scholarly journals Neural network modelling to estimate particle size distribution based on other particle sections and meteorological parameters

2021 ◽  
Author(s):  
Pak Lun Fung ◽  
Martha Arbayani Zaidan ◽  
Ola Surakhi ◽  
Sasu Tarkoma ◽  
Tuukka Petäjä ◽  
...  
Keyword(s):  
Neural Network ◽  
Particle Size ◽  
Particle Size Distribution ◽  
Size Distribution ◽  
Meteorological Parameters ◽  
Aerosol Size Distribution ◽  
Sufficient Information ◽  
Electrical Mobility ◽  
Particle Sizer ◽  
Mass Concentrations

Abstract. In air quality research, often only particle mass concentrations as indicators of aerosol particles are considered. However, the mass concentrations do not provide sufficient information to convey the full story of fractionated size distribution, which are able to deposit differently on respiratory system and cause various harm. Aerosol size distribution measurements rely on a variety of techniques to classify the aerosol size and measure the size distribution. From the raw data the ambient size distribution is determined utilising a suite of inversion algorithms. However, the inversion problem is quite often ill-posed and challenging to invert. Due to the instrumental insufficiency and inversion limitations, models for fractionated particle size distribution are of great significance to fill the missing gaps or negative values. The study at hand involves a merged particle size distribution, from a scanning mobility particle sizer (NanoSMPS) and an optical particle sizer (OPS) covering the aerosol size distributions from 0.01 to 0.42 μm (electrical mobility equivalent size) and 0.3 μm to 10 μm (optical equivalent size) and meteorological parameters collected at an urban background region in Amman, Jordan in the period of 1st Aug 2016–31st July 2017. We develop and evaluate feed-forward neural network (FFNN) models to estimate number concentrations at particular size bin with (1) meteorological parameters, (2) number concentration at other size bins, and (3) both of the above as input variables. Two layers with 10–15 neurons are found to be the optimal option. Lower model performance is observed at the lower edge (0.01 

scholarly journals Data imputation in in situ-measured particle size distributions by means of neural networks

2021 ◽  
Vol 14 (8) ◽  
pp. 5535-5554
Author(s):  
Pak Lun Fung ◽  
Martha Arbayani Zaidan ◽  
Ola Surakhi ◽  
Sasu Tarkoma ◽  
Tuukka Petäjä ◽  
...  
Keyword(s):  
Particle Size ◽  
Particle Size Distribution ◽  
Size Distribution ◽  
Meteorological Parameters ◽  
Number Concentration ◽  
The Other ◽  
Size Distributions ◽  
Ill Posed ◽  
Input Variables ◽  
Particle Sizer

Abstract. In air quality research, often only size-integrated particle mass concentrations as indicators of aerosol particles are considered. However, the mass concentrations do not provide sufficient information to convey the full story of fractionated size distribution, in which the particles of different diameters (Dp) are able to deposit differently on respiratory system and cause various harm. Aerosol size distribution measurements rely on a variety of techniques to classify the aerosol size and measure the size distribution. From the raw data the ambient size distribution is determined utilising a suite of inversion algorithms. However, the inversion problem is quite often ill-posed and challenging to solve. Due to the instrumental insufficiency and inversion limitations, imputation methods for fractionated particle size distribution are of great significance to fill the missing gaps or negative values. The study at hand involves a merged particle size distribution, from a scanning mobility particle sizer (NanoSMPS) and an optical particle sizer (OPS) covering the aerosol size distributions from 0.01 to 0.42 µm (electrical mobility equivalent size) and 0.3 to 10 µm (optical equivalent size) and meteorological parameters collected at an urban background region in Amman, Jordan, in the period of 1 August 2016–31 July 2017. We develop and evaluate feed-forward neural network (FFNN) approaches to estimate number concentrations at particular size bin with (1) meteorological parameters, (2) number concentration at other size bins and (3) both of the above as input variables. Two layers with 10–15 neurons are found to be the optimal option. Worse performance is observed at the lower edge (0.01<Dp<0.02 µm), the mid-range region (0.15<Dp<0.5 µm) and the upper edge (6<Dp<10 µm). For the edges at both ends, the number of neighbouring size bins is limited, and the detection efficiency by the corresponding instruments is lower compared to the other size bins. A distinct performance drop over the overlapping mid-range region is due to the deficiency of a merging algorithm. Another plausible reason for the poorer performance for finer particles is that they are more effectively removed from the atmosphere compared to the coarser particles so that the relationships between the input variables and the small particles are more dynamic. An observable overestimation is also found in the early morning for ultrafine particles followed by a distinct underestimation before midday. In the winter, due to a possible sensor drift and interference artefacts, the estimation performance is not as good as the other seasons. The FFNN approach by meteorological parameters using 5 min data (R2= 0.22–0.58) shows poorer results than data with longer time resolution (R2= 0.66–0.77). The FFNN approach using the number concentration at the other size bins can serve as an alternative way to replace negative numbers in the size distribution raw dataset thanks to its high accuracy and reliability (R2= 0.97–1). This negative-number filling approach can maintain a symmetric distribution of errors and complement the existing ill-posed built-in algorithm in particle sizer instruments.

scholarly journals Particle Size Distribution of E-Cigarette Aerosols and the Relationship to Cambridge Filter Pad Collection Efficiency

