Source parameter estimation of atmospheric pollution using regularized least squares

This paper presents a regularized least squares method to estimate the location and release rate of atmospheric pollution. We assume that measured pollution concentration at different ground locations and meteorological conditions such as wind speed are available so that one can solve the advection-diffusion equation for a non-steady point source. However, even finding linear parameters related to the release rate is an ill-posed problem and one has to impose certain regularization technique to avoid potential overfit. We propose to use ℓ
p
-regularization (0 ≤ p ≤ 1) and discuss its advantage over the popularly used ℓ
p
-regularization. The accuracy of source parameter estimation is examined for the cases where the number of sources and the corresponding locations are unknown.