|
Process Help INDEST - estimate grades into a block model using indicator estimation |
|
Process Name |
Menu Path |
Link to Command Table |
|
INDEST |
Introduction
|
This is a Superprocess and running it may have an effect on other Datamine files in the project. More... |
The INDEST process uses the Indicator Estimation (IE) method to estimate grades into a block model using the cumulative distribution function (CDF) of indicator transformed sample grades.
How to use
To operate, INDEST needs a series of threshold values between the smallest and largest grade values. These threshold values, referred to as cutoffs, are used to numerically build the CDF of each block in the model. For each cutoff, data in the search volume are transformed into 0s and 1s: 1s if the data are greater than the threshold, and 0s if they are less than or equal. It then estimates the probability that the block grade is greater than the threshold value, using one of the standard estimation methods. This is usually kriging, but INDEST allows other methods such as Nearest Neighbour or Inverse Power of Distance to be used. Performing this operation for each cutoff across the range of the sample data approximates the CDF for the model cell. After the CDF is built, it is post processed to calculate the indicator estimated grade.
INDESTuses theESTIMAprocess to do the estimation for each cutoff. For further details ofESTIMArefer to the help file and the Grade Estimation User Guide.
If you are using Indicator Kriging (IK) then you must already have calculated a variogram for each cutoff, and stored the models in the Variogram Model file. TheVGRAMprocess has specific features for calculating indicator variograms.
For each cutoffINDESTcalculates the following data which can optionally be stored in the Output Model file:
-
the proportion (fraction) of the model subcell which is above cutoff.
-
the grade of the proportion of the subcell which lies above cutoff.
The main output fromINDESTis the grade above a cutoff of zero, ie the indicator estimated grade of the total subcell.
Estimation Parameter File
In order to useINDESTyou must specify one record in the Estimation Parameter File &ESTPARM for each cutoff. This requires the numeric field CUTOFF to be included in the file.
The *VALUE_IN field is the grade in the input sample &IN file to which the cutoffs are applied. Note that this is different from the use of the *VALUE_IN field when using ESTIMATE for Indicator Estimation.
If the *VALUE_OU field is not specified then the *VALUE_IN field will be created in the output model file to hold the indicator estimated grade. If a *VALUE_OU field is specified in the first record of the Estimation Parameter File, then this value will be used for the indicator estimated grade in the output model file
You can only estimate one set of indicators in a single run. In other words all the VALUE_IN fields must be the same. Also when using INDEST you cannot estimate a grade using non indicator methods in the same run.
If you are using zone control then you must explicitly specify all combinations of zone field(s) in the Estimation Parameter File. You cannot use the option that is available in ESTIMA that allows you to specify an absent data zone field value that then applies to all zones that are not explicitly identified in&ESTPARM.
If you are using zone control then you may use different cutoffs for different zones. However the PRABn and GRABn fields in the output model file must then be handled very carefully! The maximum number of cutoff values is 24.
Fields for indicator estimation:
-
BINGRADE: Used when GRMETHOD=4 to set the grade below the cutoff
-
ABVGRADE: Sets the grade above the cutoff (only used for the top bin)
Grade Above Cutoff
The calculation of the grade above each cutoff requires that the average grade between each successive pair of cutoffs be specified. For example if cutoff grades of 2, 5, 6.5 and 9.5g/t are selected then average grades are required for the ranges:
|
From |
To |
|
0 |
2 |
|
2 |
5 |
|
5 |
6.5 |
|
6.5 |
9.5 |
|
9.5 |
∞ |
Four methods are available to specify the average grade for each range, controlled by parameter @GRMETHOD:
|
GRMETHOD |
Description |
|
1 |
Average of minimum and maximum cutoff values. The grade above the highest cutoff is calculated as the highest cutoff plus half the difference between the highest and second highest cutoffs. |
|
2 |
Calculated by INDEST from the grades of the samples in the &INfile. |
|
3 |
Calculated by INDEST from the grades of the samples in the &IN file. However for the top bin (above the highest cutoff) the median grade is calculated. |
|
4 |
Values are specified by the user, using the BINGRADE and ABVGRADE fields in the &ESTPARMfile. The BINGRADE field contains the grade below the cutoff and the ABVGRADE field the grade above the cutoff. The ABVGRADE field is therefore only used for the top bin. |
GRMETHOD of 4 is illustrated in the following table:
|
CUTOFF |
BINGRADE |
ABVGRADE |
|
2 |
1.3 |
- |
|
5 |
3.6 |
- |
|
6.5 |
5.7 |
|
|
9.5 |
7.8 |
11.1 |
The values used by INDEST can be recorded by specifying an output &AVGRADES file.
