Getting to know the Kriging method involves delving into the mysterious world of probability and prediction, which, depending on your understanding of statistics, could lead either to years of therapy, or to your Ph.D.
Like IDW, Kriging is a weighted average technique, except that the weighting formula in Kriging uses much more sophisticated math. Kriging measures distances between all possible pairs of sample points (that’s right, all of them) and uses this information to model the spatial autocorrelation for the particular surface you're interpolating.
In other words, Kriging tailors its calculations to your data by analyzing all the data points to find out how much autocorrelation they exhibit and then factors that information into the weighted average estimation.
When you interpolate a surface using Kriging, the distance and direction of every point pair is quantified to provide information on the spatial autocorrelation of the sample point set. Next, a best-fit model is automatically applied to the data and the unknown values are predicted.
Kriging aficionados consider the initial kriged surface a first draft—a test surface against which they compare future iterations as they search for the perfect surface. Directional influences, such as prevailing winds and random error, can be accounted for using Kriging, but you will need a statistical tool such as ArcGIS™ Geostatistical Analyst to visualize these trends.
A surface created with kriging can exceed the value range of the sample points, but will not pass through the points.
Two general and widely used Kriging methods are Ordinary and Universal Kriging. Universal Kriging assumes that there is an overriding trend in the data. For example, you may know that there is a prevailing wind or a gently sloping hillside across your study area. Ordinary Kriging assumes there is no trend in the data, which should be your standard operating assumption.