Spatial resolution and noise considerations in determining scalar dissipation rate from passive scalar image data

Abstract

The effects of photonic shot noise and finite spatial resolution on the scalar dissipation rate were investigated for the analytical profile of a passive scalar layer subjected to a compressive strain, and the results were applied to interpret measured data from spray mixing data from an internal combustion engine. A Monte Carlo approach was employed. The measured scalar dissipation rate is underestimated, and the layer width measured at 20% of the peak height is overestimated by the finite resolution. The ratio of the local scalar spread value to the noise level, the spread-noise ratio, was found to describe the noise effects, which principally results in an overestimation of the scalar dissipation rate, especially at high resolution levels. The Nyquist resolution provides a good compromise between the sampling bias at low resolution and the noise bias at high resolution. Top hat filtering the raw data prior to calculation of the scalar dissipation rate was found to, effectively, reduce spatial resolution, whereas median filtering preserved the resolution. Both filters had a comparable effect on noise reduction. The evaluation of experimental data showed that a significant fraction of data reside at low spread-noise ratio and are biased by noise. The peak scalar dissipation rate is, however, not biased by noise and a method of estimating spatial resolution based on the peak scalar dissipation rate is described.