Atmospheric Environment (v.39, #1)

The impact of congestion charging on vehicle emissions in London by Sean D. Beevers; David C. Carslaw (1-5).
The London congestion charging scheme (CCS) was successfully implemented in February 2003 and has measurably reduced traffic flows in central London. The air pollution impact of the scheme has been difficult to assess using ambient measurements alone as the air pollution concentrations in 2003 were higher than in 2002 because of unusual meteorological conditions. However, a comprehensive analysis of the impact using detailed traffic data, combined with the Environmental Research Group's road traffic emissions model, has identified a number of important results. First, between 2002 and 2003, total NO X emissions in the charging zone have reduced by −12.0% ±12% ( 2 σ ) and have increased on the inner ring road (IRR) by +1.5%. PM10 emissions have reduced by −11.9% in the charging zone and by −1.4% on the IRR. There is a significant reduction in the emissions of NO X and PM10 associated with increases in vehicle speed and that this is as important in reducing emissions as changes in vehicle numbers. There is also evidence that the speed changes in km h−1 are uniform across the whole range of average speed and therefore changes at the slower speeds have a disproportionate effect on vehicle emissions. Second, that changes in vehicle km, as a result of the scheme, are large (−15% ±4% ( 2 σ ) ) particularly in the charging zone itself. To meet the demand to travel into central London there has been increased bus use. However, the expected increase in emissions from buses have been mostly offset by the widespread introduction of particle traps to the new and existing bus fleet as well as the introduction of newer technology bus engines. Finally, there is a reduction in emissions of CO2 (−19.5%) but that unlike NO X and PM10 little additional benefit is apparent through new vehicle technology. The evidence presented shows that the congestion charging schemes could assist in attaining both the UK government's targets on air pollution as well as those relating to climate change and other international obligations.
Keywords: Environment and transport planning; Road user pricing; Climate change; Traffic emissions; Traffic congestion;

A comparison of bulk and wet-only deposition at two adjacent sites in Melle (Belgium) by Jeroen Staelens; An De Schrijver; Philip Van Avermaet; Gerrit Genouw; Niko Verhoest (7-15).
During 9 months, the weekly bulk and wet-only precipitation depositions in an urbanised region of Flanders (Belgium) were compared at two sites with a different height and separated by 1 km. The amount of rainfall at the two sites was similar, and the difference in ion deposition between the two sites was generally less than 5%. While the amount of rainfall measured was almost the same for both collector types, bulk deposition was significantly ( p < 0.02 ) higher than the wet deposition of all ions other than H+ and NH4 +. Averaged for both sites, bulk deposition was 129% (K+), 84% (Ca2+), 51% (Cl), 50% (Mg2+), 46% (Na+), 32% (SO4 2−), 27% (NO3 ), 17% (F), and 11% (NH4 +) higher than wet-only deposition. The acidity of bulk samples was significantly ( p < 0.06 ) lower than the acidity of wet-only samples. Bulk NH4 + concentrations were only significantly ( p < 0.002 ) higher than wet-only concentrations at one site because of the sensor-related, delayed closing of the wet-only lid at the second site. Although dry deposition significantly contributed to bulk precipitation measurements, bulk deposition exceeded the wet acidifying deposition of NO3 , NH4 +, and SO4 2− by less than 25%.
Keywords: Atmospheric deposition; Dry deposition; Monitoring; Precipitation chemistry; Wet deposition;

