Advanced Estimation & Variography - Introduction

The Advanced Estimation (AE) module provides a range of methods for estimating grades into a block model including both univariate and multivariate kriging methods. With univariate kriging each block estimate is calculated from the sample values for a single variable (grade). The multivariate cokriging method uses values from other grades that are correlated to the grade being estimated.

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

The two main cases where cokriging can improve the estimates (i.e. reduce the kriging variance compared to univariate kriging) are:

A variable is poorly sampled but correlates with a secondary variable that is much better sampled
A variable has poor low spatial continuity, but correlates with one that has better continuity.

In both cases the secondary variable(s) should help to improve estimates of the first variable, particularly in the first case. Cokriging can be used if there is a full set of data for both the primary and secondary grades, but there is often very little or no improvement in the estimates.

Multivariate estimation requires the variables to be well correlated, either positively or negatively. The mutual spatial behavior of regionalized variables is known asco-regionalization. Cokriging requires the same conditions to be satisfied as kriging does, but needs additional variography and modelling tools which are provided in the AE module.

Advanced Estimation: 2D vs. 3D Data

Advanced estimation can be performed using 2D or 3D data. Where samples with 2D data have been selected, panel behavior will be adjusted in some cases to ensure inputs the function are valid.

For example, axes for which it is not possible to rotate around are disabled when computing the variograms and fitting the models, and the Investigate Anisotropy panel will only display a single isocontour plot where 2D data has been specified. Other adjustments are made internally to ensure 2D data is processed correctly, particularly with regards to calculations involving rotation.

AE provides many additional features and functions that are can be used for both univariate and multivariate estimation. These include:

Guided Workflow: an integrated set of dialogs that lead the user through the steps required for a univariate or multivariate estimation study. In addition, all the standard Studio commands are available from the ribbon menu system.
Scenario Management: create a new estimation scenario, save a copy of the current scenario, edit a previously saved one or export/import a scenario for use in another project.
Supervisor: import variogram models directly from a Datamine Supervisor project file.
Custom Zones: allows selected zones to be merged into custom zones for soft boundary estimation.
Bivariate Data Analysis: to calculate and display bivariate statistics by domain for each pair of grades.
3D Variogram Model: create a 3D block model of variogram values and use the new variogram-specific task view to determine the directions of anisotropy.
Variogram Calculation: for both variograms and for cross-variograms if cokriging
Variogram Model Fitting: this includes dynamic lag control and automatic fitting options, with a choice of model fitting types (Variogram, Log Variogram, Madogram etc.)
Kriging Neighbourhood Analysis (KNA): to assist in the selection of optimal sample search volume and estimation parameters.
Estimation Methods:include ordinary and simple kriging and cokriging, nearest neighbour and inverse distance methods.
Parameter Management: includes methods to import existing COKRIG parameter files, including files compatible with ESTIMA into the Advanced Estimation dialog
Worked Example: as well as providing contextual help information for each part of the dialog, there is also a worked example and description.

If you have access to multiple monitors, float the Advanced Estimation dialog to a spare monitor so the application can be seen in full throughout the process.

The Advanced Estimation dialog can also be resized.

Datamine Supervisor Support

Your Advanced Estimation console can be used to prepare and execute uni- or multivariate estimations against any input sample data, including variography and Kriging Neighbourhood Analysis (KNA).

Alternatively, you can run variography and KNA using Datamine Supervisor instead of running them in Studio. If you choose to use Supervisor data, all variography and KNA functionality is hidden, showing you only the panels you need to complete your estimation. Set up your scenario in Studio RM, import parameter files then define your estimation and run it. You need to decide on your workflow up front, using a switch on the Scenario Setup panel.

Importing project data from Supervisor will automatically populate the Define Estimations and Define Search Volumes panels by appending them with data derived from the imported project. Variograms from the Supervisor project will also (if possible) be automatically assigned to the created estimations.

You can also extract scenario data from a Datamine Supervisor project (the project must be from Supervisor v8.14.2.3 or later). Just pick your Supervisor project using the Import option on the Scenario Setup panel. The imported project must contain a Datamine-compatible variogram.

These Supervisor integration options are independent; for example, you may choose to complete your variography in Datamine Supervisor and fill in the scenario details yourself, or use the Advanced Estimation console for variography but import scenario details from a Supervisor project, or any combination of these methods.