Targeting Workflow Report

Workflow Study: G_targeting_V2
Date: Fri Nov 19 12:42:55 2010


The Targeting Workflow is designed to guide the user through a step-by-step process of combining 3D datasets by different processes to obtain a comprehensive exploration interpretation. The result of the combination is an additional quantitative layer of interpretation that is tied with a multi-disciplinary approach to targeting.

The different combination methods can be based on either subjective empirical models or probabilistic models, referred to as knowledge-driven and data-driven approaches, respectively. Knowledge-driven approaches are based on the experience of domain experts and include processes such as Boolean Logic, Index Overlay, Multi-Class Index Overlay, Dempster-Shafer Belief Theory, and Fuzzy Logic. Data-driven approaches require that known occurrences of what we are looking for exist in the earth model, such as a set of deposits, and include the Prospector Model, Weights-of-Evidence, Weighted Logistic Regression, Likelihood Ratio, and Neural Networks, among others. In addition, hybrid models using combinations of two or more different approaches have proven to be effective in many studies. Harris and Sanborn-Barrie (2006)1 provide a comprehensive overview of the various modeling approaches.

The Targeting Workflow provides functionality to perform Boolean Logic, Index Overlay, Multi-Class Index Overlay, and Weights-of-Evidence prediction models. Each uses a combination of exploration criteria to generate a prediction model which can be used for targeting XYZ drillhole target positions to be further investigated for mineral potential. Each of the processes implemented here is based on documentation from Bonham-Carter (1994)2.


Report Tree


Targeting Method

Index Overlay involves a combination of weighted binary properties using a simple intersection algorithm where the binary classes (1 or 0) of each property are multiplied by a single weight factor, summed over all properties being combined and normalized by the sum of all weights following the equation:

Weights are defined by the expert and are based on the significance of the evidential property to the exploration model. The result is a weighted score defining favourability of mineral potential. This method allows for a simple ranking of the contributing evidences as a whole.

Targeting Approach Knowledge
Approach LogicWeights of Evidence

Pre-Processing

Evidential Properties

Evidential properties represent the exploration criteria in a targeting model. The Targeting Workflow requires that all evidential properties be prepared and stored as properties on a voxet/sgrid object prior to starting the targeting model process. Properties can be binary, multi-class (categorical) (with or without assigned classification) or continuous in type and will be treating accordingly in the reclassification within the Property Settings step. Although it is not mandatory, it is suggested that multi-class properties be associated with a classification where possible in order to simplify the display of unique classes in the reclassification step and to ease interpretation of units during modeling.
Valid true
Grid/Voxet Name G_Evidence_Layer

PropertyType
AOI_Dehua_MINFILE_showings_dist_out Binary
AOI_G_Surf_Mag_highs_dist_out Binary
AOI_G_block_fault_INTERSECTIONS_dist_out Binary
BATHOLITH Binary
CONTACTS_DIKES_ALL Binary
G_detailed_faults_Curve_dist_out Binary
G_sigma_50m_same_origin_sig1 Binary
G_sigma_50m_same_origin_sig2 Binary
JOGS_PINCH_OUTS_VTDATA Binary

Prediction Model Volumes

The prediction model volume represents the '3D study area' within which to perform the modeling. The volumes are defined by the spatial extents of each of the contributing evidential properties taking into account any no data values and inconsistent data coverage between properties. Each model volume (if more than one) represents the volume where the maximum number of contributing evidential properties overlap.
Model Volume Name Model1
Model Volume Region model_region_0
Model Volume 100

