|
Process Help |
|
Process Name |
Menu Path |
Link to Command Table |
|
PANELEST |
Estimate ribbon | Estimate | Panels |
Introduction
Estimate grade and variance of 2D or 3D panels.
Panels are defined either as a set of strings from file PERIM, or as a set of 2D or 3D discretisation points from file DISPTIN. The interpolation methods available are nearest neighbour, inverse power of distance or kriging.
In particular the process allows you to estimate a grade and a kriged variance for:
- any perimeter, without the need to create a block model;
- a subset of cells from a block model.
How to use
Defining a Panel by a Set of Strings
The PERIM file may contain one or more strings. If the strings are not closed the process will close them to form the panel. You must ensure that a string does not intersect itself as this will lead to incorrect results. String intersections are not checked by the process. There is a limit of 5000 points in any one string.
The strings must be planar and must lie in one of the three orthogonal planes. If a string does not satisfy both these criteria then a warning will be displayed and that string will not be evaluated. Although the strings must be in one of the three orthogonal planes, they do not all have to be in the same plane.
You can define independent plus and minus projection distances using the parameters DPLUS and DMINUS in order to create a volume. If either of these values are non zero then the panel will be considered as a 3D panel; otherwise it is a 2D panel.
If a panel is 2D then all sample data will be projected onto the plane of the panel even if all three coordinates (X,Y,Z) are specified for the sample data. Hence the evaluation will be in 2D. If you are kriging a 2D panel then you must take care that projecting data onto the plane of the panel does not lead to coincident samples. The process will terminate with an error if there are coincident samples and kriging is used.
Each panel is represented by a set of 2D or 3D discretisation points. You can define the spacing between discretisation points independently in each of the three dimensions using the parameters XDSPACE, YDSPACE and ZDSPACE. Alternatively if you set these parameters to zero the process will calculate a suitable spacing between points. It does this by calculating the maximum extent of the panel in the two dimensions of its plane, and then calculating the corresponding area. This area is equivalent to a rectangle surrounding the panel. Taking the square root of the area gives the side of a square with the same area. The distance between points is then calculated by dividing the length of the side of this square by 10.5. In other words if the original panel were square it would contain 11 x 11 (=121) discretisation points. XDSPACE, YDSPACE and ZDSPACE are originally set to zero then they will all be assigned the same value, as described above, so that the discretisation points are regularly distributed.
A set of discretisation points are generated within the rectangle surrounding the panel, and then the points lying within the panel are selected. If it is a 3D panel then points are also generated in the third dimension, limited by the DPLUS and DMINUS values. The total number of discretisation points lying within the panel is calculated and is compared to the minimum number of points as defined by parameter MINDISC. If the number of points is less than MINDISC then the spacing in each dimension is multiplied by 0.8, a new set of points is generated, and the total is again compared to MINDISC. This procedure is repeated until the total exceeds MINDISC. If insufficient points are found after 10 iterations then a warning message is displayed and the panel is not estimated. If this happens then reduce the spacing, and/or MINDISC, and try again.
If parameter PRINT=2 then the coordinates of the discretisation points will be displayed in the Output window. If PRINT=2 and ECHO=1 then the points will also be written to the print file. If the points are required for further analysis within Datamine, they can be imported directly from the print file.
Defining a Panel by a Set of Discretisation Points
Instead of generating the discretisation points from strings, as defined above, you can input the discretisation points directly using the DISPTIN file. In this case the parameters XDSPACE, YDSPACE, ZDSPACE, MINDISC, DPLUS and DMINUS are ignored. One method of creating the points could be to use the TRIFIL process to create a set of cells within an enclosed wireframe or below a DTM. The centre points of the model cells would then be the discretisation points, and so the panel would represent the volume within the wireframe or below the DTM. The output model from TRIFIL can be input directly to PANELEST as the DISPTIN file, specifying the XPT field as XC, YPT as YC and ZPT as ZC. It is recommended that if you use TRIFIL to create the discretisation points then you should not allow any subcelling, so that the points are on a regular grid.
The discretisation points could also be the cell centres of an existing model for which you have already interpolated grade. By selecting a subset of the cells, or the total model, you can estimate a single kriged grade and in particular a single kriged variance for that subset of cells. If the model contains subcells then it would be best to regularise the model first, so that the points are on a regular grid.
