KNA - Block Size Optimization

To access this dialog:

Using the Advanced Estimation dialog, select Optimize from the left-hand menu system and then select the Optimize Block Sizes sub-panel.

This panel is used to help determine the optimum block size. A range of statistical parameters are calculated and a chart is displayed to show the relationship between each parameter and one of the block dimensions. In the above example theBlock Sizein X is shown on the X axis and theKriging Efficiencyon the Y axis.

This panel is only visible if Supervisor data is not being imported. You decide this using the Scenario Setup screen.

Statistical Parameters

The table below shows the statistical parameters that are reported for each run. These can be chosen from the Measured value (y-axis) select box:

Name	Field	Description
Average time per block (ms)	TIME_MS	Processing time for each model block
Corr(Z, Z*)	CORZZSTR	Correlation between actual value and estimate
Cov(Z, Z*)	COVZZSTR	Covariance between actual value and estimate
Cov(Z1, Z)	COVZ1SZS	Covariance between two estimates – multivariate case only
Kriged estimate	EST	Kriged block estimate
Kriging efficiency	KRIGEFF	Comparative measure of confidence in block estimate
Lagrange parameter	LAGRANGE	Lagrange parameter when solving kriging matrix
Number of samples	NUMSAMP	Number of samples used for block estimate
Search volume index	SINDEX	Search volume index used for block estimate
Slope of regression Z/Z*	SLPZZSTR	Slope of regression of actual value on estimate
Sum of pos. weights	SUMPOSWT	Sum of positive weights
Variance	VAR	Kriging variance (error variance)
Variance of Z*	VARZSTR	Variance of estimates
Weight of mean	WTOFMEAN	Weight assigned to mean of simple kriging

Note: Advanced Estimation is part of the Studio RM toolset. Additional licensing modules aren't required.

Field Details

The user needs to define the size of a model cell in each of the X, Y and Z directions. This is done by entering the minimum size, the number of intervals and the increment between successive values for each of the three directions as described in the Optimize section. In this example only the X block size changes while the block size in Y and Z is fixed at 10m.

The calculation of the statistics also needs the number of discretization points in each direction; these are defined by the Base values selected in the Optimize Discretization sub-panel. In addition the statistics need a set of search volume parameters; these are defined by the Base values in the Optimize Search Volume sub-panel. You should check the Base values in both sub-panels before running a test.

Example

Inputs- Variogram Model

The variogram model has been selected by double clicking the required model in theVariogram Modelarea of the Optimize panel. This example is based on a single structure spherical anisotropic model with the following parameters:

· Nugget: 0.35 Sill: 3.87 Rotation: 22.5^o around Z

· Ranges: X – 82.9, Y – 93.2, Z – 25.6

Inputs- Discretization

These parameters are defined by their Base values in the Optimize Discretization sub-panel:

The number of discretization points: 5 in X, 5 in Y, 3 in Z

Inputs- Search Parameters

These parameters are defined by their Base values in the Optimize Search Parameters subpanel:

The lengths of the search volume axes: 82.9m in X, 93.2m in Y and 25.6m in Z. The initial default values are set equal to the maximum variogram ranges in each direction.
Minimum number of samples: 5
Optimum number of samples: 10
Segment method: not applied

Note that in the graphic below only the Base values are used for optimizing the block size.

These parameters are defined in theOptimize Block Sizessub-panel:

Increment values will be rounded up to the nearest integer

Inputs - Combinations

TheTest reduced combinationsoption is selected making a total of 8 KNA runs:

The combinations of values to be tested are:

ClickRun Teststo start the KNA runs.

Outputs – Slope of Regression

The Chart Option Group on separate tabs by has been selected as Test block group so the results for each location are shown on different tabs. The results for the Well informed location are shown here. It can be seen that the mean slope of regression(green line) has a value of 1 for a block size in X of 10m but reduces as the block size increases.

The corresponding charts for the Moderately informed and Sparse locations are shown below. The mean regression slope for Moderately informed is similar to Well informed, but the mean slope for Sparse is considerably lower reflecting the lower sampling density.

	Related Topics
	Select Locations Optimize Optimize Discretization Optimize Search Parameters