Cokriging with Seismic

Help From a Dense Secondary

Wells are few; seismic is everywhere. If a seismic attribute such as an impedance or an inversion correlates with porosity, it is a secondary variable that can guide the model between the wells. Cokriging brings it in: it honors the wells exactly, as kriging does, but where kriging would fade toward the global mean, cokriging leans on the seismic.

The Correlation Sets the Weight

How much the seismic is trusted depends on how well it correlates with porosity. A strong correlation lets the seismic carry real detail into the gaps and the error against the truth drops sharply, as the widget shows. Collocated cokriging is the common, efficient form: it uses the seismic value at the cell being estimated, plus the correlation, rather than the entire secondary dataset.

Garbage In

The benefit is only as good as the correlation. A weak or spurious porosity-seismic relationship adds noise dressed up as detail, and cokriging then does worse than plain kriging. The discipline is to establish the correlation at the wells first and to trust the seismic only as far as that correlation justifies.