Semivariogram - Modeling or Adjustment

This page describes the experimental semivariogram adjusting module in two different ways: Automatic or Visual.

  • The automatic mode uses the Olea et al. (1996) algorithm, which is based on least square method. This algorithm gives also the quantitative measure, denominated by Akaike information ( Akaike, 1974), which reports which model allows a better adjust.
  • The visual mode is recommended to specialists which have knowledge and are used to the studied phenomenon. In this mode, all the parameters are defined by inspection.
  • Besides the adjusting procedures, Automatic or Visual, this module defines a semivariogram theoretical model to be used by the Validation and Kriging modules.

The experimental semivariogram adjustment or modeling starts after the Semivariogram Generation. See the following procedures.

Executing the experimental semivariogram adjustment:

  • select on the Analysis option, Geostatistics -> Semivariogram Modeling...;
  • the "Semivariogram Modeling" window is presented;
  • select the Adjusting: Automatic or Visual type;
  • if the Adjustment was Automatic, define the Number of Structures (1, 2 or 3) required for the experimental semivariogram adjustment;
  • if the Adjust was Automatic, define the theoretical models using the selection buttons Model 1, Model 2 and Model 3 in the window. Select for each model one of the options Spherical, Exponential, Power, and Gaussian. See about the theoretical models;
  • press the Apply button and notice two things: that the Define... button is enabled to establish the model parameters; and the item(s) in the Adjust Verification list is active.


Adjustment Verification and Model Parameters definition:

  • press on an item in the Adjustment Verification lists, and notice that the "Adjusting Model" graphical window will be presented and the "Data Report" window.
  • The "Adjusting Model" window presents graphically the theoretical adjusting model (in black) over the experimental semivariogram (white points). Visually it is possible to say if the adjusting is good or not, as presented in the figure below.

  • The "Data Report" window presents a set of information, such as: the theoretical model type selected, the values of Lump Effect, Contribution and Range which are model compounding parameters. It is also expressed the Akaike value, which is an adjustment indicator taken; the smaller the value the better the adjustment. Then, the Lump Effect, Contribution and Range parameters are always taken related to the smallest Akaike value, as presented in the Figure below.

  • The next step is to define the semivariogram model based on the information inside the "Data Report" window. In order to do that, press the Define... button.


NOTES:

See also:
SPRING - Spatial Analysis
Theoretical Models for the experimental variogram adjustment
How to Execute? - Semivariogram Generation
How to Execute? - Structural Parameters Definition
How to Execute? - Model Validation