Indicators
Indicator values are calculated for each sample in the input sample &IN file for each cutoff. An indicator takes the value:
0 ‑ the grade is less than or equal to the cutoff.
1 ‑ the grade is above the cutoff.
The indicator values can be saved by specifying an output &INDICATE file.
Output Model
Fields PRAB1 ... PRABn will be created in the Output Model file to store the proportion of the subcell above each cutoff. These are calculated directly byESTIMA. Then the fields GRAB1 ... GRABn are calculated during the post-processing to store the corresponding grade above each cutoff. The grade above cutoff values (GRABn) are calculated from the proportion and average grade between each pair of cutoffs. For example:
|
Cutoff Number |
Cutoff |
PRABn |
Calculation |
|
4 |
9.5 |
0.1 |
GRAB4= 11.1 (This figure is taken directly from the ABVGRADE field) |
|
3 |
6.5 |
0.3 |
GRAB3= {0.1x11.1 + (0.3‑0.1)x7.8} / 0.3 = 8.9 |
|
2 |
5 |
0.6 |
GRAB2= {0.1x11.1 + (0.3‑0.1)x7.8 + (0.6‑0.3)x5.7} / 0.6 = 7.3 |
|
1 |
2 |
0.85 |
GRAB1= {0.1x11.1 + (0.3‑0.1)x7.8 + (0.6‑0.3)x5.7 + (0.85‑0.6)x3.6} / 0.85 = 6.21 |
|
0 |
0 |
1 |
Indicator Grade=0.1x11.1 + (0.3‑0.1)x7.8 + (0.6‑0.3)x5.7 + (0.85‑0.6)x3.6 + (1.0‑0.85)x1.3 = 5.30 |
The PRABn and GRABn fields will be stored in the output &MODEL file
if parameter @PGFIELDS is set to 1.
Order Relation
One of the main drawbacks of the indicator estimation method is the Order Relation Problem. This will occur if the proportion of the subcell above cutoff n is estimated to be less than the proportion above cutoff n+1. ie PRAB(n) < PRAB(n+1). Three options are available to correct for this, controlled by parameter ORDER:
=1: Downwards.
=2: Upwards.
=3: Average of methods 1 and 2.
The recommended method (and default) is 3.
Files, Fields and Parameters
Input Files
|
Name |
I/O Status |
Required |
Type |
Description |
|
PROTO |
Input |
Yes |
Block_Model_File |
Input model prototype. This is a standard block model file containing the 13 compulsory fields. It may also contain the rotated model fields. If it includes cells then it must be sorted on IJK. |
|
IN |
Input |
Yes |
Undefined |
Input sample data. This must contain X,Y and Z fields and at least one grade field. |
|
SRCPARM |
Input |
Yes |
Undefined |
Search volume parameter file. This contains 24 compulsory fields defining the search volume and the number of samples needed for grade estimation. More than one search volume may be defined. All fields are numeric:
|
|
ESTPARM |
Input |
Yes |
Undefined |
Estimation parameter file. Each record in the file describes an estimation method and its associated parameters. The fields are dependent on the estimation methods selected. All fields are optional except for VALUE_IN, SREFNUM and CUTOFF. General fields:
Fields for IPD:
Fields for kriging:
Fields for indicator estimation
|
|
VMODPARM |
Input |
No |
Variogram - Model |
Variogram model parameter file. Each record in this file defines a variogram model type and its parameters. Only the VREFNUM field is compulsory.
