CoKriging
CoKriging is a method used when you want to use another attribute information besides the attribute used in the estimation of this attribute.
For example, information relating to ash can also be used when estimating the calorific value. It has been shown in the literature that CoKriging is more advantageous than kriging if there is a high correlation between two attributes and the attribute to be used in estimation (secondary attribute) is more than the attribute to be estimated (primary attribute). However, more modelling and estimation must be carried out than the kriging because more than one attribute is considered when cokriging. For example, variogram of the second attribute and also a cross variogram is needed besides variogram of the primary attribute in cokriging.
Block model to be estimated is clicked after Geostatistical Estimation CoKriging command. Then, one of "Simple Kriging" or "Ordinary Kriging" should be chosen. If simple kriging is chosen, averages relating to primary and secondary attribute must be entered in order leaving one blank between one another.
CoKriging types are determined after primary and secondary attributes are chosen. There are three options for this:
- Multiple cokriging,
- Markov Model 1,
- Markov Model 2.
Multiple cokriging, is a technique directly used in estimation of two attributes without any assumption. These technique is mostly considered in practical applications.
In Markov Model 1, it is assumed that the cross variogram is related to the variogram of the first variable. For this reason, modelling is lesser. When Markov Model 1 is chosen, correlation coefficient between two attributes and also variance related to the second attribute should be entered.
Markov Model 2, is used when the sample size of the primary attribute is smaller than the sample size of the secondary attribute. When Model 2 is chosen, correlation coefficient between two attributes must be entered.
Clicking "Assign the Data to Kriging" can assign the data to estimation points. This will speed up the estimation but reduce the precision in estimation values. It requires the entry of information relating to the geometry of the ellipsoid to be used in the estimation of search ellipsoid, point or block. Because two attributes are used in estimation, information regarding the search ellipsoid should be entered for both attributes. While the first search ellipsoid is designed for the primary attribute, information of the second search ellipsoid should be entered for the second attribute.
After entering the information regarding the search ellipsoid, variogram model parameters relating to the first attribute, cross variogram parameters between the first and the second attribute and lastly, variogram model parameters regarding the second attribute must be entered. After entering the parameters, estimation is done by clicking "OK" and blocks will be thematized.