Selection of Samples
The MINNUM and MAXNUM parameters allow you to select the minimum and maximum number of samples to be used for making each estimate. For Nearest Neighbour and Inverse Power of Distance there is no limit on MAXNUM. However for kriging MAXNUM must not exceed 1399.
If less than MINNUM samples are found then the panel will not be estimated. If there are more than MAXNUM, then the number is reduced until only MAXNUM remain. This is done by calculating the distance of each sample from the nearest discretisation point. The samples are then sorted according to this distance, and the MAXNUM samples with the smallest distances are selected.
If panels are defined using the PERIM file, then setting parameter INSIDE=1 will ensure that only samples which lie within the panel are used for the grade estimate; if INSIDE=0 then all samples will be considered. If panels are defined using the DISPTIN file, then the parameter INSIDE is ignored.
Grade Estimation Method
The IMETHOD parameter allows you to select the estimation method. 1 = Nearest Neighbour. 2 = Inverse Power of Distance. 3 = Kriging.
(1) Nearest Neighbour
For each discretisation point the nearest sample is found. The panel estimate is then the arithmetic mean over all discretisation points. The reported variance is the classical statistical variance of all selected samples.
(2) Inverse Power of Distance
Each discretisation point is estimated using inverse power of distance, where the power is defined by parameter POWER. The panel estimate is then the arithmetic mean over all discretisation points. The reported variance is the classical statistical variance of all selected samples.
(3) Kriging
If parameter LOG=0 then normal kriging is used. If LOG=1 then log kriging is used, and the following conditions apply:
- General Case method
- a maximum of 3 iterations
- convergence tolerance is set at 0.01
- the kriged estimate is used as the mean value for the lognormal variance calculation
For further details of the log kriging method refer to the Grade Estimation User Guide.
If kriging is selected then the reported variance is the kriged variance. Note that this is different from Nearest Neighbour and Inverse Power of Distance which report the classical statistical variance of the selected samples.
Anisotropy
If Kriging is selected then anisotropy is defined by the parameters of the variogram model. For Nearest Neighbour or Inverse Power of Distance anisotropy is defined using the ANANGLEn, ANAXISn and ANDISTn parameters. Further details of this method are given in the Grade Estimation User Guide.
Totals and Averages
If parameter TOTAL=1 and more than one panel has been estimated, then the total area and/or volume and the average grade, weighted by area or volume will be reported in the Output window and saved to the OUT file. The total variance will only be reported if kriging has been used. It is calculated as the weighted average of the individual kriged variances, weighted by the square of the corresponding area or volume. In calculating the total variance it is therefore assumed that the estimates and variances of the individual panels are independent of one another. This will often be a reasonable assumption if the panels are large.
If the panels are defined using the DISPTIN file then the volume of each panel must be estimated. This is done by finding the minimum distance between discretisation points in each of the X, Y and Z directions and calculating the corresponding volume of influence of a point. The volume of the panel is then the volume of influence of a point multiplied by the number of points. This will give an accurate estimate of volume if the points are on a regular grid, but not otherwise.
Results
The results for each panel are displayed in the Output window, as illustrated below. If kriging has been selected and PRINT=1 then the panel F value (the variance of a point in the panel) and the Lagrange multiplier are also displayed. There are two output files, both of which are optional, examples of which are also shown below. The OUT file contains a single record for each panel, summarising the results. The SAMPOUT file shows the weight assigned to each sample for each panel. If kriging is used the SAMPOUT file also includes the average variogram value of the sample with the panel.
Average Sample-Panel Variogram Value
As described above, if kriging is used then the average variogram value of each selected sample with the panel is written to the sample output file SAMPOUT. Sometimes these values are required for further processing and the kriged estimate and variance are not required. The run time is then very much quicker if parameter VGONLY is set to 1, so that the kriged estimate and variance are not calculated. Another advantage is that the maximum number of samples per panel in the SAMPOUT file is increased from 1,399 to 50,000.
Timing
The execution speed of the process increases as the number of samples increases and also as the number of discretisation points increases. Also Kriging is significantly slower than Inverse Power of Distance.
