Abstract
Horizontal radial plume mapping is a cost-effective optical remote sensing method for sensitive mapping concentration distribution of atmospheric chemicals in real time. However, its sparse sampling poses challenges for reconstruction algorithms. Neither non-smooth nor smooth algorithms can recover the realistic plume shape. A new approach called Gaussian dispersion transformation (GDT) has been proposed. It first reconstructs the emission rates from unknown sources. Then concentrations are calculated through a transformation matrix defined by a Gaussian dispersion model. Smoothness regularization is also applied during the reconstruction. The method was evaluated by using randomly generated maps. It shows significant improvement over a reconstructed plume shape. The nearness shows 72%–117% better than the non-negative least-square (NNLS) algorithm and 15%–26% better than the low third derivative (LTD) algorithm. A controlled-release field experiment of methane was also conducted. The realistic concentration distribution was calculated by using a Lagrangian stochastic dispersion model. The GDT algorithm successfully recovered the realistic plume shape. The nearness shows approximately 16% better than the NNLS and the LTD algorithms. Finally, a sensitivity analysis shows that the wind direction and atmospheric stability are the main parameters that affect the performance of the GDT algorithm.
Funder
Canada Foundation for Innovation
Natural Sciences and Engineering Research Council of Canada
University of Calgary
Subject
Atomic and Molecular Physics, and Optics,Engineering (miscellaneous),Electrical and Electronic Engineering