J.K. Shultis and F. Khan, User Notes for SOILLR, Report HSRC-94-02-16, 15 pp., Report for Hazardous Substance Research Center Project 94-02, Dept. Nuclear Engineering, Kansas State University, August 1997.

In the prompt-gamma neutron activation analysis (PGNAA) of contaminated soil, a neutron source irradiates the soil, and contaminant atoms, upon absorbing a neutron, emit capture gamma photons with distinctive energies. The intensity of these capture-gamma photons are measured by a detector above the soil surface, and, from these measurements, the concentration of the contaminant can, in principle, be determined at different depths $z$ in the soil. Estimation of the contaminant profile from measured photon intensities is the problem addressed by the code SOILLR.

The expected number of counts caused by uncollided capture-gamma photons of a specific energy reaching the detector-spectrometer is related to the contaminant concentration by a Fredholm integral whose kernel depends on the source, soil, and detector properties. SOILLR solves this equation for the concentration profile u(z), approximating the integral equation by a set of (generally underdetermined) linear algebraic equations, which are then solved by imposing a regularizing smoothing constraint. In particular, SOILLR uses two solution methods: (1) the linear regularization (LR) technique, and (2) an iterative constrained linear regularization (CLR) method that guarantees positivity of the estimated profile.

SOILLR contains a large data base to permit analyses of many heavy metal contaminants in five representative soils using 14-MeV neutrons, Am/Be neutrons, or Cf-252 fission neutrons in a variety of source geometries. The following contaminants can be analyzed: N, P, Cl, Cr, Fe, Cu, Zn, Cd, Hg and Pb. The five representative soils cover the wide natural variation of soils and include: nominal, dry porous, wet dense, dry dense, and wet porous.

The program can be used to generate simulated data for the several idealized profiles. Additionally, an arbitrary contaminant profile may be specified by the user. The program uses these profiles to generate simulated data which are then analyzed to determine how accurately SOILLR can recover the profiles. Alternatively, photon intensity data, generated outside the program, can be input to the program.

A program option allows simulated statistical noise to be added to the intensity count data. Another option causes SOILLR to perform a sensitivity analysis of the estimated profile to data uncertainties by using the Latin hypercube sampling technique.

Home