2015 ◽  
Vol 26 (4) ◽  
Author(s):  
Steven L. Alderman ◽  
Chen Song ◽  
Serban C. Moldoveanu ◽  
Stephen K. Cole
Keyword(s):  
Particle Size ◽  
Particulate Matter ◽  
Particle Size Distribution ◽  
Size Distribution ◽  
Propylene Glycol ◽  
Collection Efficiency ◽  
Transmission Measurement ◽  
Average Particle ◽  
Electrical Mobility ◽  
Total Particulate Matter

AbstractThe relatively volatile nature of the particulate matter fraction of e-cigarette aerosols presents an experimental challenge with regard to particle size distribution measure-ments. This is particularly true for instruments requiring a high degree of aerosol dilution. This was illustrated in a previous study, where average particle diameters in the 10-50 nm range were determined by a high-dilution, electrical mobility method. Total particulate matter (TPM) masses calculated based on those diameters were orders of magnitude smaller than gravimetrically determined TPM. This discrepancy was believed to result from almost complete particle evaporation at the dilution levels of the electrical mobility analysis. The same study described a spectral transmission measurement of e-cigarette particle size in an undiluted state, and reported particles from 210-380 nm count median diameter. Observed particle number concentrations were in the 10Described here is a study in which e-cigarette aerosols were collected on Cambridge filters with adsorbent traps placed downstream in an effort to capture any material passing through the filter. Amounts of glycerin, propylene glycol, nicotine, and water were quantified on the filter and downstream trap. Glycerin, propylene glycol, and nicotine were effciently captured (> 98%) by the upstream Cambridge filter, and a correlation was observed between filtration efficiency and the partial vapor pressure of each component. The present analysis was largely inconclusive with regard to filter efficiency and particle-vapor partitioning of water. [Beitr. Tabakforsch. Int. 26 (2014) 183-190]

Determination of atmospheric aerosol particle size distribution and visibility using a mobile laboratory

1982 ◽  
Vol 60 (8) ◽  
pp. 1101-1107
Author(s):  
C. V. Mathai ◽  
A. W. Harrison
Keyword(s):  
Particle Size ◽  
Particle Size Distribution ◽  
Size Distribution ◽  
Atmospheric Aerosols ◽  
Suspended Particle ◽  
Aerosol Size Distribution ◽  
Size Distributions ◽  
Mobile Laboratory ◽  
Mean Values ◽  
The City

As part of an ongoing general research program on the effects of atmospheric aerosols on visibility and its dependence on aerosol size distributions in Calgary, this paper presents the results of a comparative study of particle size distribution and visibility in residential (NW) and industrial (SE) sections of the city using a mobile laboratory. The study was conducted in the period October–December, 1979. An active scattering aerosol spectrometer measured the size distributions and the corresponding visibilities were deduced from scattering coefficients measured with an integrating nephelometer.The results of this transit study show significantly higher suspended particle concentrations and reduced visibilities in the SE than in the NW. The mean values of the visibilities are 44 and 97 km for the SE and the NW respectively. The exponent of R (particle radius) in the power law aerosol size distribution has a mean value of −3.36 ± 0.24 in the SE compared with the corresponding value of −3.89 ± 0.39 for the NW. These results arc in good agreement with the observations of Alberta Environment; however, they are in contradiction with a recent report published by the City of Calgary.

scholarly journals Bayesian Retrievals of Vertically Resolved Cloud Particle Size Distribution Properties

2017 ◽  
Vol 56 (3) ◽  
pp. 745-765 ◽  
Author(s):  
Derek J. Posselt ◽  
James Kessler ◽  
Gerald G. Mace
Keyword(s):  
Remote Sensing ◽  
Particle Size ◽  
Particle Size Distribution ◽  
Size Distribution ◽  
Prior Knowledge ◽  
Unique Solution ◽  
Joint Probability Distribution ◽  
Sufficient Information ◽  
Near Infrared Reflectance ◽  
Cloud Properties

AbstractRetrievals of liquid cloud properties from remote sensing observations by necessity assume sufficient information is contained in the measurements, and in the prior knowledge of the cloudy state, to uniquely determine a solution. Bayesian algorithms produce a retrieval that consists of the joint probability distribution function (PDF) of cloud properties given the measurements and prior knowledge. The Bayesian posterior PDF provides the maximum likelihood estimate, the information content in specific measurements, the effect of observation and forward model uncertainties, and quantitative error estimates. It also provides a test of whether, and in which contexts, a set of observations is able to provide a unique solution. In this work, a Bayesian Markov chain Monte Carlo (MCMC) algorithm is used to sample the joint posterior PDF for retrieved cloud properties in shallow liquid clouds over the remote Southern Ocean. Combined active and passive observations from spaceborne W-band cloud radar and visible and near-infrared reflectance are used to retrieve the parameters of a gamma particle size distribution (PSD) for cloud droplets and drizzle. Combined active and passive measurements are able to distinguish between clouds with and without precipitation; however, unique retrieval of PSD properties requires specification of a scene-appropriate prior estimate. While much of the uncertainty in an unconstrained retrieval can be mitigated by use of information from 94-GHz passive brightness temperature measurements, simply increasing measurement accuracy does not render a unique solution. The results demonstrate the robustness of a Bayesian retrieval methodology and highlight the importance of an appropriately scene-consistent prior constraint in underdetermined remote sensing retrievals.

Sign in / Sign up

Export Citation Format

Share Document