S. Aravindan, B. Poovalinga Ganesh

In order to determine mineralogy of rock and soil samples, reflectance and emittance spectroscopy in the near-infra red and short-wave infra-red is used extensively and found to be inexpensive. Hyper spectral remote sensing satellite data are found to be prospective to deliver in depth physico-chemistry like mineralogy, chemistry, morphology of the earth’s surface. Therefore hyper spectral data is useful for mapping potential host rocks, alteration assemblages and mineral characteristics. In the present study EO-1, Hyperion data had been used for delineating magnetite mineral in Godumalai hill, Salem region, Tamil Nadu, India. The requirements for extracting magnetite from Hyperion images is to be first compensated for atmospheric effects using flag mask correction, cross track illumination correction and FLAASH model. Minimum Noise Fractionation transformation was applied to reduce the data noise and for extracting the extreme pixels. Some pure pixel end member for the target mineral and backgrounds were used in this study to account for the Spectral Angle Mapping & matched filtering techniques and the arrived results were validated with field study. Those mapping techniques have proved that the magnetite mineral can be mapped with high precaution by Hyperion preprocessing adopting methods.