I
n t e r p o l a t i o n
Idrisi
Inverse
Distance Weighting:
This was the first method of interpolation performed on the data once the database
had been properly reconstructed and is shown in the map to the right. The results
for each of the layers were fairly representative of the layer surfaces, however
extreme values were over represented. Also, on the southern side of the study
area, where there are virtually no points for interpolation, the weakness of
the IDW method is evident. What appears here on all of the layers is a circular
banding of the height of the most closely associated well values.
Kriging:
Kriging was originally planned to be the focus of our interpolation methods
for our data. However, after the many problems that surfaced, and the data was
finally ready for interpolation, the modules for Kriging in Idrisi would not
function. The first step of the process, where using the Spatial Dependence
Modeler you explore the spatial variability of the data set would not work.
When the proper input layers and the graph function was selected, the program
would either give an error message, or crash. It seemed to be very unstable.
The geostatistical modules in Idrisi, while extensive, use an open source geostatistics
program called GSTAT to perform the calculations. It is possible that this part
of the Idrisi software we were using had bugs in it. Without being able to access
the Spatial Dependence Modeler, the Model Fitting module and the Kriging and
Simulation interfaces could not be used. This was a major problem, as Kriging
would undoubtedly produce more accurate results than would Inverse Distance
Weighting, especially considering the irregular distribution of the data points.
Beyond this, there were few interpolation options left, one TREND surface was
created, however this is an inexact interpolator, which loses the original data
values. Due to the scarcity of data as is, it was clearly not a useful interpolation
method. The result of the Trend surface created for the gravel layer is show
here.