Polycyclic aromatic hydrocarbon emissions from motorcycles by Hsi-Hsien Yang; Lien-Te Hsieh; Hsu-Chung Liu; Hsiao-Hsuan Mi (17-25).
Emissions of polycyclic aromatic hydrocarbons (PAHs, 2–7 ring) and regulated air pollutants (CO, HC, NO x , PM) from 2-stroke carburetor (2-Stk/Cb), 4-stroke carburetor (4-Stk/Cb) and 4-stroke fuel injection (4-Stk/FI) motorcycles were investigated by testing these vehicles on a chassis dynamometer. Exhaust samplings were carried out on diluted exhausts in a dilution tunnel connected to a constant volume sampling system. Measurements were performed on a standard driving cycle. The results reveal that low molecular weight PAHs (especially naphthalene) dominated in the exhaust gas. The averages of soluble organic fractions were 86.4%, 46.3% and 48.9% for the 2-Stk/Cb, 4-Stk/Cb and 4-Stk/FI motorcycles, respectively. PAH emissions are greater from cold-start driving than those from hot-start driving cycle for all these three kinds of motorcycles. Total PAH emission factors were 8320, 5990 and 3390 μg km−1 for the in-used 2-Stk/Cb, 4-Stk/Cb and 4-Stk/FI motorcycles, respectively. PAH emission factors were the largest for the 2-Stk/Cb motorcycles. Besides, the 2-Stk/Cb motorcycle had the largest total BaP equivalent emission factor of 10.8 μg km−1, indicating that the emission exhaust from the 2-Stk/Cb motorcycle was most carcinogenic. HC, PM and PAH emissions were the lowest for the 4-Stk/FI motorcycles. The correlation coefficient between CO and total PAH emissions for all the test motorcycles was 0.51, indicating that CO and PAH emissions are not highly correlated.
Keywords: Toxicity equivalence factor; Emission factor; Soluble organic fraction; NO x ; CO; HC;

Fe-catalyzed photochemical oxidation of dissolved sulfur dioxide (SO2), S(IV), in the presence of oxalate has been studied under the conditions typical for acidified atmospheric water. The Fe(III)-catalyzed oxidation of S(IV) is first order with respect to S(IV) concentration in the absence of oxalate. The presence of oxalate strongly inhibits Fe(III)-catalyzed S(IV) oxidation due to the formation of Fe(III)–oxalato complexes both in the dark and under UV–visible light. However, under UV–visible light irradiation, the generation of photooxidants, such as H2O2 and OH radicals, by photochemical/chemical cycling of Fe(III)/Fe(II)–oxalato complexes, leads to the photochemical oxidation of S(IV). The initial oxidation rate increases with oxalate and Fe(III) concentration in the concentration ranges studied. The rate of Fe(III)-catalyzed photochemical oxidation of S(IV) in the presence of oxalate increases with increasing pH value of reaction solution and with decreasing irradiation wavelength. The concentrations of oxalate and other organic ligands are sufficiently large to significantly change the rates of Fe-catalyzed oxidation of dissolved SO2 in atmospheric water droplets during both day- and night time.
Keywords: Sulfite oxidation; Fe(III)–oxalato complexes; Photooxidation; Acid rain formation; Atmospheric aquatic chemistry;

Aerosol optical, chemical and physical properties at Gosan, Korea during Asian dust and pollution episodes in 2001 by Sang-Woo Kim; Soon-Chang Yoon; Anne Jefferson; John A. Ogren; Ellsworth G. Dutton; Jae-Gwang Won; Young Sung Ghim; Byung-Il Lee; Jin-Seok Han (39-50).
In order to understand the influence of dust and anthropogenic pollution aerosols on regional climate in East Asia, we analyzed the aerosol optical, chemical and physical properties for two cases with high aerosol loading and assessed the radiative forcing of these cases. The 1st case study is a heavy dust episode (DE) in April (during ACE-Asia) 2001 and the 2nd case is a regional-scale pollution event in November 2001. The Ångström exponent (Å) for DE was 0.38 from sunphotometer measurements. The mean single scattering albedo (550 nm) at the surface reported during the pollution episode (PE, 0.88) was lower than that of DE (0.91). The concentrations of organic (OC) and elemental carbon (EC) measured during the PE were about 90% and 30% higher than DE. The aerosol mass scattering efficiency (α s) of PE is a factor of about 2 higher than that of the DE. The difference in the mass absorption efficiency (α a) of EC during DE and PE is small and within the measurement uncertainty. The diurnally averaged aerosol radiative forcing efficiency (ΔDFE, W m−2τ 1) during DE is similar to results of other studies at Gosan.
Keywords: Asian dust and pollution; Light scattering; Light absorption; Aerosol radiative forcing; Gosan;

This study presents a new statistical model to predict the collection efficiency, cake thickness, cake porosity, and pressure drop across filter cakes during the particulate filtration of gases. This model is based on generation of a random distribution of particle sizes and particle falling locations. The model predicts the cake collection efficiency, which was found to be strongly dependent on the ratio of the mean particle size to the mean pore size of the filter medium. The average cake porosity decreases with increasing cake thickness and the pressure drop increases when the mean particle diameter decreases.
Keywords: Filter cake; Model; Porosity; Pressure drop;

