|
Process Help |
|
Process Name |
Menu Path |
Link to Command Table |
|
DECLUST |
Command line only |
Introduction
|
This is a Superprocess and running it may have an effect on other Datamine files in the project. More... |
The DECLUST process declusters a set of sample data.
|
|
It is common practice to take more samples in high grade areas in order to improve the level of confidence in the estimation. However, when data are not on a regular grid then using the full data set will give a biased estimate of the mean, variance and histogram. In addition, clustered data can affect an indicator variogram. For further details refer to "Non-parametric estimation of spatial distributions" by A.Journel published in Mathematical Geology, volume 15, number 3, 1983. |
How to use
Declustering is the process of adjusting the full data set i.e. by removing or weighting data points in densely sampled areas, to give a more representative and evenly spaced set of samples. The DECLUST process provides two methods of doing this, with both methods requiring a regular 3D grid to be placed over the sample data. The two methods are:
- Sample Selection - select a single sample from each grid cell.
- Declustered Weight - assign every sample a weight based on the number of samples in the grid cell
If the Sample Selection method is chosen then DECLUST provides a choice of four ways of assigning the value to the grid cell:
- select a sample at random within each grid cell. A new random sample is generated for each run.
- select a sample at random within each grid cell. The same 'random' sample is generated for each run.
- select the sample nearest to the grid cell centre.
- calculate the average value of all samples within the grid cell.
If the Declustered Weight method is chosen then the weight, DCWEIGHT,
for a sample is calculated as:
DCWEIGHT = NDATA / NCELLS
/ NPERCELL
where:
-
NDATA is the total number of samples
-
NCELLS is the number of grid cells containing one or more samples
-
NPERCELL is the number of samples in the grid cell
The sum of the weights over all samples equals the total number of samples (NDATA). Therefore if a sample lies in a high density area it will have a weight of less than 1, and if it is in a low density area it will have a weight of more than 1. The output file from the Declustered Weight method can be used to transform data into a normal distribution, for input to the NSCORE or SGSIM processes.
One of the problems with the declustering method is that different grid sizes will generate different statistics. However, in general a regular grid about the size of the average sample spacing is suggested.
The process writes a summary table for each method to the Output control bar.
-
Sample Selection: shows statistical parameters for each numeric field in the output file. These statistics can be saved to a file if the STAT_TBL output file is specified.
- Declustered Weight: shows the declustered weight
as a function of the number of samples per grid cell. These statistics
can be saved to a file if the WGTS_TBL output file is specified.
Files, Fields and Parameters
Input Files
|
Name |
Description |
I/O Status |
Required |
Type |
|
IN |
Input sample data file. This must contain a set of 3D coordinates (eg X,Y,Z) and at least one other field. |
Input |
Yes |
Undefined |
Output Files
|
Name |
I/O Status |
Required |
Type |
Description |
|
OUT |
Output |
No |
Undefined |
Output file containing declustered samples. At least one of the two output files OUT or WTOUT must be selected. |
|
WTOUT |
Output |
No |
Undefined |
Output file containing declustered weights. This will be a copy of the IN file, but will also include the field DCWEIGHT. At least one of the two output files OUT or WTOUT must be selected. |
|
WGTS_TBL |
Output |
No |
Undefined |
Output file containing summary statistics for declustered weights. |
|
STAT_TBL |
Output |