Evidential Property Settings

AOI_Dehua_MINFILE_showings_dist_out

Output Property Name AOI_Dehua_MINFILE_showings_dist_out_out

Evidential Property Reclassification

Use Lower false
Reclassification Method Manual
Manual Cutoff(s) -

ClassClassLower Range ValueUpper Range Value
1 1 0.9999 1.5
2 2 1.5 2.0001

AOI_G_Surf_Mag_highs_dist_out

Output Property Name AOI_G_Surf_Mag_highs_dist_out_out

Evidential Property Reclassification

Use Lower false
Reclassification Method Manual
Manual Cutoff(s) -

ClassClassLower Range ValueUpper Range Value
1 1 0.9999 1.5
2 2 1.5 2.0001

AOI_G_block_fault_INTERSECTIONS_dist_out

Output Property Name AOI_G_block_fault_INTERSECTIONS_dist_out_out

Evidential Property Reclassification

Use Lower false
Reclassification Method Manual
Manual Cutoff(s) -

ClassClassLower Range ValueUpper Range Value
1 1 0.9999 1.5
2 2 1.5 2.0001

BATHOLITH

Output Property Name BATHOLITH_out

Evidential Property Reclassification

Use Lower false
Reclassification Method Manual
Manual Cutoff(s) -

ClassClassLower Range ValueUpper Range Value
1 1 0.9999 1.5
2 2 1.5 2.0001

CONTACTS_DIKES_ALL

Output Property Name CONTACTS_DIKES_ALL_out

Evidential Property Reclassification

Use Lower false
Reclassification Method Manual
Manual Cutoff(s) -

ClassClassLower Range ValueUpper Range Value
1 1 0.9999 1.5
2 2 1.5 2.0001

G_detailed_faults_Curve_dist_out

Output Property Name G_detailed_faults_Curve_dist_out_out

Evidential Property Reclassification

Use Lower false
Reclassification Method Manual
Manual Cutoff(s) -

ClassClassLower Range ValueUpper Range Value
1 1 0.9999 1.5
2 2 1.5 2.0001

G_sigma_50m_same_origin_sig1

Output Property Name G_sigma_50m_same_origin_sig1_out

Evidential Property Reclassification

Use Lower false
Reclassification Method Manual
Manual Cutoff(s) -

ClassClassLower Range ValueUpper Range Value
1 1 -0.0001 0.5
2 2 0.5 1.0001

G_sigma_50m_same_origin_sig2

Output Property Name G_sigma_50m_same_origin_sig2_out

Evidential Property Reclassification

Use Lower false
Reclassification Method Manual
Manual Cutoff(s) -

ClassClassLower Range ValueUpper Range Value
1 1 -0.0001 0.5
2 2 0.5 1.0001

JOGS_PINCH_OUTS_VTDATA

Output Property Name JOGS_PINCH_OUTS_VTDATA_out

Evidential Property Reclassification

Use Lower false
Reclassification Method Manual
Manual Cutoff(s) -

ClassClassLower Range ValueUpper Range Value
1 1 0.9999 1.5
2 2 1.5 2.0001

Precombination

This is an optional step which allows the user to pre-combine two or more evidential properties prior to computing the prediction model. Pre-combining datasets will account for evidential properties which are too strongly correlated and would result in an over-biased prediction result.
Generated true
Perform Data Precombination? false

Evidence Weights

In binary and multi-class Index Overlay models, each evidential property is assigned a weight factor determined by its importance to the exploration model. Multi-Class Index Overlay further weights evidential properties by assigning scores to each of the property classes. Weights and scores are determined by the expert. They should be determined by relative importance (most important evidential property/class assigned the highest weight/score value) and are not restricted to integer values. The user is advised to use similar scoring schemes since scores are not normalized in the Multi-Class Index Overlay algorithm (i.e. scores should not range from 0-1 on one property and 0-10 on another within the same prediction model).
Scores_Out -

PropertyWeight
AOI_Dehua_MINFILE_showings_dist_out_out 1.5
AOI_G_Surf_Mag_highs_dist_out_out 1.1
AOI_G_block_fault_INTERSECTIONS_dist_out_out 1.1
BATHOLITH_out 1.2
CONTACTS_DIKES_ALL_out 1.1
G_detailed_faults_Curve_dist_out_out 1.2
G_sigma_50m_same_origin_sig1_out 1.15
G_sigma_50m_same_origin_sig2_out 1.3
JOGS_PINCH_OUTS_VTDATA_out 1.1

