980s. However,
206
exploration in the 1980s and 1990s focused primarily on gold, with the result that of
207
the 318 holes drilled over a number of campaigns not all were assayed for tungsten.
208
Most of the original, handwritten Mt Mulgine diamond core logs are available through
209
WAMEX, and contain the original assays. Handwritten logs are a valuable resource
210
as notes and drawings of individual geologists cannot be easily transferred to a
211
structured database format.
212
Data available for validation comprised legacy diamond drilling (drilled 1972–
213
1981) and RC drilling (drilled mostly in 2008). Stacked, racked, protected from the
214
weather and in excellent condition, about 84% of the legacy diamond core remains
215
(Vearncombe et al., 2016). This protected physical core is unambiguously superior to
216
digital data of uncertain legacy. As part of the validation process, core was
217
systematically re-marked, and the lithology, weathering, veining and mineralisation
218
re-logged after a conversion from feet and inches to metres. Representative
219
selections (about 6.5%) of the core were re-sampled to assess the reliability of
220
original assays, and for the first time specific gravity data were collected (Figure 3).
221
The latter is a prerequisite for present-day resource evaluation, and was easily
222
achieved using the legacy core ― this would have not been possible had the core
223
not been conserved.
224
After the completion of rigorous data validation exercises through re-logging and
225
sampling, spreadsheets of old and new data were amalgamated into a new
226
database, and the data imported into standard exploration software for geological
227
modeling. At both Mulgine Hill and Mulgine Trench, verified data were used to create
228
mineralisation wireframes and 3D geological models (Figure 4). Drill-hole data in
229
combination with georeferenced surface maps, shaft diagrams and cross-sections
230
with structural data enabled a confident interpretation of the geology.
231
Confidence in the veracity of the legacy data and geological knowledge were
232
instrumental in defining the two Resources at Mt Mulgine. Authors were able to verify
233
directly the historical information, e.g. with down-hole geology and location of drill
234
collars, and re-assaying of a representative selection of samples (taking into account
235
nugget effects). Re-logging diamond core from the 1960s to 1980s was not
236
significantly different than using cores from a recent programme and produced
237
similar results (although modern drilling would generate oriented core offering
238
detailed contact, vein and fabric data). Where legacy data did not exist (e.g. specific
239
gravity), new data were collected from the preserved core. This case study highlights
240
how resource definition was possible despite some inherent weaknesses, leading to
241
a classification of the Resource at Mulgine Trench as Inferred. This was achieved
242
without the cost of additional drilling, but simply by upcycling and assessing the
243
veracity of legacy data.
244 245
5. DATA UPCYCLING NOW AND IN THE FUTURE
246 247
The ability to assess older material and legacy datasets is a critical step in
248
unlocking their potential as part of the data upcycling process. The case studies in
249
this paper illustrate how previously collected data can successfully be integrated into
250
modern datasets, as well as the value and the cost-savings afforded by proper
251
conservation and data management in mineral exploration (Figure 5). But will the
252
data and information that explorers collect today always be suitable for upcycling
253
tomorrow?
254
In the pre-digital era, companies maintained well-documented datasets (albeit in
255
smaller volumes) and associated records. Sketches and drawings showed inter-
256
relationships, and everything was documented in reports signed by geologists who
257
took responsibility for the veracity of their work. In legacy hard copy documents, even
258
the author’s state of mind can be assessed from the quality of handwriting. Following
259
the technical and digital revolutions of the 21st century, this appears to become
260
much more difficult, especially during downturn times when staff levels are reduced.
261
New technologies generate large volumes of data. Examples include GPS
262
location data, airborne magnetics, handheld XRF, down-hole geophysical logging,
263
and spectral mineralogical assessments of drill core. They also offer highly
264
automated acquisition and capture, and almost wholly computerized input allows
265
previously
266
understanding and knowledge of a prospect or mine area. So, whereas in the past
267
geologists gathered only and exactly what was deemed essential (as it was time-
268
consuming and too expensive to collect peripheral data), today’s explorers collect
269
more data than what is specifically required, simply because they can be generated
270
rapidly, in a cost effective manner, and in great volumes.
