scholarly journals Positive matrix factorization of organic aerosol: insights from a chemical transport model

2019 ◽  
Vol 19 (2) ◽  
pp. 973-986 ◽  
Author(s):  
Anthoula D. Drosatou ◽  
Ksakousti Skyllakou ◽  
Georgia N. Theodoritsi ◽  
Spyros N. Pandis
Keyword(s):  
Matrix Factorization ◽  
Transport Model ◽  
Chemical Transport ◽  
Organic Aerosol ◽  
Positive Matrix Factorization ◽  
The Other ◽  
Chemical Transport Model ◽  
Positive Matrix ◽  
Aerosol Mass ◽  
High Volatility

Abstract. Factor analysis of aerosol mass spectrometer measurements (organic aerosol mass spectra) is often used to determine the sources of organic aerosol (OA). In this study we aim to gain insights regarding the ability of positive matrix factorization (PMF) to identify and quantify the OA sources accurately. We performed PMF and multilinear engine (ME-2) analysis on the predictions of a state-of-the-art chemical transport model (PMCAMx-SR, Particulate Matter Comprehensive Air Quality Model with extensions – source resolved) during a photochemically active period for specific sites in Europe in an effort to interpret the diverse factors usually identified by PMF analysis of field measurements. Our analysis used the predicted concentrations of 27 OA components, assuming that each of them is “chemically different” from the others. The PMF results based on the chemical transport model predictions are quite consistent (same number of factors and source types) with those of the analysis of AMS measurements. The estimated uncertainty of the contribution of fresh biomass burning is less than 30 % and of the other primary sources less than 40 %, when these sources contribute more than 20 % to the total OA. The PMF uncertainty increases for smaller source contributions, reaching a factor of 2 or even 3 for sources which contribute less than 10 % to the OA. One of the major questions in PMF analysis of AMS measurements concerns the sources of the two or more oxygenated OA (OOA) factors often reported in field studies. Our analysis suggests that these factors include secondary OA compounds from a variety of anthropogenic and biogenic sources and do not correspond to specific sources. Their characterization in the literature as low- and high-volatility factors is probably misleading, because they have overlapping volatility distributions. However, the average volatility of the one often characterized as a low-volatility factor is indeed lower than that of the other (high-volatility factor). Based on the analysis of the PMCAMx-SR predictions, the first oxygenated OA factor includes mainly highly aged OA transported from outside Europe, but also highly aged secondary OA from precursors emitted in Europe. The second oxygenated OA factor contains fresher secondary organic aerosol from volatile, semivolatile, and intermediate volatility anthropogenic and biogenic organic compounds. The exact contribution of these OA components to each OA factor depends on the site and the prevailing meteorology during the analysis period.

scholarly journals Positive Matrix Factorization of Organic Aerosol: Insights from a Chemical Transport Model

2018 ◽  
Author(s):  
Anthoula D. Drosatou ◽  
Ksakousti Skyllakou ◽  
Georgia N. Theodoritsi ◽  
Spyros N. Pandis
Keyword(s):  
Matrix Factorization ◽  
Transport Model ◽  
Chemical Transport ◽  
Organic Aerosol ◽  
Positive Matrix Factorization ◽  
The Other ◽  
Chemical Transport Model ◽  
Positive Matrix ◽  
Aerosol Mass ◽  
High Volatility