Processing

Prediction Model Generation

At this step in the workflow the knowledge-driven data combination algorithm is executed on the selected evidential properties, taking into account weights and scores if applicable, generating the final prediction model. Multiple prediction models can be generated at this step by selecting various combinations of evidential properties.
ModelName G_Targeting_Model_V2, G_Targeting_Model_Final

ModelPrediction Model Properties
G_Targeting_Model_V2 AOI_Dehua_MINFILE_showings_dist_out_out, AOI_G_Surf_Mag_highs_dist_out_out, AOI_G_block_fault_INTERSECTIONS_dist_out_out, BATHOLITH_out, CONTACTS_DIKES_ALL_out, G_detailed_faults_Curve_dist_out_out, G_sigma_50m_same_origin_sig1_out, G_sigma_50m_same_origin_sig2_out, JOGS_PINCH_OUTS_VTDATA_out
G_Targeting_Model_Final AOI_Dehua_MINFILE_showings_dist_out_out, AOI_G_Surf_Mag_highs_dist_out_out, AOI_G_block_fault_INTERSECTIONS_dist_out_out, BATHOLITH_out, CONTACTS_DIKES_ALL_out, G_detailed_faults_Curve_dist_out_out, G_sigma_50m_same_origin_sig1_out, G_sigma_50m_same_origin_sig2_out, JOGS_PINCH_OUTS_VTDATA_out

Post-Processing

Targeting

Once the prediction model is generated, the workflow allows for an advanced targeting approach by allowing the user to refine and interpret the model down to the drillhole targets level. The prediction model can be refined by looking only at cells within a specific sub-region of the model, for example, in only the undrilled portion of the model, or within the spatial extents of a claim block area, etc.

A target value cutoff should be applied here which will show only the top percentage of the target result. From this top, typically 1-5%, of data, clusters of values can be generated using a connectivity type factor and ranked based on size. The cells of each cluster can be analyzed separately and the top cells within those clusters can be exported as a set of points representing the XYZ values of your drillhole targets.

Prediction Model Name G_Targeting_Model_V2
Target Cutoff 0.860465109348297 (99.9749%)
Cluster Connectivity Type corners

Target Clusters


Target_RankTarget_VolumeTarget_CellsTarget_MinTarget_MaxTarget_MeanTarget_Median
1 0 1 0.6418605 0.6418605 0.6418605 0.6418605
2 0 1 0.5767442 0.5767442 0.5767442 0.5767442
3 0 1 0.5767442 0.5767442 0.5767442 0.5767442
4 0 6 0.5767442 0.5767442 0.5767441 0.5767442
5 0 15 0.6744186 0.6744186 0.6744185 0.6744186
6 0 1 0.6511628 0.6511628 0.6511628 0.6511628
7 0 10 0.6465116 0.6511628 0.6488372 0.6465116
8 0 27 0.6465116 0.6465116 0.6465117 0.6465116
9 0 22 0.6465116 0.6465116 0.6465118 0.6465116
10 0 6 0.6465116 0.6465116 0.6465116 0.6465116
11 0 1 0.6465116 0.6465116 0.6465116 0.6465116
12 0 7 0.6511628 0.6511628 0.6511628 0.6511628
13 0 2 0.6465116 0.6465116 0.6465116 0.6465116
14 0 2 0.6511628 0.6511628 0.6511628 0.6511628
15 0 250 0.6325582 0.8604651 0.6984743 0.6511628
16 0 1 0.6325582 0.6325582 0.6325582 0.6325582
17 0 14 0.6465116 0.6465116 0.6465116 0.6465116
18 0 20 0.6511628 0.6511628 0.6511629 0.6511628
19 0 4 0.6325582 0.6325582 0.6325582 0.6325582
20 0 8 0.6325582 0.6325582 0.6325582 0.6325582
21 0 6 0.6511628 0.6511628 0.6511628 0.6511628
22 0 28 0.6325582 0.6325582 0.6325582 0.6325582
23 0 2 0.6511628 0.6511628 0.6511628 0.6511628
24 0 1 0.6511628 0.6511628 0.6511628 0.6511628
25 0 2 0.6511628 0.6511628 0.6511628 0.6511628
26 0 18 0.6511628 0.6511628 0.6511629 0.6511628
27 0 13 0.6511628 0.6511628 0.6511627 0.6511628
28 0 67 0.6325582 0.6511628 0.6358902 0.6325582
29 0 8 0.6511628 0.6511628 0.6511628 0.6511628
30 0 9 0.6511628 0.6511628 0.6511628 0.6511628