Air toxics in ambient air of Delhi by Anjali Srivastava; A.E. Joseph; S. Patil; A. More; R.C. Dixit; M. Prakash (59-71).
Volatile organic compounds (VOCs) are major group of air pollutants which play critical role in atmospheric chemistry. It contributes to toxic oxidants which are harmful to ecosystem human health and atmosphere. Data on levels of VOCs in developing countries is lacking. In India information at target VOCs as defined in USEPA compendium method TO-14 is almost totally lacking. The present work deals with estimation of target VOCs at 15 locations in five categories namely residential, industrial, commercial, traffic intersections and petrol refueling stations in Delhi, the capital of India. The monitoring was carried out during peak hours in morning and evening each month for a year in 2001. Ambient air was adsorbed on adsorbent tubes, thermally desorbed and analyzed on GC–MS. The results show that levels of VOCs are high and stress the need for regular monitoring programme of VOCs in urban environment.
Keywords: Air toxics; VOCs; Delhi; Ambient air;

Waste heat released from human activities (anthropogenic heating) can be a significant contributor to the urban energy balance, and can thus play an important role in affecting the urban thermal environment, ambient air quality, and other attributes of the urban climate system. To quantify the impacts of anthropogenic heating we have incorporated it as a source term in the near-surface energy balance within the MM5 mesoscale atmospheric model. This energy balance is calculated as part of the planetary boundary layer (PBL) module within the MM5. Because of the multiple PBL scheme options available within the MM5 and other atmospheric modeling systems we have enabled anthropogenic heating within two commonly used PBL modules—Blackadar (BL) and Gayno–Seaman (GS). Results from a case study series of simulations for Philadelphia suggest that anthropogenic heating plays an important role in the formation of the urban heat island, particularly during the night and winter. Control simulations (without anthropogenic heating) consistently underestimated urban air temperatures and the observed urban heat island effect. Simulations for winter suggest that anthropogenic heating contributes 2–3 °C to the nighttime heat island. In addition, anthropogenic heating is also found to have impacts on the nocturnal PBL stability and PBL structure during the morning transition. The choice of PBL scheme affects the magnitude of these modeled impacts. In winter, for example, the addition of anthropogenic heating in the BL scheme resulted in a 3 °C temperature increase at night compared with about a 2 °C temperature increase in the corresponding GS simulation.
Keywords: Waste heat; Urban heat islands; Mesoscale modeling; PBL schemes;

Impact of the smoke aerosol from Russian forest fires on the atmospheric environment over Korea during May 2003 by Kwon H. Lee; Jeong E. Kim; Young J. Kim; Jhoon Kim; Wolfgang von Hoyningen-Huene (85-99).
Extensive forest fire activities occurred in May 2003 across Siberia, Russia, particularly in the area between the Amur and Lena rivers east of Lake Baikal. These forest fires released large amounts of particulates and gases into the atmosphere, resulting in adverse effects on regional air quality and radiation budget. On certain occasions, a smoke pollution plume from these forest fires was transported through Mongolia and eastern China, down to the Korean peninsula. In this study, satellite data and ground-based radiation measurement data were analyzed to estimate the smoke aerosol's impact on the local atmospheric environment over Korea. Aerosol optical depth (AOD) values retrieved using the Bremen Aerosol Retrieval method from Moderate Resolution Imaging Spectroradiometer data were compared with those derived from ground-based radiation measurements. Large AOD values in the range 2.0–4.0 were observed on 20 May 2003 over Korea due to the influence of the long-range transported smoke aerosol plume from the Russian fires, resulting in a surface-observed short-wavelength direct aerosol radiative forcing efficiency of −90 to −200 W m−2. This smoke aerosol plume also resulted in a decrease in the solar visible irradiance of up to 57%, and increased the surface PM10 concentration by up to 258 μg m−3.
Keywords: AOD; MODIS; Radiation; CO; PM10;