Abstract. Factor analysis of Aerosol Mass Spectrometer measurements (organic aerosol mass spectra) is often used to determine the sources of organic aerosol (OA). In this study we aim to gain insights regarding the ability of positive matrix factorization (PMF) to identify and quantify the OA sources accurately. We performed PMF and multilinear engine (ME-2) analysis on the predictions of a state-of-the-art chemical transport model (PMCAMx-SR) during a photochemically active period for specific sites in Europe in an effort to interpret the diverse factors usually identified by PMF analysis of field measurements. Our analysis used the predicted concentrations of 27 OA components, assuming that each of them is chemically different from the others. The PMF results based on the chemical transport model predictions are quite consistent (same number of factors and source types) with those of the analysis of AMS measurements. The estimated uncertainty of the contribution of fresh biomass burning is less than 30 % and of the other primary sources less than 40 %, when these sources contribute more than 20 % to the total OA. For contributions between 10 and 20 % the corresponding uncertainties increase to 50 %. Finally, when these sources are small (less than 10 % of the OA) the corresponding error is a factor of two or even three. One of the major questions in PMF analysis of AMS measurements concerns the sources of the two or more oxygenated OA (OOA) factors often reported in field studies. Our analysis suggests that these factors include secondary OA compounds from a variety of anthropogenic and biogenic sources and do not correspond to specific sources. Their characterization in the literature as low and high volatility factors is probably misleading, because they have overlapping volatility distributions. However, the average volatility of the one often characterized as low-volatility factor is indeed lower than that of the other (high volatility factor). Based on the analysis of the PMCAMx-SR predictions, the first oxygenated OA factor includes mainly highly-aged OA transported from outside Europe, but also highly aged secondary OA from precursors emitted in Europe. The second oxygenated OA factor contains fresher SOA from volatile, semi-volatile, and intermediate volatility anthropogenic and biogenic organic compounds. The exact contribution of these OA components to each OA factor depends on the site and the prevailing meteorology during the analysis period.

scholarly journals Development of a versatile source apportionment analysis based on positive matrix factorization: a case study of the seasonal variation of organic aerosol sources in Estonia

2019 ◽  
Vol 19 (11) ◽  
pp. 7279-7295 ◽  
Author(s):  
Athanasia Vlachou ◽  
Anna Tobler ◽  
Houssni Lamkaddam ◽  
Francesco Canonaco ◽  
Kaspar R. Daellenbach ◽  
...  
Keyword(s):  
Matrix Factorization ◽  
Organic Aerosol ◽  
Positive Matrix Factorization ◽  
Water Soluble ◽  
Oxidation Products ◽  
Positive Matrix ◽  
Bootstrap Analysis ◽  
Seasonal Behavior ◽  
Yearly Cycle ◽  
Aerosol Mass

Abstract. Bootstrap analysis is commonly used to capture the uncertainties of a bilinear receptor model such as the positive matrix factorization (PMF) model. This approach can estimate the factor-related uncertainties and partially assess the rotational ambiguity of the model. The selection of the environmentally plausible solutions, though, can be challenging, and a systematic approach to identify and sort the factors is needed. For this, comparison of the factors between each bootstrap run and the initial PMF output, as well as with externally determined markers, is crucial. As a result, certain solutions that exhibit suboptimal factor separation should be discarded. The retained solutions would then be used to test the robustness of the PMF output. Meanwhile, analysis of filter samples with the Aerodyne aerosol mass spectrometer and the application of PMF and bootstrap analysis on the bulk water-soluble organic aerosol mass spectra have provided insight into the source identification and their uncertainties. Here, we investigated a full yearly cycle of the sources of organic aerosol (OA) at three sites in Estonia: Tallinn (urban), Tartu (suburban) and Kohtla-Järve (KJ; industrial). We identified six OA sources and an inorganic dust factor. The primary OA types included biomass burning, dominant in winter in Tartu and accounting for 73 % ± 21 % of the total OA, primary biological OA which was abundant in Tartu and Tallinn in spring (21 % ± 8 % and 11 % ± 5 %, respectively), and two other primary OA types lower in mass. A sulfur-containing OA was related to road dust and tire abrasion which exhibited a rather stable yearly cycle, and an oil OA was connected to the oil shale industries in KJ prevailing at this site that comprises 36 % ± 14 % of the total OA in spring. The secondary OA sources were separated based on their seasonal behavior: a winter oxygenated OA dominated in winter (36 % ± 14 % for KJ, 25 % ± 9 % for Tallinn and 13 % ± 5 % for Tartu) and was correlated with benzoic and phthalic acid, implying an anthropogenic origin. A summer oxygenated OA was the main source of OA in summer at all sites (26 % ± 5 % in KJ, 41 % ± 7 % in Tallinn and 35 % ± 7 % in Tartu) and exhibited high correlations with oxidation products of a-pinene-like pinic acid and 3-methyl-1, 2, 3-butanetricarboxylic acid (MBTCA), suggesting a biogenic origin.

scholarly journals Analysis of secondary organic aerosol formation and aging using positive matrix factorization of high-resolution aerosol mass spectra: application to the dodecane low-NO<sub>x</sub> system

