Application of artificial neural network and satellite images for potential exploration mapping in GIS environment

Mineral exploration is a complex task that often requires the use of satellite imagery, geochemical and geophysical data in order to achieve a better result. The following study is a result of a research about Artificial Neural Network (Back propagation) in GIS environment to determine areas of Cr anomalies in Baneh (Kurdistan). Images of Landsat TM (the date 11/2002) has been used to determine rock alterations in the region. For determination of rock alterations and pyrite oxidation, we used band ratios 1 +2 +3 / 4 +3 and 2-3 +1 / 8 +4. Also field observations and petrography of thin section were used to control the results. GIS data and the results of the geochemical analysis were used in neural network data processing. Four types of neurons were used in the first layer. Also, 50 neurons in the second layer and 1 neuron in output layer were used. By using of the network, Cr potential maps with high accuracy of 86% were estimated (compared to ground data). This method is based on the numerical data and compared to conventional techniques (knowledge base) which has more advantages. The results showed that the use of satellite imagery and neural cells may play a significant role in geochemical mapping of mineral potential.