Simulation by Indication

This window presents the Simulation by Indication which aims at obtaining a regular grid of values acquired from point sample data. This grid is generated from a set of simulated grids following the procedure of sequential simulation described by Deutsch e Journel, 1992.

This module was implemented in SPRING based on the indicator kriging using the "sisim" code from GSLIB - Geostatistical Software Library (Deutsch and Journel, 1992). This module relies on measures of spatial variability according to regular grids, of spatial attributes, continuous or categorical. Along with an attributes map, an uncertainty map is generated with an estimation of a regular grid representation.

Executing Simulation by Indication :

  • Select on the "Control Panel" a numeric IL that contains point samples;
  • On the SPRING main menu choose Analysis - Geostatistics - Simulation by Indication...; Two windows will be presented: Model Parameters and Kriging by Indication. The window Model Parameters can also be activated by pressing the button Models/Probability... inside the window Kriging by Indication. Follow the instructions described at Models/Probability...to fill out the fields;
  • On top of the window "Simulation by Indication" you will see the variable correspondent to the active IL;
  • the structural parameters can be changed at any time by clicking the button Models/Probability...;
  • define the attribute nature by choosing Continuous or Categorical on the field Variable;
  • choose between Ordinary or Simple at the field Kriging Type;
  • Define an option for Kriging by Indication: Full IK or Median. If the Option isMedian, type the median value at the field Threshold;
  • Press the button Prior Prob. ... to define a set of prior probabilities valid for the points of the grid to be created. This option is not implemented yet;
  • the button Indirect Data... must be activated in case you want to choose a file with auxiliary data of attributes that are correlated with the attribute being studied. These indirect data must be in a file with Geo-EAS ( Deutsch e Journel, 1992 ) format, X and Y coordinates on the columns 1 and 2 respectively, and indirect attribute values already transformed by indication, placed on the column 3.
  • Press the Bounding Box... button in case you want to define a new area for the grid to be generated. The default value is the active Information Layer;
  • The fields X Res. and Y Res. contain predefined values for the active IL. Change these values in case you want to use different resolution. Notice that the values defined for the bounding box and resolution determine the size of the grid and can not be greater than 1000 rows or 1000 columns;
  • The parameters of Maximum Samples in the Searching Area must be defined by filling out the fields: Original which determines the maximum number of original samples of the attribute, Prev. Nodes which defines the maximum number of previously simulated nodes of the grid and Soft Nodes which determines the maximum number of indirect nodes inside the searching area. By default, these fields have predefined values of 8, 4 and 4, but the user can change them if he wishes.
  • Next, define the radius and orientation for the Searching Ellipsoid. The fields R.min, R.max and Angle have predefined values for the Isotropic case: R.min and R.max (in meters) are equivalent to the reach of the isotropic variogram and Angle is equal to zero. If there is anisotropy, the parameters must be adjusted accordingly (Deutsch e Journel, 1992);
  • The Simulation Parameters must be defined in a sequence. Fill out the field N. Realiz. which corresponds to the number of times the simulation was executed. Define the minimum (Min. Value) and maximum (Max. Value) values for the attribute, to be used by the probability distribution function. Provide an initial value for the simulation at the Seed: field. The type of search (simple, multiple 2, multiple 4, multiple 6, multiple 8 or multiple 10) determines the number of previously simulated grids (1, 2, 4, 6, 8 or 10 respectively). Finally you should decide if you want or not to make a previous association of the sample values to the nodes, by marking or not the box of Data in Nodes. This option helps to optimize the original data search decreasing the simulation total time.
  • As outputs of Kriging by Indication we have two Information Layers with regular grid representations, one with attribute values and another with estimation uncertainties. To define these ILs it is necessary to choose a Category, type in a name for the IL, IL values (different from the active IL), choose the type of value and uncertainty. For continuous attributes you should choose for Value mean or median and for Uncertainty confidence intervals based on standard deviations or quartiles. For categorical attributes, the distribution mode is chosen for Value and the maximum probability or Shannon entropy for Uncertainty . The uncertainty Information Layer has a _Inc suffix added to the IL values field.
  • Click on Apply.

[ Procedures] [ Concepts] [ Exploratory Analysis ] [ Semivariogram Analysis ]
[ Semivariogram Modeling] [ Model Validation ] [ Kriging]

 

See also:
SPRING - Spatial Analysis
Theoretical Models for adjusting experimental variogram