Understanding the Different Types of Kriging in Mining: A Beginner’s Guide

kriging illustration

When you start learning geostatistics in mining, one of the first techniques you'll encounter is Kriging. It’s a powerful tool that helps geologists and mining engineers make better decisions based on limited data, especially when it comes to estimating the distribution of minerals underground.

But Kriging isn’t just one method — there are several types, each with its own strengths and purposes. If you’re a beginner, this article will help you understand why different types of Kriging exist, and when to use them.

What is Kriging (in simple words)?

Kriging is a statistical interpolation method used to estimate the value of a variable (like gold grade, copper concentration, etc.) at unsampled locations based on data from surrounding drill holes or samples.

It’s more than just averaging: Kriging considers both the distance between samples and the spatial structure (variability) of the data, captured through something called the variogram.

Why Different Types of Kriging?

Not all data and geological situations are the same. Some deposits are well-drilled and easy to model; others are sparse, complex, or risky to estimate. Different types of Kriging help adapt to these realities.

Think of Kriging types as different tools in a toolbox — you pick the right one depending on:

• How much data you have

• How variable the deposit is

• What kind of decision you’re trying to make (planning, reporting, exploration)

The Main Types of Kriging in Mining

1. Simple Kriging (SK)

Assumption: You already know the global mean of the variable across the whole deposit.

Usage: Rare in mining because we usually don’t know the global mean with certainty.

Advantage: Mathematically straightforward.

Limitation: Unrealistic in real-life geology.

2. Ordinary Kriging (OK)

Assumption: The local mean is unknown but assumed constant within the search neighborhood.

Usage: The most common Kriging method in mining.

Why? It doesn’t require knowing the global mean, just relies on local data.

Best for: Estimating resources where you have moderate to good drilling density.

3. Universal Kriging (UK)

Assumption: The mean is not constant but changes gradually over space (like a trend).

Usage: When there’s a clear geological trend (for example, increasing grade along a vein).

Best for: Large, structured deposits with visible spatial trends.

Extra: Includes modeling a drift or trend surface in addition to the variogram.

4. Indicator Kriging (IK)

Assumption: Rather than estimating precise values, it estimates categories (lithos, domains, probability of exceeding a threshold, etc...).

Usage: For creating probabilistic models, or when data doesn’t follow a normal distribution.

Best for: Estimating the likelihood of mineralization above/below cut-offs.

5. Co-Kriging

Assumption: You have secondary data that correlates with your primary data (e.g., copper grades and density measurements).

Usage: Incorporates both primary and secondary variables.

Best for: Enhancing estimates when secondary data is easier to obtain or denser than the primary data.

How to Choose the Right Kriging?

Type	When to Use
Simple	When the global mean is well-known (rare)
Ordinary	Standard choice in most mining contexts
Universal	When there’s a visible spatial trend in the data
Indicator	When estimating categories not values

Want to understand geostatistics and geomodeling: it's here