Drillhole Target Centroids List

The following table displays a list of targets within the Gocad pointset object generated by the 'Add to Centroid Target List' command. These points represent the centroid of the selected Target Cluster. They are ranked by the order they were selected in the Target Clusters Table.


Centroid_IdCentroid_XCentroid_YCentroid_Z
0 351775 6.96692e+06 1401
1 352175 6.97452e+06 1401
2 352075 6.97442e+06 1401
3 352025 6.9743e+06 1401
4 351218 6.974e+06 1401
5 350525 6.97138e+06 1401
6 351055 6.97136e+06 1401
7 351803 6.97111e+06 1401
8 351266 6.97109e+06 1401
9 349767 6.97071e+06 1401
10 351275 6.97062e+06 1401
11 352861 6.9706e+06 1401
12 351425 6.9705e+06 1401
13 352200 6.9704e+06 1401
14 350443 6.97086e+06 1401
15 352075 6.97012e+06 1401
16 348661 6.97004e+06 1401
17 351080 6.96984e+06 1401
18 352250 6.9698e+06 1401
19 352000 6.96982e+06 1401
20 351875 6.9696e+06 1401
21 352888 6.96968e+06 1401
22 353375 6.9695e+06 1401
23 351425 6.96948e+06 1401
24 352175 6.9694e+06 1401
25 353719 6.96902e+06 1401
26 350571 6.96886e+06 1401
27 351900 6.96882e+06 1401
28 351588 6.96848e+06 1401
29 351436 6.9678e+06 1401

Drillhole Target Cells List

The following table displays a list of targets within the Gocad pointset object generated by the 'Add to Drillhole Target List' command. These points represent the individual cells selected from the target region. They are ranked by the order they were selected in the Property Viewer Table.

References:

1 Harris, J.R., Sanborn-Barrie, M., 2006, Mineral Potential Mapping: Examples from the Red Lake Greenstone Belt, Northwest Ontario, in Harris, J.R., ed., GIS for the Earth Sciences: Geological Association of Canada, Special Publication 44, p. 1-21.

2 Bonham-Carter, G.F., 1994, Geographic Information Systems for Geoscientists: Modeling with GIS: Pergamon, Oxford, 398 p.

3Thiart, C., Bonham-Carter, G.F., Agterberg, F.P., Cheng, Q., and Panahi, A., 2006, An application of the new omnibus test for conditional independence in weights-of-evidence modelling, in Harris, J.R., ed., GIS for the Earth Sciences: Geological Association of Canada, Special Publication 44, p. 131-142.

4 Agterberg, F.P., Bonham-Carter, G.F., Wright, D.F., 1990, Statistical Pattern Integration for Mineral Exploration: in Gaal, G. and Merriam, D.F., ed., Computer Applications in Resource Estimation: Prediction and Assessment for Metals and Petroleum, Pergamon Press, Toronto, p. 1-21.



Produced by Mira Geoscience. For technical support contact support@mirageoscience.com