CONCLUSION

The simple analysis and field verification reported in this paper show that it is possible - at least in the Syrdarya area of Kazakhstan - to use spectral image data extracted from Landsat ETM+ to assist in the search for fossils in the ground. Using image data we were able to locate areas of likely fossiliferous Cretaceous-aged sediment prior to prospecting on the ground, saving both time and money. Limitations are implied, however, particularly in the accuracy of the spectral analysis and its ability to distinguish closely related spectral sets: for example, our attempts to recognize Zhirkindek Formation deposits within the field area with Landsat ETM+ images failed because these deposits are represented mostly by grey-colored clays that are hard to determine accurately from images. This is because the spectral characteristics of this unit are similar (in Landsat images) to those of the overlying Paleogene strata in the field area.

Nevertheless, if treated with caution, the classification and interpretation of satellite multispectral images could be very helpful to paleontologists in particular for determining likely areas for fruitful field exploration. The spectral libraries created for our approach are different from those used for other kinds of geological exploration in that they are created without reference to mineral compositions. Our approach uses just the reflected spectrum of beds exposed at the surface.

As this project develops, we will build spectral libraries for other fossiliferous locations across Kazakhstan that are of large enough thickness to be seen from space. Spectral libraries may potentially have a huge significance to the future extrapolation of prospective fossiliferous sites.