2012 ◽  
Vol 12 (24) ◽  
pp. 11795-11817 ◽  
Author(s):  
J. S. Craven ◽  
L. D. Yee ◽  
N. L. Ng ◽  
M. R. Canagaratna ◽  
C. L. Loza ◽  
...  
Keyword(s):  
High Resolution ◽  
Matrix Factorization ◽  
Mass Spectra ◽  
Secondary Organic Aerosol ◽  
Organic Aerosol ◽  
Positive Matrix Factorization ◽  
Aerosol Formation ◽  
Positive Matrix ◽  
Aerosol Mass ◽  
Van Krevelen Diagram

Abstract. Positive matrix factorization (PMF) of high-resolution laboratory chamber aerosol mass spectra is applied for the first time, the results of which are consistent with molecular level MOVI-HRToF-CIMS aerosol-phase and CIMS gas-phase measurements. Secondary organic aerosol was generated by photooxidation of dodecane under low-NOx conditions in the Caltech environmental chamber. The PMF results exhibit three factors representing a combination of gas-particle partitioning, chemical conversion in the aerosol, and wall deposition. The slope of the measured high-resolution aerosol mass spectrometer (HR-ToF-AMS) composition data on a Van Krevelen diagram is consistent with that of other low-NOx alkane systems in the same O : C range. Elemental analysis of the PMF factor mass spectral profiles elucidates the combinations of functionality that contribute to the slope on the Van Krevelen diagram.

scholarly journals Sources and production of organic aerosol in Mexico City: insights from the combination of a chemical transport model (PMCAMx-2008) and measurements during MILAGRO

2011 ◽  
Vol 11 (11) ◽  
pp. 5153-5168 ◽  
Author(s):  
A. P. Tsimpidi ◽  
V. A. Karydis ◽  
M. Zavala ◽  
W. Lei ◽  
N. Bei ◽  
...  
Keyword(s):  
Mexico City ◽  
Biomass Burning ◽  
Urban Areas ◽  
Transport Model ◽  
Three Dimensional ◽  
Chemical Transport ◽  
Organic Aerosol ◽  
Basis Set ◽  
Chemical Transport Model ◽  
Aerosol Mass

Abstract. Urban areas are large sources of organic aerosols and their precursors. Nevertheless, the contributions of primary (POA) and secondary organic aerosol (SOA) to the observed particulate matter levels have been difficult to quantify. In this study the three-dimensional chemical transport model PMCAMx-2008 is used to investigate the temporal and geographic variability of organic aerosol in the Mexico City Metropolitan Area (MCMA) during the MILAGRO campaign that took place in the spring of 2006. The organic module of PMCAMx-2008 includes the recently developed volatility basis-set framework in which both primary and secondary organic components are assumed to be semi-volatile and photochemically reactive and are distributed in logarithmically spaced volatility bins. The MCMA emission inventory is modified and the POA emissions are distributed by volatility based on dilution experiments. The model predictions are compared with observations from four different types of sites, an urban (T0), a suburban (T1), a rural (T2), and an elevated site in Pico de Tres Padres (PTP). The performance of the model in reproducing organic mass concentrations in these sites is encouraging. The average predicted PM1 organic aerosol (OA) concentration in T0, T1, and T2 is 18 μg m−3, 11.7 μg m−3, and 10.5 μg m−3 respectively, while the corresponding measured values are 17.2 μg m−3, 11 μg m−3, and 9 μg m−3. The average predicted locally-emitted primary OA concentrations, 4.4 μg m−3 at T0, 1.2 μg m−3 at T1 and 1.7 μg m−3 at PTP, are in reasonably good agreement with the corresponding PMF analysis estimates based on the Aerosol Mass Spectrometer (AMS) observations of 4.5, 1.3, and 2.9 μg m−3 respectively. The model reproduces reasonably well the average oxygenated OA (OOA) levels in T0 (7.5 μg m−3 predicted versus 7.5 μg m−3 measured), in T1 (6.3 μg m−3 predicted versus 4.6 μg m−3 measured) and in PTP (6.6 μg m−3 predicted versus 5.9 μg m−3 measured). The rest of the OA mass (6.1 μg m−3 and 4.2 μg m−3 in T0 and T1 respectively) is assumed to originate from biomass burning activities and is introduced to the model as part of the boundary conditions. Inside Mexico City (at T0), the locally-produced OA is predicted to be on average 60 % locally-emitted primary (POA), 6 % semi-volatile (S-SOA) and intermediate volatile (I-SOA) organic aerosol, and 34 % traditional SOA from the oxidation of VOCs (V-SOA). The average contributions of the OA components to the locally-produced OA for the entire modelling domain are predicted to be 32 % POA, 10 % S-SOA and I-SOA, and 58 % V-SOA. The long range transport from biomass burning activities and other sources in Mexico is predicted to contribute on average almost as much as the local sources during the MILAGRO period.

