Assessment of particle size magnifier inversion methods to obtain particle size distribution from atmospheric measurements

Abstract

Abstract. Determining the particle size distribution of atmospheric aerosol particles is an important component to understand nucleation, formation and growth. This is particularly crucial at the sub 3 nm range because of the growth of newly-formed nanoparticles. The challenge in recovering the size distribution is due its complexity and the fact that not many instruments currently measure at this size range. In this study, we used the particle size magnifier (PSM) to measure atmospheric aerosols. Each event was classified into one of the three types: new particle formation (NPF), non-event and haze events. We then compared four inversion methods (step-wise, kernel, Hagen and Alofs and expectation-maximization) to determine its feasibility to recover the particle size distribution. In addition, we proposed a method to pre-treat measured data and introduced a simple test to estimate the efficacy of the inversion itself. Results showed that all four methods inverted NPF events well; but the step-wise and kernel methods fared poorly when inverting non-event and haze events. This was due to their algorithm, such that when encountering noisy data (e.g. air mass fluctuations) and under the influence of larger particles, these methods overestimated the size distribution and reported artificial particles during inversion. Therefore, using a statistical hypothesis test to discard noisy scans prior to inversion is an important first step to achieve a good size distribution. As a first step after inversion, it is ideal to compare the integrated concentration to the raw estimate (i.e., the concentration difference at the lowest supersaturation and the highest supersaturation) to ascertain whether the inversion itself is sound. Finally, based on the analysis of the inversion methods, we provide recommendations and codes related to the PSM data inversion.

Variation of size-segregated particle number concentrations in wintertime Beijing

“…Traditionally, NPF events occur during the time window between sunrise and sunset by photochemical reactions Dada et al, 2018). The binary or ternary nucleation between sulfuric acid and water, ammonia, or amines are usually thought of as sources of atmospheric cluster mode particles, espe-cially in heavily polluted environments Yao et al, 2018;Chu et al, 2019). The burst of cluster mode particle number concentration outside the traditional NPF time window, especially during the rush hour in the afternoon, suggests a very different source of cluster mode particles from traditional nucleation, e.g., nucleation from gases emitted by traffic (Rönkkö et al, 2017).…”

“…Apart from their health effects, the temporal and spatial variation of particle number concentrations of different sizes is a good indicator of the strength of their emission sources. Aerosols are emitted directly as primary particles, such as sea salt or dust particles; as a result of natural phenomena (Solomos et al, 2011); or they can be formed through new particle formation (Kulmala, 2003;Kulmala et al, 2004Kulmala et al, , 2013Kerminen et al, 2018;Chu et al, 2019). Newly formed particles can grow up to diameters of 20-100 nm within a day (Kulmala et al, 2004), and they have been found to contribute to a major fraction of the global cloud condensation nuclei (CCN) population, thus indirectly affecting the climate .…”

Long-term measurement of sub-3nm particles and their precursor gases in the boreal forest

“…Before the PSM raw data was inverted, the data was run through a quality control algorithm in which we assumed that the measured total concentration during each scan should have a positive and statistically significant (p < 0.05) correlation with the PSM saturator flow rate (Chan et al, 2020). Increasing the flow rate should always result in increased total concentration because we are activating a larger size range of particles.…”

Supplementary material to &quot;Towards understanding the mechanisms of new particle formation in the Eastern Mediterranean&quot;

“…The two most popular methods to invert PSM data are the kernel function method and the step inversion method. The expectation-maximization (EM) method has been recently recommended over the kernel method because it is less sensitive to random errors (Cai et al, 2018;Chan et al, 2020). Here, we compare the kernel method and the EM method using PSM data from the whole measurement period.…”

“…4) Noise filtering procedure achieved by applying a 6 th order median filter on the one second resolution data. 5) Quality check using the method suggested by Chan et al (2020). 6) Minimization of the inversion matrix using a saturator flow inversion window of 0.08 lpm which minimized the saturator flow (corresponding to cut-off diameter) scans from ~120 to 16 per onedirection of the scan.…”