Académique Documents
Professionnel Documents
Culture Documents
Page 1
This same procedure outputs a detailed report with detailed composite information (as used in interpolation). See figure 2.
Kriging matrix:
0.0850 0.0112 0.0607 0.0607 0.0118 0.0103 0.0473 1.0000 0.0112 0.0850 0.0104 0.0118 0.0585 0.0584 0.120 1.0000 0.0607 0.0104 0.0850 0.0473 0.0105 0.0098 0.0368 1.0000 0.0607 0.0118 0.0473 0.0850 0.0128 0.0104 0.0607 1.0000 0.0118 0.0585 0.0105 0.0128 0.0850 0.0438 0.0135 1.0000 0.0103 0.0584 0.0098 0.0104 0.0438 0.0850 0.0103 1.0000 0.0473 0.0120 0.0368 0.0607 0.0135 0.0103 0.0850 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
0.0283 0.0241 0.0275 0.0275 0.0238 0.0228 0.0252 1.0000
Covariance between samples Kriging Variance = 0.05082 # PTS = 7 Lagrange Multiplier = -0.01091
Kriggin estimate, kriging variance, distance to the closest composite, # if composites, # of drill holes Local Error (Variance) = 0.000102 Row: 75 5490.0 Comps: 1 64 #Calc 1 End of BENCH 35 2435.0 #Blocks Calculated 1 100.0% Completed Figure 2. Detailed Kriging Debug Report
Page
2. Ellipsoidal Search Versus Variogram Ranges There are two places where you will be asked to define length of a major, minor and vertical axes. One is for the ellipsoidal search and applies only to the selection of the composites (see figure 3).
Page 3
The second one is for the variogram ranges and it applies only to the calculation of the kriging weights (see figure 4).
Figure 4. Variogram Parameters Panel Those ranges are likely the same. If not, try to use the same major versus minor versus vertical length relationship. 3. Ellipsoidal Search Versus Primary Search Parameters The primary search parameters define a 3-D volume around a block (see figure 5 on next page). They are used to initially screen out the composites to be used for a given block. Four major search parameters have to be defined: X-search parameter (par1) Y-search parameter (par2) Z-search parameter (par3) Max 3-D search parameter (par4)
Page
The search for the composites around the block includes the volume from the center of the block the distances in x, y, and z directions. If the max 3-D search distance (par4) is less than or equal to the smallest of par1-3, then the search volume has been reduced to a sphere (see figure 5).
Figure 5. Major search parameters If par3 is smaller than the bench height, the search volume is reduced to a 2-D plane rectangular or circular search. The ellipsoidal search follows the primary search, however this is optional. If you want to have a true ellipsoid, then the search box around the block should be big enough to include the ellipsoid; otherwise the smaller box will truncate the ellipsoid. The safest way to apply this is to ensure par4 is equal to the length of the major axes of the ellipsoid. See figure 6.
Page 5
4. Effect of Weighting Factor The most common use of a special weight during estimation is to adjust the weights to take into account aberrant composite lengths or densities. If a composite is only defined for part of its length, then it should only contribute to the interpolation according to its length. In M624V1 the composite length weighting is done after the Kriging weights are computed as follows: Grade = [sum (grd x Kwt x wt)] / [sum (Kwt x wt)] grd = grade wt = weighting item Kwt = Kriging weights In the debug report, the program prints the kriging weights before using the weighting factor. Therefore you see the original weights. If you use weighting factors in IDW, the grades will be calculated as follows: grd1 x (wt1 /d1)2 + grd2 x (wt2 /d2)2 (wt1/d1)2 + (wt2/d2)2 where grd = grade d = distance wt = weighting item A new option was added to apply a weighting factor from a weight item after using the inverse distance: grd1 x wt1 /(d1)2 + grd2 x wt2 /(d2)2 wt1/(d1)2 + wt2/(d2)2
Page
5. Search from Mid-Point of the Bench If the check for fixed length/ZMID composites switch is ON, this indicates the bench midpoint elevation will be used. This option will work if the composites are indeed stored at the bench midpoint elevation option (see figure 7).
Figure 7. Composite Stored at Mid-bench. Note: Center of block to mid bench is correct distance. ZMID option is needed.
Page 7
The reason for this is to take into account that in bench composites, although composites are stored at the toe or the mid bench, they represent the grade of the whole bench. Therefore, the distance to the block needs to be measured from the center of the composite interval to the center of the block (or from the toe of the composite to the toe of the block; see figure 8).
Figure 8. Composite Stored at Toe. Note: Toe to toe is correct distance. No need for ZMID option.
Page
If fixed length composites are used, use the center of the block for distance measurements (see figure 9).
Figure 9. Composite Stored at Mid Bench. Note: Center of block to mid-bench is correct distance. ZMID option is needed. Without the mid-bench option, programs will calculate distance from toe of block. If the mid-bench option is used, programs will calculate the distance from the center of the block.
