Kriging Neighborhood Analysis (KNA)

Kriging Neighborhood Analysis (KNA) is meant for neighborhood parameters optimization. This is an optional step to be run prior to kriging. It allows the quick evaluation of the quality of a neighborhood in order to select the most adequate one regarding the actual input dataset and variogram.

The principle of KNA is to perform a series of kriging. Several geostatistical criteria are then computed in order to classify each kriging run/test and so select the best neighborhood parameter combination. Special kriging options (local anisotropies, faults, filtering...) are ignored when running KNA. Only the calculation mode, point or block, is taken into account.

To access this option, click on the KNA button in the Moving Neighborhood Parameters section. If the selected output data table is a 3D grid, KNA can be applied on a grid section (XOY, XOZ or YOZ with a Level specified in the Section part of the interface) or on the whole file. If the output data table is not a 3D grid, the KNA is calculated considering the whole file. After choosing the mode, the Kriging Neighborhood Analysis Panel pops up. The current neighborhood should be set to a moving neighborhood.