No |
Undefined |
Output file containing summary statistics for declustered and clustered WTFIELD samples. |
Fields
|
Name |
Description |
Source |
Required |
Type |
Default |
|
X |
X coordinate of sample data |
IN |
Yes |
Numeric |
X |
|
Y |
Y coordinate of sample data |
IN |
Yes |
Numeric |
Y |
|
Z |
Z coordinate of sample data |
IN |
Yes |
Numeric |
Z |
|
WTFIELD |
Field to be used for calculating declustered weights. This is only relevant if a WTOUT file has been specified and one or more of the grade fields in the IN file contain absent data values. Specifying a WTFIELD field ensures that records containing absent data values for that field will be ignored. If a WTFIELD field is not specified but a WTOUT file has been selected then the Z field is used. |
IN |
No |
Numeric |
Undefined |
Parameters
|
Name |
Description |
Required |
Default |
Range |
Values |
|
METHOD |
Declustering method if OUT file specified: 1 = random selection within grid (different selection each run) 2 = pseudo random selection within grid (repeatable) 3 = nearest to grid centre 4 = average of samples within grid |
No |
1 |
1,4 |
1,2,3,4 |
|
XGRID |
Grid size in X |
Yes |
Undefined |
0.00001,+ |
Undefined |
|
YGRID |
Grid size in Y |
Yes |
Undefined |
0.00001,+ |
Undefined |
|
ZGRID |
Grid size in Z |
Yes |
Undefined |
0.00001,+ |
Undefined |
|
XORIG |
X coordinate of grid origin |
No |
0 |
Undefined |
Undefined |
|
YORIG |
Y coordinate of grid origin |
No |
0 |
Undefined |
Undefined |
|
ZORIG |
Z coordinate of grid origin |
No |
0 |
Undefined |
Undefined |
|
CENTRE |
Flag to show whether the X, Y and Z coordinates of the grid centre should be included in the OUT file. If selected the names of the fields in the file will be XCENTRE, YCENTRE and ZCENTRE: |
No |
0 |
0,1 |
0,1 |
Notes
The process DECLUST has a limit of only being able to use input sample files with a maximum of 53 fields. If the process is run with an input sample file of greater than 53 fields, a warning message is displayed which prompts the user to reduce the number of fields.
Example
|
!DECLUST
|
&IN(sampleb),&OUT(dsampleb),&OUT(dcwts), *X(X),*Y(Y),*Z(Z),@METHOD=2, @XGRID=10,@YGRID=10,@ZGRID=10,@XORIG=0,@YORIG=0,@ZORIG=0,@CENTRE=0 |
DECLUST - Decluster a set of sample data.
... input validation
... declustering using pseudo random selection method (2)
... sorting by grid position
FORMAT TIME >12:38:28
______________________________________________________________________________
SUMMARY STATISTICS FOR DECLUSTERED FILE dsampleb
================================================
Input file sampleb contains 1077 records
Output file dsampleb contains 325 records
Field
Number Samples
Number Absent
Number >0
Minimum
Maximum
Mean
Variance
St.Dev
X
325
0
325
8645.74
8750.43
8699.92
672.63
25.94
Y
325
0
325
3358.24
3466.30
3407.14
605.41
24.61
Z
325
0
325
159.65
244.96
208.47
294.48
17.16
325
0
325
4.00
7.00
4.40
1.03
1.02
AU
325
0
325
0.80
13.90
5.98
6.87
2.62
CU
325
0
325
0.27
2.99
1.50
0.41
0.64
ROCK
325
0
325
1.00
2.00
1.13
0.11
0.34
______________________________________________________________________________
... calculating declustered weights
FORMAT TIME >12:38:44
_____________________________________
STATISTICS FOR DECLUSTERED WEIGHTS
==================================
1077 samples assigned declustered weight
Samples per Grid Cell
Declustered Weight
No. of Grid Cells
1
3.31
88
2
1.66
140
3
1.10
165
4
0.83
128
5
0.66
110
6
0.55
84
7
0.47
119
8
0.41
104
9
0.37
63
10
0.33
30
11
0.30
33
13
0.25
13
_____________________________________
... declustered file dsampleb
contains 325 records
... declustered weights file dcwts
contains 1077 records
... process complete
Error and Warning Messages
|
Message |
Description |
Solution |
|
|
The input sample file has greater than 53 fields. |
Reduce the number of fields in the input sample file e.g. by using SELCOP or SELDEL to generate a new file. |
|
|
|
|
