Define an Estimation

To access this screen:

This screen is used to define grade estimation parameters for the current scenario.

If you have imported data from Datamine Supervisor, estimations derived from the imported project appear here automatically.

Define Estimations Screen Overview

Estimations that use the same parameters (estimation type, discretization level, output level, search volume) can be grouped together and given an Estimation case name. The case parameters can then be assigned to a single grade or to multiple grades in multiple zones.

  • An Estimation Case can be created automatically from the Save Models tab on the Fit Models panel. This is the recommended method if the cokriging option has been selected. Alternatively, a new case can be created and parameters defined for any combination of grade and zone. See Advanced Estimation - Fit Models - Save Models.

  • An Estimation Group can be copied to a new group and its parameters edited. The results for both groups will be written to the output model so it is then easy to compare the estimates, for example, ordinary kriging and simple kriging.

Estimations List

The area on the left of the screen lets you add, delete and copy an estimation scenario. This can be useful when testing sensitivity to grade estimation parameters or input data.

All estimation scenarios that have been committed for estimation using the Save Models panel (Fit Models screen) appear here.

To clear the list of all estimations (after confirmation), use Clear all estimations. This action cannot be undone.

Centre Panels

The centre of the Define Estimations screen hosts 3 panels, of which one is displayed at a time.

  • Estimation Setup - define the core parameters for the estimation case. See Estimation Setup.
  • Field Names - a summary of the fields that will be created in the output model file. See Field Names.
  • Soft Boundary Setup - if custom zones have been configured (using the Define Custom Zones screen) you can associate them with a model zone here. If multiple zone (or zone combinations) are represented by the custom zone, samples from all custom zone members will be used to influence the model cell values within the model zone. See Soft Boundary Setup.

Grade Variable(s) and Zone(s)

The Available Variables and Available Zones areas on the right of the panel allow you to add grades and zones to the current case. When a new grade or zone is selected the value appears in the corresponding area in the lower central portion of the panel using the Estimation Setup panel.

Estimation Setup Panel

This panel is used to define the core parameters for the estimation case. Settings applied here can have a significant effect on the outcome of your estimation run.

To define general estimation parameters:

  1. Display the Estimation Setup panel. Only one centre panel is visible at a time.
  2. On the left, select the Estimation you wish to configure.

  3. Enter a Description for the estimation scenario, or accept the default text.

    Tip: Use a consistent naming convention for estimation scenarios. Include the grade variable and key parameters, e.g. "AU Univariate No Zones SVOL 4".

  4. Choose the general estimation method:
    • Ordinary Kriging - arguably, the most widely used kriging method. It serves to estimate a value at a point of a region for which a variogram is known (Kriging methods depend on mathematical and statistical models), using data in the neighbourhood of the estimation location.

    • Simple Kriging - Simple kriging is more restrictive than ordinary kriging because you have to know the mean value of the surface, but it produces more realistic prediction error maps.

      If Simple Kriging is selected, you can choose the method for calculating the local mean grade to be used for the estimation. If you want the process to calculate the local mean automatically from the data in the search volume then you will need to select the required local mean attribute using the Local mean (SK) entry in the Field Names panel. See Field Names.

      If either Kriging method is chosen and two or more variables have been selected for the current case (and all combinations of variogram and cross-variogram models are available), you can choose to perform either Univariate or Multivariate estimation:

      1. Select Multivariate to perform cokriging.

      2. Select Univariate to perform a single grade variable estimation.

    • Nearest neighbour - The nearest neighbor method assigns grade values to blocks from the nearest sample point to the block. Closest sample gets a weight of one; all others get a weight of zero.
    • Inverse distance - a mathematical (deterministic) method assuming closer known grade values are more related than further values. Essentially, the 'power' of a known sample wanes over increasing distance. If selected, you can choose the weighting Power of known sample points.  For IPD estimates, choose which transformed distance to use, either using anisotropy or not:
      • If Use Anisotropy is checked, COKRIG will use the transformed distances defined by the search volume.
      • If Use Anisotropy is unchecked, no transformation is considered. Distances are calculated from the coordinate system used in the input Samples File (Select Samples panel).
  5. For any estimation method, choose if subcelling is performed:
    • Choose Sub-cells estimation to permit sub-cells to be constructed to create a more organic grade shell boundary within the model. This will typically produce larger model data than regular, parent-only cell models.
    • Choose Parent cell estimation to generate a regular model without sub-cells. The model prototype will determine size of blocks throughout the model. (Default setting).

  6. Select the number of discretisation points for each cell in the X, Y and Z directions. The optimum number of points can be determined using the options described in Kriging Neighbourhood Analysis. This setting is not supported if using the Nearest neighbour estimation method.

    If using 1x1x1 discretization and either of the kriging estimation methods, the system will default to not using point discretization. You can override this setting if you wish, and force point (not block) kriging by checking Point Kriging, which only appears where a 1x1x1 discretization and kriging is chosen.