Files, Fields and Parameters
Input Files
Name | I/O Status | Required | Type | Description |
IN | Input | Yes | Undefined | Input sample data file. This must contain the fields X , Y , and VALUE. A Z field is also required for 3D panels. |
VMODPARM | Input | No | Variogram - Model | Variogram model parameter file. |
PERIM | Input | No | String | The input string file. This must contain the 5 fields PANEL , PTN , XP , YP , ZP . The strings may be open or closed, but must be planar and must lie in one of the three orthogonal planes. Either a PERIM file or a DISPTIN file must be specified. |
DISPTIN | Input | No | Point Data | The input file containing discretisation points. This must include the 3 fields XPT , YPT , ZPT A fourth field PANEL is optional, and is used to identify different sets of discretisation points representing different panels. Either a PERIM file or a DISPTIN file must be specified. |
Output Files
Name | I/O Status | Required | Type | Description |
OUT | Output | No | Undefined | The output results file, containing a record for each panel estimated. The fields include the panel identifier, the estimated grade, the variance, and other associated information. |
SAMPOUT | Output | No | Undefined | The sample output file. This will contain the samples used to estimate each panel, and the weight assigned to each sample. |
Fields
Name | Description | Source | Required | Type | Default |
X | X coordinate of sample data in the IN file. | IN | Yes | Numeric | X |
Y | Y coordinate of sample data in the IN file. | IN | Yes | Numeric | Y |
Z | Z coordinate of sample data in the IN file. | IN | No | Numeric | Z |
VALUE | Field to be estimated in the IN file. | IN | Yes | Numeric | Undefined |
PANEL | Panel identifier. This is a numeric or alpha field (max 40 characters) in the PERIM or DISPTIN file. If a perimeter file is used then the PVALUE field may be used. | PERIM,DISPTIN | No | Numeric | PVALUE |
XPT | X coordinate of discretisation points in the DISPTIN file. | DISPTIN | No | Numeric | Undefined |
YPT | Y coordinate of discretisation points in the DISPTIN file. | DISPTIN | No | Numeric | Undefined |
ZPT | Z coordinate of discretisation points in the DISPTIN file. | DISPTIN | No | Numeric | Undefined |
Parameters
Name | Description | Required | Default | Range | Values | ||||||||
MINNUM | Minimum number of samples for panel to be estimated.(1). | No | 1 | 1,+ | Undefined | ||||||||
MAXNUM | Maximum number of samples for panel to be estimated. If kriging is selected then the maximum cannot exceed 1399. | No | 100 | 1,+ | Undefined | ||||||||
INSIDE | If set to 1 samples must lie inside panel. This only applies if a PERIM file has been specified. | No | 1 | 0,1 | 0,1 | ||||||||
XDSPACE | The distance between discretisation points in X. If set to zero then a suitable spacing will be generated automatically. | No | 0 | Undefined | Undefined | ||||||||
YDSPACE | The distance between discretisation points in Y. If set to zero then a suitable spacing will be generated automatically. | No | 0 | Undefined | Undefined | ||||||||
ZDSPACE | The distance between discretisation points in Z. If set to zero then a suitable spacing will be generated automatically. | No | 0 | Undefined | Undefined | ||||||||
MINDISC | Minimum number of discretisation points in panel before it can be estimated. | No | 50 | 1,+ | Undefined | ||||||||
DPLUS | Perimeter projection distance measured in the increasing direction of the perpendicular axis. | No | 0 | 0,+ | Undefined | ||||||||
DMINUS | Perimeter projection distance measured in the decreasing direction of the perpendicular axis. | No | 0 | 0,+ | Undefined | ||||||||
IMETHOD | Interpolation method: 1 - nearest neighbour 2 - inverse power of distance 3 - ordinary kriging | No | 3 | 1,3 | 1,2,3 | ||||||||
VMODNUM | Variogram model reference number as defined by the VREFNUM field in the VMODPARM file. (1). | No | 1 | Undefined | Undefined | ||||||||
LOG | Flag to indicate whether log kriging is selected. 0 = ordinary kriging 1 = log kriging | No | 0 | 0,1 | 0,1 | ||||||||
POWER | Power for inverse power of distance method. | No | 2 | Undefined | Undefined | ||||||||
TOTAL | If TOTAL is set to 1 then values for the total volume and area over all panels will be reported and saved to the OUT file. If kriging is selected then the average variance will be calculated as the weighted average of the individual variances, weighted by the square of the area or volume. | No | 0 | 0,1 | 0,1 | ||||||||
VGONLY | Flag controlling estimation (0).