Page 9
Figure 10. Panel for Geologic Matching Assume the following composites fall within the search parameters:
Composite #
Value
Rock Type
1 2 3 4 5 6
50 60 70 80 90 100
1 2 2 1 1 2
If you want to use block rock type = 2, max # of composites allowed = 3, and max distance to the closest composite = 55 , if you use the option above the block will not be interpolated.
Page 10
7. Rotation Changes Primary search distances are with respect to model coordinates (see figure 11)
Figure 11. Primary Search in a Model Rotated 45 degrees Options for angles in variograms and ellipses recently added to accommodate the model rotation updates include:
Page 11
1.
Variogram rotation angles can be defined with respect to Project Coordinates or Model Coordinates (see figure 12).
Page 12
2.
Search rotation angles can also be defined with respect to Project Coordinates or Model Coordinates (see figure 13).
Page 13
8. Octant/Quadrant Searches An octant/quadrant search can now be defined with respect to project axes or anistropic axes (see figures 14, 15 and 16).
Page 14
Page 15
9. Alternative Coordinate Items The composite coordinate items are used to determine the position of composites in space. The default coordinate items retrieved from the composite file are EAST, NORTH, ELEV. (3-D Block Models). You can now optionally use other composite items such as XBOT, YBOT, and ZBOT (see figure 17).
Page 16
10. Outliers The maximum search distance can be changed for these composites greater or smaller than a specified cutoff. See figure 18.
Figure 18. Panel for Outliers PAR31 = Outlier cutoff to change the search distance PAR32 = Max 3-D distance for outliers This means that composites greater than PAR31 will be included in the calculation only if the distance to the block is less than PAR32. If PAR32 is negative, then the outlier high grades are cut-down to PAR31. PAR33 = Low Grade cutoff to change the search distance PAR34 = Max 3-D distance for low grades This means that composites less than PAR33 will be included in the calculation only if the distance to the block is less than PAR34. All of the above logic was added to both the kriging and inverse distance programs.
Page 17
11. Variograms in MineSight 3-D In MineSight Compass, you can run procedure PVGCTR.DAT to create a 2-D contour map of variogram values. If you want to see those contours in MineSight 3-D and use colors for different variogram values you can load them to a grided surface (file 13 item). You can then make a colored MineSight 3-D model view of the variogram values. Procedure PVGCTR.DAT outputs an ASCII points file (x, y coordinates along with the variogram values). This file can be loaded to file 13 within the procedure (see figure 19). You need to have a file 13 item with the appropriate minimum and maximum variogram values. If you normalize the values before contouring, the maximum value of the item needs to be 100.
Page 18
After you run the procedure, create a file 13 view in MineSight 3-D. The best way to view the contours is on a -90 degrees dip view with the viewer lights off (see Figure 20).
Figure 20. Variogram Contours in MineSight 3-D 12. Composite Data Distance Analysis Mintec now has a new program to perform composite data distance analysis to be used in, among other things, the interpolation search strategy. Procedure P52201.DAT can be found under group composites/operation calculations. On the following pages is a sample report and a graph of bench elevation per average distance:
Page 19
Sample Report
Composite Data Distance Analysis by Bench Based on the Specified Search Window Moving From Point to Point: Max. Search Distance: PAR4 = 100.0 Max. No. of Data Points = 8 Available Data Distance Analysis Bench No. of Item Nearest Average Toe Data Average Distance Distance 2660.0 4 0.447 40.86 60.94 2645.0 30 0.193 39.30 68.41 2630.0 73 0.333 42.05 69.65 2615.0 81 0.554 41.31 67.97 2600.0 58 0.698 46.52 72.88 2585.0 49 0.610 47.90 75.74 2570.0 48 0.585 50.92 76.89 2555.0 44 0.755 50.77 76.00 2540.0 45 0.798 50.47 75.71 2525.0 48 0.700 50.86 75.22 2510.0 50 0.753 51.82 75.00 2495.0 50 0.760 52.18 74.62 2480.0 47 0.799 52.00 74.43 2465.0 38 0.831 52.76 76.31 2450.0 33 0.803 53.54 77.17 2435.0 26 0.848 52.36 75.31 2420.0 25 0.733 49.53 76.35 2405.0 28 0.799 47.39 74.77 2390.0 28 0.811 45.27 73.40 2375.0 26 0.825 44.55 72.80 2360.0 12 0.723 52.97 74.64 2345.0 8 0.731 51.79 74.60 TOTAL 851 0.670 48.19 73.68
Page 20
13. Multiple Indicator Kriging Program Using GSLIB Routines. Mintec now has a new Multiple Indicator Kriging program using the GSLIB routines. One advantage of this program over the standard M624IK is the user can specify the extrapolation type assumed at either end of the distribution. In M624IK, only the linear extrapolation is assumed. The program can be found in the MineSight Compass menu under group MIK.
Page 21
Page 22