Kriging is a geostatistical method that generates or interpolates a probability
surface that fits best to a scattered set of point values in two-dimensional space. It is different from other interpolation methods Spatial Analyst supported by ArcGIS as it involves an interactive investigation of the spatial behaviour of the point values the best estimation method is selected to map the output surface. Since, Kriging presumes a uniform pattern of distribution of point values which is never found in nature, therefore, the spatial variation is quantified by the semivariogram. First, plot a scatter diagram representing distances of the point values on the X- axis and point values on Y-axis. It will be observed that after a certain distance the scatter of point starts somewhat flattening that point values are almost similar and don�t show much variation with distance, though, some wild variations cannot be over ruled, yet average pattern would show somewhat flattening. The distance where the model first flattens is known as the RANGE. The point values separated by smaller distances are spatially autocorrelated, while values beyond the RANGE are not. The point value on the Y-axis at which the scatter of points reaches the RANGE is called the SILL. A PARTIAL SILL is the SILL minus the NUGGET. The NUGGET is mostly due to measurement errors. Either draw a smooth curve through a scatter plot or fit a function to the scatter point values against their distances. In the former case where the smooth curve intersect the Y-axis or in the later case the value of intercept (constant of the fitted function) will define the NUGGET. After finding out RANGE and NUGGET and SILL values define either through inspection of scatter diagram or fitting a curve, define the range of SEARCH RADIUS. The SEARCH RADIUS may be either fixed or variable. The kriging operation is of two types; Ordinary and Universal. Ordinary Kriging is the most general and widely used of the kriging methods and is the default. It assumes the constant mean is unknown. Universal Kriging assumes that there is a strong trend in data throughout the region. Therefore, Universal Kriging should only be used when one knows that there is a trend in data and one can give a scientific justification to describe it. Creating a predicted surface map using Ordinary Kriging 1. Click the point layer on which you wish to perform Ordinary Kriging in the ArcMap table of contents. 2. Start the Geostatistical Analyst. 3. Click the Attribute dropdown list, and click the attribute on which you wish to perform ordinary kriging. 4. Click the Kriging method. 5. Click Next. 6. Click Prediction under Ordinary Kriging. 7. Click Next. 8. Specify the desired parameters in the Semivariogram/Covariance Modelling dialog box and click Next. 9. Specify the desired parameters in the Searching Neighbourhood dialog box and click Next. 10. Examine the results on the Cross Validation dialog box and click Finish. 11. On the Output Layer Information dialog box, click OK.