    Note: The purpose of block kriging is to produce direct block averages from point or quasi-point measurements, not from known block averages. It is useful to achieve the estimation of a linear average for an attribute inside supports that are intermediate in size between the support of the sampling and the sampling domain.

  7. If you have configured attributes for Dynamic Anisotropy (attributes containing angle data for block model cells) whilst setting up a prototype model you should indicate which structure(s) will be oriented with this data; either Search volume,Variogram model or both.

    Note: If you haven't defined orientation fields for a selected Dynamic Anisotropy option, you'll see a warning to this effect.

Field Names Panel

This panel displays the field names that appear in the output grade model for each grade variable specified on the Select Samples screen. You can choose whether to include the field or not using the corresponding check box (or select all or none using the check box in the respective header cell).

Each field is assigned a default name based on the selected grade field. The default name can't be changed using the Auto name field and any field can be renamed by editing the name in the grid.

  • Local Mean: this only appears if Simple kriging is performed and is used to select the local mean attribute for the variable.
  • Number of holes: this is only supported if a Hole ID is selected on the Select Samples panel.

Available field names and default suffixes

Description

Default suffix

Auto name
See "Using Auto name", below.

-

Local mean (SK) *

-

Estimates

(Grade Name, no suffix)

Variance

_V

Number of samples

_N

Weight of mean

_WM

Sum of the positive weights

_WP

Corr(Z, Z*)
Correlation between actual and estimate

CRZ

Cov(Z, Z*)
Covariance between actual and estimate

CVZ

Cov(Z1*, Z*)
Covariance between two estimates – multivariate case only

CZ1

Slope of regression Z/Z*
Actual on estimate

_SR

Variance of Z*
Variance of estimate

VZS

Kriging efficiency

_KE

Lagrange parameter

_LG

Search volume index
Used for estimate

_SI

Distance to nearest sample
Calculated field value output with estimation

_MD

Average distance for all samples
Average distance of all samples used for estimation of block

_AD

* If you have previously created a field in the input model file it can be selected from the drop-down list. All numeric fields are listed. Once selected, the check box to the right is automatically enabled, but you can disable it if you wish to calculate the local mean based on the input sample data instead.

Using "Auto name"

A prefix can be applied to your output data files to help identify the attribute and the grade variable to which the attribute values apply. By default, your grade variable is used as the prefix, e.g. "AU", meaning output attribute names will be prefixed with "AU_".

If the Auto name check box is enabled:

  • All text fields below (which excludes the Local mean drop-down list, if Simple kriging is selected), will not be directly editable but can be enabled or disabled using their corresponding check box.
  • Any changes to the Auto name prefix will be applied to all enabled output attributes (i.e. those with an enabled check box) when <ENTER> is pressed.

If the Auto name check box is disabled:

  • All text fields below (which excludes the Local mean drop-down list, if Simple kriging is selected), can be edited by typing in any value
  • Any changes to the Auto name prefix will not be applied to enabled output attributes when <ENTER> is pressed.

Auto name field specification

The Auto name prefix can be up to 5 characters for Short field systems or 21 characters for Long-field systems.

In order to associate output fields with their sample variable you can define up to five characters that will be used as the first characters in the output field name. The process automatically creates the full output field name by adding an additional two or three characters. For example; N for the number of samples used for the estimate and _WP for the sum of the positive weights. Details of all suffixes are given in the Output Field Names view, above.  If you do not define a prefix the system will assign one automatically based on the input variable name.

Selecting and Deselecting Variables

Estimation variables can be globally enabled or disabled using the check box in the Input Variable header row, e.g.:

Enabling or disabling this check box will forcibly enable or disable all check boxes below with the exception of the Auto name row and (if it is displayed as part of Simple kriging), the Local mean (SK) row.

Deselecting or selecting variables will automatically trigger the same behaviour if there are dependent variables. For example, disabling the Variance attribute will automatically disable, amongst others, the Sum of pos. weights attribute, as variance is required to calculate the kriged weights.

Soft Boundary Setup Panel

As a resource geologist, you may want to use different combinations of sample zones for estimating a domain with soft boundaries.  

Soft boundary estimation is used when there is a gradational grade transition at a contact, rather than a sharp grade transition. This indicates that samples from an adjacent zone influence grades within that zone, therefore those samples can be used to inform the estimate.  

Using custom zones lets you estimate from different zones into your block model. Custom zones are defined using the Define Custom Zones panel. If configured, you can assign them to a model zone here.

For example, zone values 1, 2 and 3 exist. You want your model zone (1) to be influenced by samples across a larger area than just zone 1 samples. In fact, you want samples from zones 1 and 2 to contribute. To do this, you set up a custom zone representing zones 1 and 2, and assign it to model zone 1 using the Soft Boundary Setup table.

Model Zone: this field shows you the zone that was chosen to generate the estimation case (as determined by the zone used to create the Candidate Model Set on the Fit Models panel). You can't edit this value as it is part of the model set definition.

Custom Zone: all predefined custom zones are listed here. Select one to assign it to the Model Zone.

Related Topics and Activities