scholarly journals Interpretation of organic components from Positive Matrix Factorization of aerosol mass spectrometric data

2009 ◽  
Vol 9 (9) ◽  
pp. 2891-2918 ◽  
Author(s):  
I. M. Ulbrich ◽  
M. R. Canagaratna ◽  
Q. Zhang ◽  
D. R. Worsnop ◽  
J. L. Jimenez
Keyword(s):  
Detailed Analysis ◽  
Matrix Factorization ◽  
Synthetic Data ◽  
Organic Aerosol ◽  
Positive Matrix Factorization ◽  
Positive Matrix ◽  
Spectrometric Data ◽  
Residual Structure ◽  
Aerosol Mass ◽  
Number Of Factors

Abstract. The organic aerosol (OA) dataset from an Aerodyne Aerosol Mass Spectrometer (Q-AMS) collected at the Pittsburgh Air Quality Study (PAQS) in September 2002 was analyzed with Positive Matrix Factorization (PMF). Three components – hydrocarbon-like organic aerosol OA (HOA), a highly-oxygenated OA (OOA-1) that correlates well with sulfate, and a less-oxygenated, semi-volatile OA (OOA-2) that correlates well with nitrate and chloride – are identified and interpreted as primary combustion emissions, aged SOA, and semivolatile, less aged SOA, respectively. The complexity of interpreting the PMF solutions of unit mass resolution (UMR) AMS data is illustrated by a detailed analysis of the solutions as a function of number of components and rotational forcing. A public web-based database of AMS spectra has been created to aid this type of analysis. Realistic synthetic data is also used to characterize the behavior of PMF for choosing the best number of factors, and evaluating the rotations of non-unique solutions. The ambient and synthetic data indicate that the variation of the PMF quality of fit parameter (Q, a normalized chi-squared metric) vs. number of factors in the solution is useful to identify the minimum number of factors, but more detailed analysis and interpretation are needed to choose the best number of factors. The maximum value of the rotational matrix is not useful for determining the best number of factors. In synthetic datasets, factors are "split" into two or more components when solving for more factors than were used in the input. Elements of the "splitting" behavior are observed in solutions of real datasets with several factors. Significant structure remains in the residual of the real dataset after physically-meaningful factors have been assigned and an unrealistic number of factors would be required to explain the remaining variance. This residual structure appears to be due to variability in the spectra of the components (especially OOA-2 in this case), which is likely to be a key limit of the retrievability of components from AMS datasets using PMF and similar methods that need to assume constant component mass spectra. Methods for characterizing and dealing with this variability are needed. Interpretation of PMF factors must be done carefully. Synthetic data indicate that PMF internal diagnostics and similarity to available source component spectra together are not sufficient for identifying factors. It is critical to use correlations between factor and external measurement time series and other criteria to support factor interpretations. True components with <5% of the mass are unlikely to be retrieved accurately. Results from this study may be useful for interpreting the PMF analysis of data from other aerosol mass spectrometers. Researchers are urged to analyze future datasets carefully, including synthetic analyses, and to evaluate whether the conclusions made here apply to their datasets.

scholarly journals Analysis of secondary organic aerosol formation and aging using positive matrix factorization of high-resolution aerosol mass spectra: application to the dodecane low-NO<sub>x</sub> system

2012 ◽  
Vol 12 (7) ◽  
pp. 16647-16699 ◽  
Author(s):  
J. S. Craven ◽  
L. D. Yee ◽  
N. L. Ng ◽  
M. R. Canagaratna ◽  
C. L. Loza ◽  
...  
Keyword(s):  
High Resolution ◽  
Matrix Factorization ◽  
Mass Spectra ◽  
Secondary Organic Aerosol ◽  
Organic Aerosol ◽  
Positive Matrix Factorization ◽  
Aerosol Formation ◽  
Positive Matrix ◽  
Aerosol Mass ◽  
Van Krevelen Diagram