| No | 0 | 0,1 | 0,1 | ||||||||
ANANGLE1 | First rotation angle for defining anisotropy when nearest neighbour or inverse power of distance methods are selected ie IMETHOD = 1 or 2. | No | 0 | -360, 360 | Undefined | ||||||||
ANAXIS1 | First rotation axis for defining anisotropy when nearest neighbour or inverse power of distance methods are selected ie IMETHOD = 1 or 2. This parameter has a value 1 for rotation about the X axis, 2 for rotation about the Y axis, and 3 for rotation about the Z axis. If it is set to 0 then there will be no rotation, irrespective of the value of ANANGLE1. | No | 3 | 0,3 | 0,1,2,3 | ||||||||
ANANGLE2 | Second rotation angle for defining anisotropy when nearest neighbour or inverse power of distance methods are selected ie IMETHOD = 1 or 2. | No | 0 | -360, 360 | Undefined | ||||||||
ANAXIS2 | Second rotation axis for defining anisotropy when nearest neighbour or inverse power of distance methods are selected ie IMETHOD = 1 or 2. This parameter has a value 1 for rotation about the X axis, 2 for rotation about the Y axis, and 3 for rotation about the Z axis. If it is set to 0 then there will be no rotation, irrespective of the value of ANANGLE2. | No | 1 | 0,3 | 0,1,2,3 | ||||||||
ANANGLE3 | Third rotation angle for defining anisotropy when nearest neighbour or inverse power of distance methods are selected ie IMETHOD = 1 or 2. | No | 0 | -360, 360 | Undefined | ||||||||
ANAXIS3 | Third rotation axis for defining anisotropy when nearest neighbour or inverse power of distance methods are selected ie IMETHOD = 1 or 2. This parameter has a value 1 for rotation about the X axis, 2 for rotation about the Y axis, and 3 for rotation about the Z axis. If it is set to 0 then there will be no rotation, irrespective of the value of ANANGLE3. | No | 3 | 0,3 | 0,1,2,3 | ||||||||
ANDIST1 | Anisotropy distance measured along rotated X axis, when nearest neighbour or inverse power of distance methods are selected ie IMETHOD = 1 or 2. This corresponds to the range of influence in that direction. | No | 1 | 0.0001,+ | Undefined | ||||||||
ANDIST2 | Anisotropy distance measured along rotated Y axis, when nearest neighbour or inverse power of distance methods are selected ie IMETHOD = 1 or 2. This corresponds to the range of influence in that direction. | No | 1 | 0.0001,+ | Undefined | ||||||||
ANDIST3 | Anisotropy distance measured along rotated Z axis, when nearest neighbour or inverse power of distance methods are selected ie IMETHOD = 1 or 2. This corresponds to the range of influence in that direction. | No | 1 | 0.0001,+ | Undefined | ||||||||
Output control flag (1).
| No | 1 | 0,2 | 0,1,2 |
Notes
No additional notes.
Example
!panelest &IN(holes.c),&VMODPARM(VMODEL),&DISPTIN(points),&OUT(RESNEW),
&SAMPOUT(sampout),*X(X),*Y(Y),*Z(Z),*VALUE(Fe),*PANEL(AZONE),
@MINNUM=1,@MAXNUM=1399,@INSIDE=0,@XDSPACE=0,@YDSPACE=0,@ZDSPACE=0,
@MINDISC=20,@DPLUS=0,@DMINUS=0,@IMETHOD=3,@VMODNUM=1,@LOG=0,
@POWER=2,@TOTAL=1,@ANANGLE1=0,@ANAXIS1=3,@ANANGLE2=0,@ANAXIS2=1,
@ANANGLE3=0,@ANAXIS3=3,@ANDIST1=1,@ANDIST2=1,@ANDIST3=1,@PRINT=0,
@VGONLY=0
Variogram model file VMODEL includes fields for up to 9 structures.
The file contains 2 records.
