The key task of interpolating the lithology layers proved to be one of the more difficult aspects of the project. The original intention was to use a variety of geo-statistical and non geo-statistical methods to derive lithology surfaces, which would then be reconstructed into a 3D model. The data was broken up into 5-foot intervals via manual database editing, and later a VB script. The main problem however, is that our database, after standardization and classification, was set up to interpolate between nominal layer classes. This is impossible, as the numbers for the class names have no numeric value.

The problem outlined above became evident in Idrisi after the first 5-foot layer was interpolated. The concentric rings in the figure above illustrate the problem of using nominal data for interpolation. These rings are caused by the interpolation method trying to estimate the unknown values by selecting values between the known data points. The problem becomes more apparent when the image is zoomed to the area containing the 3rd and 4th class, orange and red areas.

In this case, there are points with layer class 4 surrounded by other points with layer class 2. The interpolation process naturally derives values of 3 between them, as indicated by the concentric rings of layer class 3. The nature of interpolation, namely estimating unknown values based on known data observations assumes that the data under analysis is continuous, and that between any two data points of differing values, there will be a continuum of values. This was undoubtedly a major unforeseen problem.

The irregularity of the distribution of data points caused problems when the Thiessen polygon method of deriving catchment zones was attempted. The interpolation using Thiessen polygons in below illustrates the overly generalized output due to the data point distribution.

Much of the study area was made into large contiguous polygons of one material where there was simply no litho data available. This method could clearly not improve upon previous work done visualizing the layers and was abandoned after one layer.