Upcoming TALKS

Title: Advances in NASVD – to an automated data analysis and isotope specific map

Date: Tuesday September 10th at 4:00 PM EDT

Speaker: Dr. Nicolas Martin-Burtart (Radiation Solutions)

Location: University of Toronto, Department of Earth Sciences, Room ES2093 (22 Ursula Franklin Street)/Virtual, please register here

Abstract: 

Noise Adjusted Singular Value Decomposition (NASVD) was developed in 1997 by Jens Hovgaard and has been widely used in airborne gamma spectrometry. NASVD determines all statistically significant spectra for an entire gamma-ray data set and then reconstructs the original data set with a higher signal to noise ratio (Hovgaard 1997). The technique uses up to 1024 channels of data from the entire survey data set to identify all statistically significant spectral shapes which are then used to reconstruct new potassium, uranium, and thorium windows.

After this reconstruction, the new windows are found to have significantly less noise than the original raw windows. Since the procedure makes use of all the counts in the spectrum as well as the correlation between potassium, uranium and thorium, the reduction in statistical noise is found to be significantly greater than for multi-channel spectral fitting. Maps produced using these new windows were found to show significantly more geological information than those produced by the standard 3 window method.

The NASVD components computed to describe the dataset are a combination of isotopes’ signatures, and spectral components will have both positive and negative “real” isotope contributions. The NASVD component concentrations represent the proportion of the different spectral components used to recreate each point of the data set, but do not represent individual isotope contribution. However, if the NASVD reconstructed (filtered) spectra can be analyzed for isotope content in a single point, then all data can be filtered and analyzed for isotope specific content. It means that “real” spectral components can be derived from NASVD spectral components and quantified.

This new, automated method analyzes the data set using NASVD and then the spectral components are automatically analyzed for nuclide specific content using a template library. This enables the mapping of specific isotopes, allowing the end-user to generate a map of the elements of their choice. This is applicable to the geological world as well as the nuclear safety world.

The template library has a limited number of isotopes and additional work is being performed to include “unknown” isotope mixtures. In a complex fall out mixtures it may not be possible to identify all isotopes, but it is still possible to quantify the concentration of the then unknown mixture.

Author Biography: 

Nicolas Martin-Burtart obtained his PhD in Nuclear Physics in 2012 from the University of Strasbourg, France. Under the direction of the French Atomic Commission (CEA), Nicolas completed and successfully defended his thesis on airborne gamma ray spectrometry, developing algorithms for nuclide identification. In 2013, he joined Radiation Solutions (RSI) as Senior Scientist. His current focus includes nuclide identification, detector characterization and neutron detection. Nicolas is experienced in airborne and ground radiometric mapping and data processing, with a particular interest in the application of noise reduction techniques to airborne gamma spectrometry.