Model 1.00 has been retrieved from file.
Number of fully defined structures = 1
N Nugget Variance = 0.0000
All rotation angles are zero.
Structure Model Type Range X Range Y Range Z Spatial Var
T AX AY AZ C
1 1 100.0 100.0 100.0 1.0000
_______________________________________________________________________________
The grade estimation method is ordinary kriging.
Volumes will be estimated by finding the minimum
distance between discretisation points in the X,Y and Z
directions, calculating the volume of influence of a
discretisation point, and multiplying by the total number of
points. It is therefore assumed that the points are on a
regular grid.
_______________________________________________________________________________
AZONE 1.0000
Number of discretisation points = 845
Volume of panel = 845000.0000
Kriged grade = 46.2306
Kriged variance = 0.0432
Arithmetic mean of samples = 45.1471
Number of samples used = 183
_______________________________________________________________________________
AZONE 3.0000
Number of discretisation points = 160
Volume of panel = 159999.9844
Kriged grade = 50.8482
Kriged variance = 0.0644
Arithmetic mean of samples = 45.1471
Number of samples used = 183
_______________________________________________________________________________
AZONE 4.0000
Number of discretisation points = 25
Volume of panel = 12799.9512
Kriged grade = 47.5234
Kriged variance = 0.3584
Arithmetic mean of samples = 45.1471
Number of samples used = 183
_______________________________________________________________________________
AZONE 5.0000
Number of discretisation points = 90
Volume of panel = 6184.7759
Kriged grade = 43.6955
Kriged variance = 0.5730
Arithmetic mean of samples = 45.1471
Number of samples used = 183
_______________________________________________________________________________
Average / Totals Over All Panels
Total discretisation points = 1120
Total volume = 1023984.6875
Average kriged grade,all panels = 46.9530
Kriged variance, all panels = 0.0440
Total number of samples = 732
_______________________________________________________________________________
>>> 5 RECORDS IN FILE RESNEW <<<
>>> 732 RECORDS IN FILE SAMPOUT <<<
>>> PANELEST Completed <<<
_______________________________________________________________________________
======================================================================
FILENAME RESNEW
FILE CREATED BY SYSTEM USING PANELEST ON 99/06/03
----------------------------------------------------------------------
FILE CONTAINS 5 RECORDS EACH OF LENGTH 6
----------------------------------------------------------------------
FIELD TYPE WORD.NO STORED START DEFAULT
----------------------------------------------------------------------
AZONE N 1 Y 1 0.0
Fe N 1 Y 2 0.0
VARIANCE N 1 Y 3 -
NUMSAM N 1 Y 4 -
FVALUE N 1 Y 5 -
NDISCPTS N 1 Y 6 -
======================================================================
======================================================================
AZONE Fe VARIANCE NUMSAM FVALUE NDISCPTS
======================================================================
1.0 46.23064 0.04319 183.0 0.801666 845.0
3.0 50.8482 0.064417 183.0 0.513691 160.0
4.0 47.52341 0.358402 183.0 0.277084 25.0
5.0 43.69546 0.573013 183.0 0.234732 90.0
- 46.953 0.044021 732.0 - 1120.0
5 RECORDS LISTED
>>> LIST Completed <<<
_______________________________________________________________________________
======================================================================
FILENAME SAMPOUT
FILE CREATED BY SYSTEM USING PANELEST ON 99/06/03
----------------------------------------------------------------------
FILE CONTAINS 732 RECORDS EACH OF LENGTH 7
----------------------------------------------------------------------
FIELD TYPE WORD.NO STORED START DEFAULT
----------------------------------------------------------------------
AZONE N 1 Y 1 0.0
X N 1 Y 2 0.0
Y N 1 Y 3 0.0
Z N 1 Y 4 0.0
Fe N 1 Y 5 0.0
WEIGHT N 1 Y 6 -
AV-VGRAM N 1 Y 7 -
======================================================================
======================================================================
AZONE X Y Z Fe WEIGHT AV-VGRAM
======================================================================
1.0 363.625 480.5 38.0 40.3 0.011093 0.928843
1.0 340.5 373.0 29.5 40.3 0.008356 0.987864
etc
Error and Warning Messages
Message | Description | Solution |
|
|
|
|
|
|
