Validation of spectral data: A critical step towards accurated prediction

Ramanaidou, Erick1

1Mineral Resources – Discovery | CSIRO, Kensington, WA, Australia

The last three decades have seen the emergence of spectral mineralogy as a valued tool for the exploration and mining companies. From the past luggable GER IRIS and portable PIMA we now have access to an extended suite of small, fast, light and accurate field spectrometers such as the ASD and Spectral Evolution spectrometers covering the visible, near infrared and short-wave infrared – VNIR-SWIR 380 to 2500 nm range. In parallel, automated systems to scan diamond cores and drill chips have been developed such as the hyperspectral point analyser, the HyLogging System™, the hyperspectral imaging spectrometers, the Corescan™ Core Imager Mark III, SpecIm SisuROCK, Neo Hyspex and HCIS Terracore are commercially available for use by the exploration and mining companies. As well as VNIR-SWIR, the thermal infrared -TIR range (8 to 14 µm) is now available with the HyLogging System™ 3 providing the detection minerals such as quartz, felspar, olivine, pyroxene and garnet. Large volumes of diamond cores and drill chips have been measured by the exploration and mining companies and the spectral geologist research community has responded by providing automated ways of processing large number of spectra. The CSIRO- developed the spectral geologist or TSG™ offers two ways of processing the spectra (1) through the automated spectral analysis program or The Spectral Assistant (TSA) or through customised scripts, algorithms that use depth and minimum wavelength of absorptions to uniquely identify specific minerals. The TSA is applied for HyLogging System™ spectra on measured areas on around a few cm2 where mineral mixture is likely. On the other hand, the hyperspectral imaging Corescan™ Core Imager Mark III captures many pure pixels at a resolution of 500 µm and the mineral mapping processing is performed using a dedicated expert system program.

Reflectance spectra acquired using these systems are often the complex results of many absorptions embedding not only mineralogical but also particle size information. Although quite powerful, the processing methods previously mentioned require validation by more classical methods such as x-ray diffraction, Raman spectroscopy, X-ray fluorescence (XRF) and µXRF mapping to improve prediction.

Through examples selected from the iron ore and nickel laterite industries, it will be demonstrated that complementary and cross validated methods are essential to ensure that validation of spectral data is undertaken as a critical step towards accurate mineralogical prediction and that it is good to have redundant information.


Dr. Erick Ramanaidou is the Commodity Research Leader for iron ore and nickel laterite. He has been involved in spectral research for the last 25 years and has concentrated his effort to the understanding of the spectral properties of iron oxides and gangue minerals in iron ores.

About the GSA

The Geological Society of Australia was established as a non-profit organisation in 1952 to promote, advance and support Earth sciences in Australia.

As a broadly based professional society that aims to represent all Earth Science disciplines, the GSA attracts a wide diversity of members working in a similarly broad range of industries.