unavailable
visualization
of
data,
enabling
new
perspectives,
271
Because it is easier to collect and store more data, now more than ever it is also
272
necessary to understand and record exactly how and why data were collected. This
273
is important not only for QA/QC purposes but to maximize their application without
274
exceeding the data limitations, now and in the future. QA/QC is essential to
275
determine the quality of data collected. With large volumes of digital data it is often
276
already impossible for a geologist to assess if the data are reliable. Procedures need
277
to be in place to ensure continued integrity of data, starting well before the point of
278
collection. For example, when using a handheld XRF, decisions need to be made on
279
choice of sample, how and how often to calibrate the tool and how calibrations are
280
documented, what mode to use and how to ensure fairness between samples before
281
data are collected (Arne et al., 2014; Brand and Brand, 2014). All this is in addition to
282
an assessment of whether the samples analysed answer the question(s) being
283
asked. As well as using QA/QC procedures to ensure raw data are reliable, data
284
must also be fit-for-purpose. Data collected for one specific reason may not be
285
suitable for a different purpose, a key consideration in the upcycling process.
286
Large datasets should always be accompanied by a comparable amount of
287
metadata, i.e. sets of fields and values that describe the original purpose of the data
288
collected, and categorise content and managed objects (in other words, data and
289
information about data). Metadata accompanying datasets may describe features
290
such as what coordinate system was used, what settings were used on a handheld
291
XRF, and what filters or processing parameters were applied to geophysical data.
292
The Greek “meta” means “after”, as in its use in metamorphism. Looking back to the
293
collection and defining with hindsight is better than nothing. But this is not the same
294
as active documentation of information about the data environment, before, during
295
and after collection. Undoubtedly the phrase metadata is today used in a broad
296
context, but the unfortunate name highlights the low-levels of attention given to data
297
about data. To preserve the integrity of data for optimal usage at any time (and
298
specifically during upcycling), it is also critical that the links between metadata and
299
their related datasets are maintained.
300
A possible solution to ensure future trust in today’s datasets may lie in a
301
confidence analysis system that assigns data a confidence score that remains with
302
the raw data, thus enabling future generations to know how much confidence they
303
can have in using or upcycling data. For example, if magnetic images are acquired
304
with no metadata to detail how the image was produced it would get a low score,
305
whereas the same image with details of the processing in the image title may get a
306
moderate score, and a complete dataset with raw data, processed images and
307
metadata explaining processing steps taken would be given a high score. This
308
method documents data quality but does not in itself encourage quality.
309
The digital revolution has massive benefits. But the future cannot be simply more
310
and more data collected by machines or unqualified technicians. We suggest that all
311
geological data be a geologist’s personal direct responsibility. Specifically, we
312
suggest that the geologist be required to formally report on data quality. A full written
313
report with professional signature accepting responsibility should include details on:
314
(1) project environment and geological purpose, (2) original data request and set-up,
315
(3) data widget calibration(s), (4) widget use, especially at what material it was
316
directed, how, when and where, (5) data collection and data upcycling, (6) that all
317
data collected have been entered digitally with specified quality control, (7) that data
318
fall within established and documented QA/QC constraints (e.g. 2 sigma and similar
319
tests), (8) that all data have been plotted in the context of local geology and make
320
sense scientifically, and (9) the authors name and qualifications. We believe that
321
(formal or informal) requirements for personal responsibility will enhance data
322
veracity, in much the same way that the JORC code requirements have improved the
323
veracity of Resource and Reserve estimations and reporting.
324 325 326
6. CONCLUSION
327
To preserve future confidence in the large datasets assembled today, geologists
328
should take full responsibility for the design, implementation, collection, upcycling,
329
compilation, analysis and reporting of geological data, information and knowledge.
330
As part of the re-instatement of geologists to the heart of geoscience data we
331
emphasize the contributing role of data upcycling in the mineral exploration industry.
332
New discoveries and mine developments, and advances in geological understanding
333
resulting from the upcycling process are increasingly common. Data collected now
334
should have the veracity to maintain its value for future generations.
335 336
ACKNOWLEDGEMENTS
337
We thank Jun Cowan, Jack Hallberg, Dave Lawie and Stephen Sugden for
338
discussions that have significantly impacted our views and the content of this paper.
339
Anonymous referees made significant comment to this paper, which we have taken
340
into account. This article is published with the permission of the Director of the
341
Geological Survey of Western Australia.
342 343 344
REFERENCES
345
Arne, D., Jeffress, G., Segeev, N. and Margereson, A., 2014, What reporting
346
Portable XRF Data to 2012 JORC Code Guidelines Means. Australian Institute of
347
Geoscientists (AIG) News, v. 115, p. 1–6.
