GISdevelopment.net ---> AARS ---> ACRS 1989 ---> Water Resources

Ground water targeting using digitally enhanced imagery

Dr. K. K. Rampal
Professor Department of Civil Engineering
Indian Institute of Technology Kanpur
Kanpur (UP) India 208 016

K.V.G. Rao
Research Scientist Centre for Study
in Resources Engineering (CSRE)
Indian Institute of Technology
Bombay, India 400 076


Abstract
Morphology of features and permeability of rock formations seems to have a definite relationship with the availability of under ground water. Limited studies performed in this field have established that permeability is directly proportional to the infiltration umber. The ground water targeting thus depends upon the identification and mapping of fractures, lithological units and regional geology of the area.

To identify such features from the available remotely sensed data is presently not possible. It is expected that the image processing techniques applied to such data may be found useful in lineament detection.

By selecting MSS data pertaining to LANDSAT-5, the DODDAGUNI area in topographica sheet No. 57C in the Karnataka State of South India, Programs have been developed to enhance the images using histogram equalization, combination and ratioing of bands and high and low pass filtering etc., which are capable of image enhancement upon the general purpose computer including PCs. To suit the requirement the 256 gray levels have been compacted in 16 grey levels to suit a Tektronic-4107/4109A colour graphic terminal.

Using the various enhanced images, a geological map of the area has been prepared and based upon the above-mentioned understanding a relationship between various lithological units and estimated water features is made.

The availability of water is partially checked through he Bay's classifier and it is found that the classified areas estimated to be having water can be clearly marked through above mentioned enhanced images. In both cases the results were found encouraging.

Introduction
Morphology of features and permeability of rock formations seem to have a definite relationship with the availability of underground water. Limited studies performed in this field have established that permeability is directly proportional to the infiltration number. The ground water targeting thus depends upon the identification and mapping of fractures, lithological units and regional geology of the area.

In the present study the Digital Image Processing of LANDSAT digital available on Computer Compatible Tape (CCT) was carried out to enhance the geological features, such as, lineaments lithological units etc. related to ground water. In particular, study was made for the development and application of Image Enhancement and Image Classification procedures a general purpose computer systems and way of the assessing their performance and understanding how the ground water targeting is acheived. The area selected for study is the MSS data partaining to LANDSAT-5 to DODDAGUNI area in the topographical sheet no. 57C in the Karnataka State of South India. (Latitute 130N, Longitude 760E path No. and Row No. of CCT as 144-51).

Geological description of the study area.
As per available geological map, the village Doddaguni is situated on Chitradurga schist belt. Tee rock types in the area around Doddaguni can be subdivided into two groups. The older group is made up of amphibolites and chlorite schist interceded with cross-bedded guartizites and the younger group contains bedded iron formation (BIL), marble, peletic and simplistic schist along with amphibolts. The other group can be correlated with Babadudan group of earlier works (Swaminath et al 1976, Rama Krishna et al 1976) and the young group with the Chitra durga group (Mukhopadhaya - 1981). Fig. 1 shows the geological map.

Previous Work
The work of S. K. Sharma (1986) shows that there is definite relation between morphology of features (density and frequency) and permeability of the formation in bard rock terrain based on the infilteration number, which is the product of fracture density and frequency. Stephen and Myner ( 1985) used five image enhancement techniques to LNADSAT digital data for lineament detection namely, Mean Value of Four Bands reflectances, Principal Component (PC), Analysis, band retioing, Histogram equalization and High pass Filtering. They found that all the five techniques identified significant amount of lineament features.

Study Procedure
The present study was carried out in two phases. The first phase was to extract the various geological features related to the ground water through digital image process and the second phase involved the mapping of the ground water potential area.

In the study emphasis was given enhance the geological features. The enhancement used can be classification into three categories:
  1. Contrast manipulation by Histogram equalization and other stretching techniques.

  2. Multi Image Manipulation by Multi Band addition and Band Rationing.

  3. Spatial Feature Manipulation by Image Smoothing through low pass filtering and edge enhancement through high pass filtering.
Results
Fig. 2 shows the Landsat MSS band 4 subimage with 15 grey levels before histogram equalization and Fig. 3. Gives after histogram. Qualization. Fig. 4 shows the effect combining or adding Band 1 and Band 2 on the depiction of drainage pattern and Fig. 5 shows the result of combination of Band 3 and Band 4 bringing out clearly the fracture zones. Fig. 6 shows the results of all four MSS bands combined which results in much better depiction of lithological units.

Fig. 7 shows the results of low pass filtering for Band 4 (with equal weights). Fig. 8 shows the Digital Thermatic Map obtained with elysian Classifier. The structural map obtained as a result of the comparison between existing geological map and the imagery obtained as result of enhancement, making use of the regions targeted as A, B, C & D on the structural map and suitably interpreting the enhanced imagery a map showing the ground water potential zones was prepared. This is shown in Fig. 10. The ground water data of about 10 observations well above map. All the targeted water-bearing zones have been marked on Fig. 10. The limited amount of ground data confirm, the results obtained in their investigation.

Conclusions
  1. The Density Slicing and histrogram equalization methods used to generate the digital image with sixteen colours at a time on the graphic terminal proved that the graphical terminals could be for displaying the remote sensed images. The contrast of the displayed images is greatly increased by suing the maximum available sixteen colours at a time throught he histrogram equalization method.

  2. Band 4 image shows relatively more geological features i.e., discrimination between lithological units, lineaments and surface water bodies are comparatively more clear after histogram equalization technique.

  3. The combination of Band 3 and Band 4 has enhanced fractures in granite terrain considerably. The image obtained by combining all the four bands shows clear demarcation between different lithological units.

  4. While the advantage of two pass Band Ratioing method gives better resolution the one pass may be preferred.

  5. Low Pass Filter is found to be useful to enhance the sharpness of the iamge. By using this technique the fracture zones, lineaments and linear details, i.e. roads and railways which are generally not visible in the raw image can be clearly seen.

  6. Baye's Classifier was utilized to obtain the data in four classes viz. water, granite rocks, sedimentary rocks and vegetation. While the earlier three classes fully conformed to the image obtained through enhancement technique, the vegetation ground water potential zones.
References
  1. Stanley. N, , Dewitt, J.M. 1986, "Hydrogeology" John Wiley & Sons.

  2. Stephen. J, 1986., "Landsat Digital Enhancement for Lineament Detection", Environmental Geology Vol. 8 No. 3 pp. 123-128.

  3. William P., M. Karalick & Jeffersion, A. Stephen Whartan. 1987, "A Methodology for evaluation of an interactive multispectral Image Processing System, Photogrammetric Engineering & Remote Sensing",