Water absorption spectroscopy is a well-established and widely used laser diagnostic which can be applied to measure temperature information in various environments. A recently developed spectroscopic objective function is used to select wavelengths that optimize the temperature precision of water absorption thermometry.
Typical line-of-sight absorption thermometry is only able to provide integrated gas properties along the beam path with the assumption of uniformity along the path. This project develops a novel technique combining water absorption spectroscopy with a specific computed tomography (CT) algorithm, named as total-variation based compressed sensing (TVCS), to reconstruct two-dimensional temperature image under limited sampling conditions.
A new noise characteristic is derived empirically in this work through a series of simulation studies to extend the temperature precision analysis to absorption-based tomography. Phantoms with two Gaussian-shape temperature distributions are employed to validate the modified wavelength selection approach. Studies confirm that use of more than 2 wavelengths dramatically improves the final temperature imaging results when the range of temperatures in the domain is large (e.g. 300 – 2000 K). An experimental heated disk setup is implemented to generate a steady temperature distribution, and complete absorption datasets are measured by translating and rotating a single tunable laser beam through the test region. This experiment demonstrates the capability to reconstruct the temperature distribution with mean relative errors of only 2.5% when using 3 wavelengths at 64 beams and 6 view angles.