Estimate of mercury emission from gasoline and diesel fuel consumption, San Francisco Bay area, California by Christopher H. Conaway; Robert P. Mason; Douglas J. Steding; A. Russell Flegal (101-105).
Gasoline samples ( n = 20 ) collected from San Francisco Bay area refineries and service stations had mercury concentrations ranging from 0.08 to 1.4 ng g−1; diesel samples ( n = 19 ) had concentrations of 0.05–0.34 ng g−1. These relatively low levels show little evidence for mercury enrichment or contamination into these fuels from the refining process. Combustion of these fuels in the San Francisco Bay area contributes 0.7–13 kg Hg yr−1 to the environment, with an average of 5 kg Hg yr−1. Assuming the fate of this emission is to the atmosphere, the total flux from the combustion of these fuels represents less than 3% of the total atmospheric emissions in the San Francisco Bay area.
Keywords: Mercury; Gasoline; Diesel; San Francisco Bay;

Particle measurements were performed in the exhaust of five light-duty vehicles (Euro-3) at +23, −7, and −20 °C ambient temperatures. The characterization included measurements of particle number, active surface area, number size distribution, and mass size distribution. We investigated two port-injection spark-ignition (PISI) vehicles, a direct-injection spark-ignition (DISI) vehicle, a compressed ignition (CI) vehicle with diesel particle filter (DPF), and a CI vehicle without DPF. To minimize sampling effects, particles were directly sampled from the tailpipe with a novel porous tube diluter at controlled sampling parameters. The diluted exhaust was split into two branches to measure either all or only non-volatile particles. Effect of ambient temperature was investigated on particle emission for cold and warmed-up engine. For the gasoline vehicles and the CI vehicle with DPF, the main portion of particle emission was found in the first minutes of the driving cycle at cold engine start. The particle emission of the CI vehicle without DPF was hardly affected by cold engine start. For the PISI vehicles, particle number emissions were superproportionally increased in the diameter size range from 0.1 to 0.3 μm during cold start at low ambient temperature. Based on the particle mass size distribution, the DPF removed smaller particles ( dp < 0.5 μ m ) more efficiently than larger particles ( dp > 0.5 μ m ). No significant effect of ambient temperature was observed when the engine was warmed up. Peak emission of volatile nanoparticles only took place at specific conditions and was poorly repeatable. Nucleation of particles was predominately observed during or after strong acceleration at high speed and during regeneration of the DPF.
Keywords: Volatile nanoparticles; Particulate matter (PM); Soot; Subzero; Diesel particle filter (DPF);

It is difficult to conduct the assessment of dust storms under the influence of land surface indices with accurate statistical theory in the dust source region of China due to problems, especially those that emerge in the data of land surface that has no accurate data for soil moisture and hardness from meteorological records, no quantitative data for dust emission and a deficient proper theory using the capacity to combine quantitative data with qualitative data. This paper selected Erlianhaote of Inner Mongolia in northern China as a study example, and compiled the data of land surface condition of soil moisture and hardness as well as the data of dust storms and wind during 2001–2003. The authors determined the threshold value from the influences in different land patterns on dust emission processes by discrimination, quantitatively described the integrated influence of wind speed and soil moisture as well as soil hardness impact on dust storms using Hayashi's quantification theory type 2, and also discussed the related problems of potential errors in the data sets. This theorem could provide the obvious numeral materials about multi-factor cause–effect relationship for decision making and is useful in the quantificational assessment for factors impacting dust storms during the study period. Land surface condition diversity research had obvious scale effect characteristics. A clear evidence is that the combined contribution of soil moisture and hardness in dust emission is 48.79% of that of total contribution of wind speed, approximately the same as that of wind speed.
Keywords: Hayashi's quantification theory; Dust storms; Arid and semi-arid; Wind speed; Soil moisture; Soil hardness;