|
Output Files
|
Name |
I/O Status |
Required |
Type |
Description |
|
MODEL |
Output |
No |
Block Model File |
Output model containing estimated grades, variance etc. |
|
AVGRADES |
Output |
No |
Undefined |
Output file containing cutoff grade ranges and average grade used for each range. It will include zone field(s), if any, plus the following fields: BIN: bin or grade range number LO_CUT: lower cutoff grade UP_CUT: upper cutoff grade NSAMPLES: number of samples in IN file lying within the bin BINGRADE: bin grade used for indicator kriging. This is dependent on the GRMETHOD parameter . SAMPMEAN: mean grade of samples in IN file lying within the bin |
|
INDICATE |
Output |
No |
Undefined |
Output indicator file. This is a copy of the sample input IN file, but also includes the 0/1 indicator values for each cutoff |
|
SAMPOUT |
Output |
No |
Undefined |
Output sample file containing details of weights for each sample for each cell estimated. |
Fields
|
Name |
Description |
Source |
Required |
Type |
Default |
|
X |
X coordinate of sample data in IN file. If not specified, then X is assumed. |
IN |
No |
Numeric |
Undefined |
|
Y |
Y coordinate of sample data in IN file. If not specified, then Y is assumed. |
IN |
No |
Numeric |
Undefined |
|
Z |
Z coordinate of sample data in IN file. If not specified, then Z is assumed. |
IN |
No |
Numeric |
Undefined |
|
ZONE1_F |
First field for zonal control. |
IN |
No |
Any |
Undefined |
|
ZONE2_F |
Second field for zonal control. |
IN |
No |
Any |
Undefined |
|
KEY |
Key field used to specify the field limiting the number of samples for estimation. The field must exist in the IN file. |
IN |
No |
Numeric |
Undefined |
|
LENGTH_F |
Field used for length weighting in IPD. The field must exist in the IN file. |
IN |
No |
Numeric |
Undefined |
|
DENS_F |
Field used for density weighting in IPD. The field must exist in the IN file. |
IN |
No |
Numeric |
Undefined |
Parameters
|
Name |
Description |
Required |
Default |
Range |
Values |
||||||||||
|
DISCMETH |
Cell discretisation method:
|
No |
1 |
1,2 |
1,2 |
||||||||||
|
XPOINTS |
Number of discretisation points in X. (1) |
No |
1 |
Undefined |
Undefined |
||||||||||
|
YPOINTS |
Number of discretisation points in Y. (1) |
No |
1 |
Undefined |
Undefined |
||||||||||
|
ZPOINTS |
Number of discretisation points in Z. (1) |
No |
1 |
Undefined |
Undefined |
||||||||||
|
XDSPACE |
Distance between discretisation points in X if DISCMETH=2. The default gives just one point. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
YDSPACE |
Distance between discretisation points in Y if DISCMETH=2. The default gives just one point. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
ZDSPACE |
Distance between discretisation points in Z if DISCMETH=2. The default gives just one point. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
PARENT |
Flag to control parent cell estimation:
|
No |
0 |
0,2 |
0,1,2 |
||||||||||
|
MINDISC |
Minimum number of discretisation points. Only used if PARENT=2. The default is (1). |
No |
1 |
Undefined |
Undefined |
||||||||||
|
COPYVAL |
Flag controlling copying of values from PROTO to MODEL if there is insufficient data to estimate them:
|
No |
0 |
0,1 |
0,1 |
||||||||||
|
FVALTYPE |
Flag for cell size approximation for F values:
|
No |
1 |
1,2 |
1,2 |
||||||||||
|
FSTEP |
Step size for cell approximation. This is only used if FVALTYPE=2. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
XMIN |
Minimum X value for model updating. The default is the X model origin. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
XMAX |
Maximum X value for model updating. The default is the maximum X value for PROTO. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
YMIN |
Minimum Y value for model updating. The default is the Y model origin. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
YMAX |
Maximum Y value for model updating. The default is the maximum Y value for PROTO. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
ZMIN |
Minimum Z value for model updating. The default is the Z model origin. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
ZMAX |
Maximum Z value for model updating. The default is the maximum Z value for PROTO. |
No |
Undefined |
Undefined |
Undefined |
||||||||||
|
XSUBCELL |
Number of subcells per parent cell in X if PROTO is empty. The default is (1). |
No |
1 |
Undefined |
Undefined |
||||||||||
|
YSUBCELL |
Number of subcells per parent cell in Y if PROTO is empty. The default is (1). |
No |
1 |
Undefined |
Undefined |
||||||||||
|
ZSUBCELL |
Number of subcells per parent cell in Z if PROTO is empty. The default is (1). |
No |
1 |
Undefined |
Undefined |
||||||||||
|
ORDER |
Order relation correction method:
|
No |
3 |
1,3 |
1,2,3 |
||||||||||
|
GRMETHOD |
Method for specifying average grade within each cutoff range:
|
No |
3 |
1,4 |
1,2,3,4 |
||||||||||
|
PGFIELDS |
Flag to select whether the proportion above cutoff fields (PRABn) and the grade above cutoff fields (GRABn) should be included in the output MODEL file:
|
No |
0 |
0,1 |
0,1 |
Notes
It is also used by ESTIMATE.
Example
!INDEST
&PROTO(modelb1),&IN(sampleb1),&SRCPARM(spar4),&ESTPARM(eparind),
&VMODPARM(vmodelb),&MODEL(modelind),&AVGRADES(gradeind),
&INDICATE(indind),*X(XPT),*Y(YPT),*Z(ZPT),*ZONE1_F(ROCK),
@DISCMETH=1.0,@XPOINTS=3.0,@YPOINTS=3.0,@ZPOINTS=3.0,
@PARENT=0.0,@COPYVAL=0.0,@FVALTYPE=1.0,@ORDER=3.0,
@GRMETHOD=3.0,@PGFIELDS=0.0
Output Window:
|
INDEST |
Error and Warning Messages
|
Message |
Description |
Solution |
|
|
|
|
|
|
|
|