348 349
Bevan, J., Elias, M. and Vearncombe, J.R. 2016. The Yilgarn Retrospective 1950 -
350
1999 Symposium. Applied Earth Sciences (Transactions Institute Mining, Mineral
351
and Metallurgy B), v. 125: 190–201. DOI:10.1080/03717453.2016.1258101
352 353
Brand, N.W. and Brand, C.I., 2014. Performance comparison of portable XRF
354
instruments,
355
DOI:10.1144/geochem2012-172.
Geochemistry:
Exploration,
Environment,
Analysis
14:125‒138,
356 357
Conner, G., Bright, S., Vearncombe, J.R. and Butler-Blaxell, T., 2012, Structural
358
Control and Exploration for Tungsten Ore, in Vearncombe, J.R., eds., Structural
359
Geology and Resources (ed.) Australian Institute of Geoscientists (AIG) Bulletin v.
360
56, p. 28–42.
361
362
Diviacco, P., Wardell, N., Forlin, E., Sauli, C., Burca, M., Busato, A., Centonze, J.
363
and Pelos, C., 2015, Data rescue to extend the value of vintage seismic data: The
364
OGS-SNAP experience, GeoResJ, v. 6, p. 44–52, DOI: 10.1016/j.grj.2015.01.006.
365 366
Griffin, E.R. 2015, When are Old Data New Data?, GeoResJ, v. 6, p. 92–97, DOI:
367
10.1016/j.grj.2015.02.004.
368 369
GSWA, 2012, Inventory of abandoned mine sites: progress 1999–2002: Geological
370
Survey of Western Australia.
371 372
Hallberg, J.A., 2015. A Different Perspective on the WA Mining Boom. Australian
373
Institute of Geoscientists (AIG) Bulletin v. 60, p. 32–33.
374 375
Hazelwood Resources Ltd., 2014, Hazelwood Continues to Increase Tungsten
376
Resource. Announcement to Australian Stock Exchange, ASX, 5 November 2014.
377 378
Isles, D. and Cunneen, P., 2015, From a Mile to a Chain: The Story of
379
Aeromagnetics in the Yilgarn, in Bevan, J., Elias, M. and Vearncombe, J.R., eds.,
380
Yilgarn Retrospective, Extended Abstracts, Australian Institute of Geoscientists (AIG)
381
Bulletin v. 60, p. 34–38.
382 383
Isles, D., Hallberg, J., Rankin, L. and Hitch, B., 2016, Integration of Detailed
384
Geological Mapping and Aeromagnetics in Mature Exploration Environments. in
385
Vearncombe, J.R. ed., Brownfields Exploration: Deep and Meaningful, Australian
386
Institute of Geoscientists (AIG) Bulletin v. 6, p. 29-31.
387 388
JORC, 2012, The JORC Code 2012 Edition. Australasian Code for Reporting of
389
Exploration Results, Mineral Resources and Ore Reserves. 44pp.
390 391
Migisha, C.J.R. and Both, R.A., 1991, Mineralogy and genesis of the Trench
392
tungsten-molybdenum deposit, Mount Mulgine, Western Australia. Mineralium
393
Deposita, v. 26, p. 247–256.
394 395
Minty, B. and McFadden, P., 1998, Improved NASVD smoothing of airborne gamma-
396
ray spectra, Exploration Geophysics, v. 29, p. 516–523. DOI:10.1071/EG998516
397
398
Occhipinti, S.A., Metelka, V., Lindsay, M.D., Hollis, J.A., Aitken, A.R., Sheppard,
399
S., Orth, K., Tyler, I.M., Beardsmore, T., Hutchinson, M. and Miller, J.M., 2016,
400
Prospectivity analysis of the Halls Creek Orogen, northern Western Australia —
401
using a Mineral Systems Approach. Geological Survey of Western Australia, Report
402
159,
403
analysis-of-the-halls-creek-orogen-western-australia-using-a-mineral-systems-
404
approach.do
51pp.
http://dmpbookshop.eruditetechnologies.com.au/product/prospectivity-
405 406
Ormsby, W.R., Howard, H.M. and Eaton, N.W., 2003, Inventory of abandoned mine
407
sites: progress 1999–2002: Geological Survey of Western Australia, Record 2003/9,
408
76pp.
409
abandoned-mine-sites-progress-1999-2002-geographical-product-n03a.do.
http://dmpbookshop.eruditetechnologies.com.au/product/inventory-of-
410 411
Ormsby, W.R., 2010, The application of abandoned mine site data to the
412
interpretation of regional geology and gold mineralization in the Kalgoorlie Terrane,
413
Western Australia: Geological Survey of Western Australia, Record 2010/9, 112p.