The performance of a predictive emission monitoring system (PEMS) as a technique for NO x monitoring on medium speed marine diesel engines has been evaluated for 16 similar engines on four different ships. The PEMS function tested measured O2 concentration in the exhaust gas, engine load, combustion air temperature and humidity, and barometric pressure to calculate the NO x concentration. Emission measurements were carried out by means of a conventional continuous emission monitoring system (CEMS) and the measured NO x concentrations were compared with those calculated by the PEMS function.For 11 of the 16 engines, the average error between measured and calculated NO x concentration was <10% of the calibration range (1725 ppm). In addition, 10 of the engines displayed correlation coefficients between measured and calculated NO x as 0.90 or higher. For two of the ships, the predicted NO x concentrations from all engines on board gave good agreement with those measured (2.6–4.7% and 2.6–8.0% average error). In other cases however, the performance of the PEMS function was poor e.g. the four engines of ship D showed average errors of 10.3–17.7%. Although similar engine models, fuel and load characteristics were compared in the tests, the specific NO x emissions at steady-state loads used varied from 12.6 up to 15.8 g k−1Whcorr. Although a single PEMS function may prove universal and adequate for calculating NO x emissions from similar engines on board the same ship, an engine specific PEMS function is recommended. The form of the PEMS function, i.e. using exhaust O2 and engine load as inputs, is however likely to be applicable to most propeller-law diesel engines.Bearing in mind the performance criteria for using PEMS at land-based installations, the results from this study are promising. Viewed as a single data set of 56 h with 16 separate engine comparisons between CEMS and PEMS, the data set shows a relative accuracy of 14.5% i.e. within the 20% requirement of the US Environmental Protection Agency. In light of the increased interest and international guidelines for continuous NO x monitoring on board ships, the PEMS technique can offer a simple but cost-effective option.
Keywords: NO x ; Emissions; Monitoring; PEMS; Ship;

Particle emissions from district heating units operating on three commonly used biofuels by A. Wierzbicka; L. Lillieblad; J. Pagels; M. Strand; A. Gudmundsson; A. Gharibi; E. Swietlicki; M. Sanati; M. Bohgard (139-150).
The aim of this study was to characterise particle emissions from district heating units operating on three commonly used biofuels: forest residues, pellets and sawdust. Boilers used in the three district heating units were of moving grate type, with the maximum thermal output between 1 and 1.5 MW. The measurements were done after multicyclones, the only particle removal devices installed, therefore the direct emissions to ambient air were characterised. Number and mass size distributions were determined. Elemental composition of the particles was determined by particle induced X-ray emissions analysis (PIXE) and thermal-optical analysis. Particles’ morphology was assessed on the basis of transmission electron microscopy (TEM). Total number concentration of emitted particles with aerodynamic diameter smaller than 5 μm (PM5) at medium operation load ranged from 6.3 to 7.7×107 particles/cmn 3, with the slightly higher values from combustion of forest residues. PM5 mass concentration at medium load from low pressure impactor measurements ranged between 51 and 120 mg/mn 3, with the highest values from unit operating on forest residues. Both PM5 mass and total number concentrations were dominated by fine mode contributions i.e. particles with aerodynamic diameter smaller than 1 μm (PM1). Elements determined by PIXE (Z>12) contributed to 21–34% of PM1 mass, of which K, S, Cl and Ca contributed to 18–33% of PM1 mass, and Zn, Mn, Fe, Cr, Pb and Cd to 1–3%. Emitted concentrations of heavy metals depended on type of the fuel and operating load. Particulate organic (OC) and elemental (EC) carbon contribution to PM1 ranged from 1–19% and 0–56%, respectively. Particulate OC concentrations strongly depended on the operation load regardless the type of the fuel, while EC concentrations seemed to depend both on load and the type of the fuel. Considering the potential public health implications of the obtained results, further research is needed to carefully assess the impact of particle emissions from biofuels combustion on human health and environment.
Keywords: Particle emissions; Biofuel; Combustion; District heating;

Monitoring data indicate that organic compounds with high octanol-air partition coefficients (K OA), such as polybrominated diphenyl ethers (PBDEs) and polychlorinated biphenyls (PCBs) exhibit seasonally variable air concentrations, especially during early spring, shortly after snow melt and before bud-burst when levels are elevated. This variability can complicate the interpretation of monitoring data designed to assess year-to-year changes. It is suggested that relatively simple dynamic multimedia mass balance models can assist interpretation by “factoring out” variability attributable to temperature and other seasonal effects as well as identifying likely contaminant sources. To illustrate this approach, high-volume air samples were collected from January to June, 2002 at a rural location in southern Ontario. Gas-phase concentrations for both ΣPBDE and ΣPCB rose from below the detection limit during the winter to 19 and 110 pg m−3, respectively, in early spring, only to decrease again following bud-burst. Passive air samples (PAS), deployed at seven urban, rural and remote sites for two one-month periods prior and following bud-burst, indicate a strong urban–rural gradient for both the PBDEs and PCBs. Calculated air concentrations from the PAS are shown to agree favorably with the high-volume air sampling data, with concentrations ranging 6–85 pg m−3 and 6–360 pg m−3 for ΣPBDE and ΣPCB, respectively. Concentrations in urban areas are typically 5 times greater than in rural locations. These data were interpreted using simulation results from a fate model including a seasonally variable forest canopy and snow pack, suggesting that the primary source is urban and that the “spring pulse” is the result of several interacting factors. Such contaminants are believed to be efficiently deposited in winter, accumulate in the snow pack and are released to terrestrial surfaces upon snow melt in spring. Warmer temperatures cause volatilization and a rise in air concentrations until uptake in emerging foliage leads to a decline in late spring. Implications for monitoring are discussed.
Keywords: Polybrominated diphenyl ethers; Spring pulse effect; Passive air samples; Long-range transport; Snow; Model; Monitoring;

