Richard H.

(Tick) Knight
AlphaGeoscience, 679PlankRoad, CliftonPark, NY 12065, USA
tknight@alphageoscience.com
Richard G. Lane
Leapfrog, 47HerefordSt., Christchurch8013, NewZealand
rlane@leapfrog3d.com
Hughan J. Ross
Leapfrog, 47HerefordSt., Christchurch8013, NewZealand
hross@leapfrog3d.com
Andrew P. G . Abraham
ZaparoLtd., P.O. Box84611, 2336Bloor Street West,
Toronto, Ontario, M6S4Z7Canada
aabraham@leapfrog3d.com
Jun Cowan
ZaparoLtd., Perth, WesternAustralia, Australia
jcowan@leapfrog3d.com
INTRODUCTION

Mineral resourceestimationrequires accurate geometric models of irregular 3D orebody
boundariesthat arecreatedusingandefficient and flexible modelingtechniques.

Theobjectiveof this study was to comparetheefficiency, flexibility andaccuracy of an
alternative, implicit geometric modelingapproach(employedby Leapfrog software)
tothoseof traditional explicit contour methodsusedby industry-standardgeneral min-
ingsoftwarepackages(GMPs) suchasMineSight.

Implicit modeling is based on a fast method of global interpolation using Radial Basis
Functions.
METHOD

Geometric Modeling

Preliminary(phase1) andfollow-up(phase2) surfacedrillinginformationoutliningthe
DorisHingeveingolddeposit wasprovidedbytheMiramar MiningCorporation. A total
of 80different geometricmodelsof theDorisHingeveinwerecreatedusingonlythe
phase1drill holehangingwall andfootwall veincontact points.

One model was created using the traditional, contour modeling method using Mine-
Sightsoftware. A total of 79implicit models werecreatedusingLeapfrog software.
Implicit models fall into two categories: semi-automatic and interpretation. Semi-
automatic models aregeneratedusingonly drill holecontact points. Interpretationmod-
elsincorporatesubjectivegeological interpretationintheformof digitizedpolylines.

CONCLUSIONS

The implicit method of geometric modeling is as accurate as the traditional modeling
method.

The implicit method is efficient, which allows for the creation and continuous update
of multiple geometric models in a fraction of the time required to construct a single
model using traditional techniques.

Implicit modelling is much more flexible since it allows incorporation of multiple geo-
logic interpretations that are conditional to the same data.

The new conditional geometric modeling workflow used in this study provides a se-
ries of accurate models that represent a range of geologically-realistic orebody
boundaries that can be used in mine planning or for quantifying the uncertainty of re-
source estimations.

Implicit resource and reserve models can be updated with new drilling information on
a daily, rather than a semi-annual or annual basis. Maintenance of evergreen geo-
metric models provide for regular mine production/reserve reconciliations that in-
crease the efficiency of mining operations.

Efficiency

Theexplicit referencemodel requirednearlyeight hourstoconstruct. Incontrast, the
seventy-nineimplicit modelscreatedinthestudy, onaverage, requiredonly13minutes
eachtocreate. Inthetimeit takestocreateonesinglemodel usingthetraditional
method, between30and40implicit modelscouldbeproduced.
Flexibility

Thetraditional model wasinflexibleinthat onesubjectiveinterpretationwasbuilt intoa
singledeterministicmodel. Hoursof workisrequiredtoregeneratethemodel usinga
different interpretation. Multipleimplicit models reflectingdifferent geological inter-
pretationsthat areconditional tothesamedataweregeneratedinminutes.
7:55 Total modeling time (single model)
0:30 Contour linking/ model validation
4:15 Contour Refinement
2:10 Contour (32) digitizing
1:00 TIN(10) creation/revision

0:30 Importing data / project set-up
Time(hours)
Traditional (MineSight) Modeling Efficiency
0:13 0:09 0:07 Total modeling time
(per model)
0:05 0:03 0:03 Model run
0:08 0:06 0:04 Model set-up

0:50 0:30 0:30 Total preparation time
0:20 Polylinedigitizing
0:30 0:30 0:30 Importingdata/
project set-up
Interpretation
Models
(S2, S3)
Semi-automatic
Models
(Preliminary, S1)
Isometric
Models

Time(hours)
Implicit (Leapfrog) Modeling Efficiency
Model Evaluation

Modelingefficiencywasmeasuredmy
recordingthetimerequiredtoperform
eachstepof thegeometricmodelling
process.

Model accuracywasdeterminedby
measuringtheperpendicular distance
betweenthenearest triangle(vertex,
edgeor face) onthegeometricwire-
framemodel surfaceandthefollow-up,
phase2hangingwall andfootwall vein
contact points.

2.03 2.09 2.15 Stage 3 Modeling
Trend/ Interpretationmodels
(18models)
2.05 2.15 2.40 2.27 Stage 2 Modeling
Interpretationmodels
(16models+1isometricmodel)
2.15 2.36 3.12 3.48 Stage 1 Modeling
Semi-automaticmodels
(16models+1isometricmodel)
2.29 2.58 3.06 Preliminary Modeling
Semi-automaticmodels
(27models)
Minimum
(most accurate)
Mean Maximum
(least accurate)
Isometric
Root Mean Square Error (RMSE) distance (meters)
Implicit (Leapfrog) Modelling Accuracy
Accuracy

Theimplicit methodgeneratedawiderangeof geometricmodelswithaccuraciesthat
werecomparabletothat of theMineSightmodel.

ThemeasuredRMSE distanceaccuracyof thetraditional MineSight model was2.06m.

RESULTS
a

b

c

d

a

c

b

d

Traditional and Implicit Model Comparison
Traditional Model (blue) Implicit Model (green)
west limb

east limb