Abstract. Positive matrix factorization (PMF) of high-resolution laboratory aerosol mass spectra is applied for the first time, the results of which are consistent with molecular level MOVI-HRToF-CIMS aerosol-phase and CIMS gas-phase measurements. Secondary organic aerosol was generated by photooxidation of dodecane under low-NOx conditions in the Caltech environmental chamber. The PMF results exhibit three factors representing a combination of gas-particle partitioning, chemical conversion in the aerosol, and wall deposition. The slope of the measured high-resolution aerosol mass spectrometer (HR-ToF-AMS) composition data on a Van Krevelen diagram is consistent with that of other low-NOx alkane systems in the same O:C range. Elemental analysis of the PMF factor mass spectral profiles elucidates the combinations of functionality that contribute to the slope on the Van Krevelen diagram.

scholarly journals Interpretation of organic components from positive matrix factorization of aerosol mass spectrometric data

2008 ◽  
Vol 8 (2) ◽  
pp. 6729-6791 ◽  
Author(s):  
I. M. Ulbrich ◽  
M. R. Canagaratna ◽  
Q. Zhang ◽  
D. R. Worsnop ◽  
J. L. Jimenez
Keyword(s):  
Time Series ◽  
Detailed Analysis ◽  
Matrix Factorization ◽  
Synthetic Data ◽  
Organic Aerosol ◽  
Positive Matrix Factorization ◽  
Positive Matrix ◽  
Rotational Parameter ◽  
Aerosol Mass ◽  
Number Of Factors

Abstract. The organic aerosol (OA) dataset from an Aerodyne Aerosol Mass Spectrometer (Q-AMS) collected at the Pittsburgh Air Quality Study in September 2002 was analyzed for components with Positive Matrix Factorization (PMF). Three components – hydrocarbon-like organic aerosol OA (HOA), a highly-oxygenated OA (OOA-I) that correlates well with sulfate, and a less-oxygenated, semi-volatile OA (OOA-II) that correlates well with nitrate and chloride – are identified and interpreted as primary combustion emissions, aged SOA, and semivolatile, less aged SOA, respectively. The complexity of interpreting the PMF solutions of unit mass resolution (UMR) AMS data is illustrated by a detailed analysis of the solutions as a function of number of components and rotational state. A public database of AMS spectra has been created to aid this type of analysis. A sensitivity analysis with realistic synthetic data is also used to characterize the behavior of PMF for choosing the best number of factors, rotations of non-unique solutions, and the retrievability of more (or less) correlated factors. The ambient and synthetic data indicate that the variation of the PMF quality of fit parameter (Q, a normalized chi-squared metric) vs. number of factors in the solution is useful to identify the minimum number of factors, but more detailed analysis and interpretation is needed to choose the best number of factors. The maximum value of the rotational matrix is not useful for determining the best number of factors. In synthetic datasets, factors are "split" into two or more components when solving for more factors than were used in the input. Elements of the "splitting" behavior are observed in solutions of real datasets with several factors. Significant structure remains in the residual of the real dataset after physically-meaningful factors have been assigned and an unrealistic number of factors would be required to explain the remaining variance. This residual structure appears to be due to variability in the spectra of the components (especially OOA-II in this case), which is likely to be a key limit of the retrievability of components from AMS datasets using PMF and similar methods that need to assume constant component mass spectra. Methods for characterizing and dealing with this variability are needed. Values of the rotational parameter (FPEAK) near zero appear to be most appropriate for these datasets. Interpretation of PMF factors must be done carefully. Synthetic data indicate that PMF internal diagnostics and similarity to available source component spectra together are not sufficient for identifying factors. It is critical to use correlations between factor time series and external measurement time series to support factor interpretations. Components with <5% of the mass or with high correlation (R>0.9) with other components are suspect and should be interpreted with care. Results from this study may be useful for interpreting the PMF analysis of data from other aerosol mass spectrometers.

Sign in / Sign up

Export Citation Format

Share Document