Hourly mean concentration data for nitrogen oxides (NO X ), nitrogen dioxide (NO2) and ozone (O3) have been used to derive a method for estimating the mean primary NO2 fraction from vehicle exhausts in London. A set of simple chemical equations has been used to explain the differences in NO2 concentrations between background and roadside site pairs and to estimate the proportion of NO2 that is likely to be derived from primary NO2 vehicle emissions and the NO2 formed through the reaction of nitric oxide (NO) with O3. These results suggest that there is a wide range of primary NO2 emission fractions observed on different roads in London. The estimates ranged from 3.2 to 23.5 vol% with a median value of 10.6 vol% for the 43 measurement sites considered. A median primary NO2 fraction of 10.6% accounts for an average of 21% of the observed NO2 concentration at roadside sites. The range in calculated primary NO2 fractions explains much of the variation in annual mean NO2 concentrations between monitoring sites in London. The estimated primary NO2 fraction is considerably higher than the 5.0 vol% typically used in modelling studies that aim to predict ambient NO2 concentrations. These results have implications for the assessment of NO2 concentrations in London and whether London will meet the limit values set by the European Union. Further work is required to explain the wide range of primary NO2 fractions estimated in terms of road traffic composition and activity.
Keywords: Nitrogen dioxide; Dispersion modeling; Air quality assessment; Ozone; EU Daughter Directive;

Changes in Nordic surface ozone episodes due to European emission reductions in the 1990s by Sverre Solberg; Robert Bergström; Joakim Langner; Tuomas Laurila; Anne Lindskog (179-192).
Based on analyses of model calculations with a regional scale CTM for two different years and measurement data from background locations in northern Europe, we have found several indications that peak ozone values in the Nordic countries have been reduced during the 1990s as a result of reduced emissions of precursors in Europe. Official European emission data for 1999 gave a better model performance than the emission data for 1990 when modelling 1999 and 2000. A bootstrap resampling technique indicated that the improvement in performance was significant. The model predicted a reduction in peak ozone values of the order of 30 μg m−3 due to European emission reductions during the 1990s in the Nordic countries. It is thus likely that the number of exceedances of hourly threshold values has been reduced, although the small number of episodes does not allow strict statements. The number and magnitude of the ozone episodes as well as the model performance was clearly higher for southern Sweden and Norway compared to Finland, presumably reflecting differences in meteorological transport and emission source regions.
Keywords: EMEP; Ground level ozone; Trends; Model calculations; MATCH;

A graphical diagnostic method for assessing the rotation in factor analytical models of atmospheric pollution by Pentti Paatero; Philip K. Hopke; Bilkis A. Begum; Swapan K. Biswas (193-201).
Factor analytic tools such as principal component analysis (PCA) and positive matrix factorization (PMF), suffer from rotational ambiguity in the results: different solutions (factors) provide equally good fits to the measured data. The PMF model imposes non-negativity of both source profiles and source contributions in order to reduce the rotational problem. Such constraints are generally insufficient to ensure a unique solution. In the Unmix approach, edges of the multidimensional distribution of source contributions define the variable relationships in the factors. The present work extends this idea into an easy-to-use graphical procedure called G space plotting for PMF modeling. Scatter plots are created of pairs of source contribution factors. When factors are plotted in this way, unrealistic rotations appear as oblique edges that define the distribution of points away from one (or both) of the coordinate axes. With a correct rotation, the limiting edges usually coincide with the axes or lay parallel with them. Inspection of the plots helps one in choosing a realistic rotation.
Keywords: Edges; Unmix; Independent component analysis; ICA; Positive matrix factorization; PMF; G space plotting;