414
http://geodocs.dmp.wa.gov.au/document/launchViewer.do?documentId=325700&cab
415
inetId=1101&forward=documentSearch&from=topNav.
416 417
Perrons, R.K. and McAuley, D., 2015, The case for “n<
418
revoluton will probably happen differently in the mining sector. Resources Policy
419
v.46: p. 234-238
420 421
Riganti, A., Farrell, T., Johnson, S. and Hocking, R., 2012, Managing the geological
422
data behind the maps — digital fieldnotes in WAROX, in Proceedings of the 34th
423
International Geological Congress 2012, 5–10 August 2012, Brisbane, Australia, p.
424
3296.
425 426
Riganti, A., Farrell, T.R., Ellis, MJ, Irimies, F., Strickland, C.D., Martin, S.K. and
427
Wallace, D.J. 2015, 123 years of Legacy Data at the Geological Survey of Western
428
Australia:
429
10.1016/j.grj.2015.02.015.
Capture
and
Delivery,
GeoResJ
v.
6:
p.
175–194,
DOI:
430 431
Riganti, A. and Vearncombe, J.R., in press, Managing data for mineral exploration, in
432
Australian Institute of Mining and Metallurgy, Australian Ore Deposits Monograph,
433
434
Ronald, E. 2016, How to Train Your Geologist,
435
Geoscientists
436
content/uploads/2016/03/aig_news_126_01_v17_MR_FINAL_spreads.pdf
(AIG)
News,
126,
19–25.
Australian Institute of https://www.aig.org.au/wp-
437 438
Vearncombe, J.R., Conner, G. and Bright, S., 2016, Value from Legacy Data.
439
Applied Earth Sciences (Transactions Institute Mining, Mineral and Metallurgy B), v.
440
125, p. 231–246.
441 442
Vollgger S.A., Cruden A.R., Ailleres, L. and Cowan, E.J., 2015, Regional dome
443
evolution and its control on ore-grade distribution: insights from 3D implicit modelling
444
of the Navachab gold deposit, Namibia. Ore Geology Reviews, v. 69, p. 268–284.
445
DOI:10.1016/j.oregeorev.2015.02.020
446 447
Wilde, A., 2015, Data Collection and Management in the Uranium Industry; A Paladin
448
Perspective. In Big Data, Inspiring Information and Strategic Knowledge, Australian
449
Institute of Geoscientists, Bulletin 61, p. 20–21.
450
451
Figure Captions
452 453
Figure 1. Example of data layers upcycled by GSWA and made available through
454
GeoVIEW.WA, a custom web application for viewing and querying multiple datasets
455
simultaneously. Highlight windows are return queries for the WAMEX, Company
456
Mineral Drillhole, and Surface Geochemistry datasets. Other government agencies in
457
Australia have developed similar applications.
458 459
Figure 2. Integrated 1:50,000 scale solid geological interpretation of the Comet Vale
460
area (near Goongarrie, 100 km north of Kalgoorlie) based on Hallberg 1:25,000 scale
461
mapping and GSWA aeromagnetic imagery. The pink and red colours in the geology
462
maps are granitic rocks, purple and mauve are ultramafic rocks, greens mafic rocks,
463
and the grey and yellow tones show sediments.
464 465
Figure 3. Photographs of data upcycling at Mt Mulgine. (a) Half core from Mt Mulgine
466
in good condition stored in covered racks at the core yard; (b) core being washed
467
down; (c) remarking core trays (d) Specific gravity measurement at Mt Mulgine, using
468
legacy core samples.
469 470
Figure 4. Mulgine Trench 3D model of tungsten mineralisation wireframes (WO3%>
471
0.10) and the relationship to geology.
472 473
Figure 5. Motivated by the journal requirement to provide a pictorial abstract, we
474
summarize in schematic form the difference between legacy data and information,
475
and how they may be upcycled with re-collected data, extra data and novel analysis,
476
plus QA/QC and fit-for-purpose tests into New Data and Knowledge.
477
478 479
480 481
482 483
484 485
486 487
488
Data Upcycling
489
Julian Vearncombe, Angela Riganti, David Isles and Sian Bright
490 491
Highlights:
492 493 494
This paper offers perspectives and examples on how “data upcycling” benefits mineral exploration, with short case studies from Western Australia highlighting the role of government, service providers and resource explorers.
495 496
497