Binary Value Kriging
Binary Value Kriging is a method used in the calculation of the possibility essentially related to the unknown value in one point.
For example; instead of estimating an unknown value in a non-sampled point, it may be required to calculate the possibility of it being higher or lower than a specified value. In these cases, limit value kriging method is used. For this, original data (composites) is coded with 1s and 0s according to the considered limit value. If the original value is lower than the limit value, it is a 1, otherwise it is a 0. Then, the variogram of the data, which is coded in 1s and 0s, is calculated and modelled. Values in non-sampled points are estimated using this converted model and limit values (Binary Value). This estimation values give the possibility of the variable being lower than the limit value.
Binary Value estimation can be done for more than one limit value. In this way, user can model the complete cumulative distribution function. For this, if there are not any technical or economical limitation, nine limit values are chosen monotonically representing the original data distribution. In general, these limit values are equal to the decile values of distribution. For every limit value, limit value (Binary Value) variogram is modelled and then, the possibility of them being low is independently estimated with limit value kriging. As a result of this, values of cumulative distribution function of nine limit values are obtained. This is an inconvenient and a time-consuming operation. Instead, only the Binary Value variogram model at the median limit value is extracted and this model can be regarded as variogram model for all limit values. This method is called median limit value (Binary Value) kriging. However, variograms in all other limit values must be the same as the variogram of median limit value in order to use this method.
