MODELLING AND ZONING GROUNDWATER POTENTIALITIES 
The main objective is to mapping favorable areas for groundwater occurrence on a regional scale. The 
model technique is adopted from AGTERBERG et.al(1988) who developed it for mineral explorations. The 
methodology can be used to estimate how much the prior probability that high yielding wells are present 
within a neighborhood is increased because of new evidence of factors which are favourable to it's 
occurrence. In this context, some spatial features used to establish decision rules have been considered as 
influencing factors on well yield. This can be explained as in figure 5 where a lineament was taken in to 
account as a influencing factor. 
The priori probability (unconditional probability) of high yielding wells within a small arbitrary area can 
be calculated by area of occurrences of high yielding wells divided by the total area (D/T). Figure 6 is the 
venn diagram showing the association between an arbitrary of a lineament and are of high yielding wells. 
  
  
  
  
  
  
  
: T m T - Study area 
EE B 
EE BND 
ura B - Area of map pattern 
pupa n BND (lineament corridor) 
/ P a = ^ 
pet NS 
ad a e D - Area of wells 
‘Tube well BND 
Figure 5 Figure 6 
Two weights can be defined for each map pattern for a quantitative estimation of the association between 
the binary map pattern and well location, as follows: 
W = for those areas on the binary map pattern 
W- - for those areas off the binary map pattern 
The above two weights have been calculated as log ratios of conditional probabilities using following 
equations: 
ln 2D) (BND)/ D 
P(B/D) (BND)/ D 
eh P(B/D) an (BND)/D 
By By Dy nen gpg D 
The measure of association between a map pattern and tube well points is given by W* - W” and denoted 
by C. Considering a lineament, as a spatial feature, the dialation operation available in the used geographic 
information system have been carried out create "buffer zones" to optimize the spatial association between 
lineament and well point. This was done by making a series of buffer zones around the lineament in order 
to create a distance map. This distance map, as a predictor map, would be very useful in determining 
optimum value for C, the reliable association, in order to include required weights for the model. For 
example for different dialated areas, it possible to calculate different C values as a function of distance to 
lineament. These predictor maps could be calculated for all spatial features which have been included in the 
362 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996