Continuing developments in detector technology have given rise to a new class of imagers combining the advantages of Broadband thermal infrared (TIR) imaging, commonly used to investigate the physical processes taking place during eruptions, and spectroscopic methods such as Fourier transform infrared (FTIR) or dispersive gratings, which can provide high-resolution spectral information in the same wavelength range. The combination yields hyperspectral imagery (a full spectrum for each pixel), which can be used to observe of volcanic activity at or to quantify surface mineralogies. I have used a hyperspectral imager to observe volcanic activity at Stromboli volcano in Italy. Simple techniques based on curve-fitting and principle component analysis (PCA) help quickly extract spectral information from large high-resolution datasets and implement fast and reliable pixel classification. Those straightforward statistical methods can be implemented unilaterally to datasets acquired over a variety of observing conditions. They are computationally economical and more accurate when compared to more traditional Brightness Temperature Difference (BTD) indicators, which are commonly used to identify potential pixels of interest in large datasets from multispectral and hyper-spectral satellite instruments. Quantifcation of the gases is also possible, using methods similar to those employed for OP-FTIR. However, the process is resource-intensive given the large number of pixels to treat. I am developing simplified algorithms to perform this analysis in near real time.
