Forecasts using neural network versus Box-Jenkins methodology for ambient air quality monitoring data

This study explores ambient air quality forecasts using the conventional time-series approach and a neural network. Sulfur dioxide and ozone monitoring data collected from two background stations and an industrial station are used. Various learning methods and varied numbers of hidden layer processing units of the neural network model are tested. Results obtained from the time-series and neural network models are discussed and compared on the basis of their performance for 1-step-ahead and 24-step-ahead forecasts. Although both models perform well for 1-step-ahead prediction, some neural network results reveal a slightly better forecast without manually adjusting model parameters, according to the results. For a 24-step-ahead forecast, most neural network results are as good as or superior to those of the time-series model. With the advantages of self-learning, self-adaptation, and parallel processing, the neural network approach is a promising technique for developing an automated short-term ambient air quality forecast system.     

Preliminary data from the USEPA dry deposition network: 1989

The objective of the National Dry Deposition Network is to determine patterns and trends of dry deposition for various sulfur and nitrogen species at roughly 50 locations throughout the continental USA. Each site is equipped for collection of continuous meteorological and ozone data and weekly average concentrations of SO4(2-), NO3-, SO2 and HNO3, using a three-stage filter pack. Results from 40 eastern US sites operational throughout 1989 show species-dependent variability from site to site, season to season, and day to night. Annual average concentrations of atmospheric SO4(2-), NO3-, SO2 and HNO3 ranged from 2.7 to 7.9, 0.2 to 3.9, 2.4 to 23.2 and 0.7 to 3.6 microg/m(-3), respectively. Seasonal variability was considerable for all constituents. Day/night data indicate that SO2 and HNO3, but not SO4(2-) and NO3-, are typically found at moderately to substantially lower concentrations at night, especially during spring and summer. Estimated dry deposition for SO2 and HNO3 appear to be much greater than for SO4(2-) and NO3-, respectively. Comparison of measured wet deposition and estimated dry deposition at numerous sites suggests that the two are similar in magnitude over much of the eastern USA.     

Ground-level ozone in the Pearl River Delta region: Analysis of data from a recently established regional air quality monitoring network

The ambient air quality monitoring data of 2006 and 2007 from a recently established Pearl River Delta (PRD) regional air quality monitoring network are analyzed to investigate the characteristics of ground-level ozone in the region. Four sites covering urban, suburban, rural and coastal areas are selected as representatives for detailed analysis in this paper. The results show that there are distinct seasonal and diurnal cycles in ground-level ozone across the PRD region. Low ozone concentrations are generally observed in summer, while high O₃ levels are typically found in autumn. The O₃ diurnal variations in the urban areas are larger than those at the rural sites. The O₃ concentrations showed no statistically significant difference between weekend and weekdays in contrast to the findings in many other urban areas in the world. The average ozone concentrations are lower in urban areas compared to the sites outside urban centers. Back trajectories are used to show the major air-mass transport patterns and to examine the changes in ozone from the respective upwind sites to a site in the center of the PRD (Wanqingsha). The results show higher average ozone concentrations at the upwind sites in the continental and coastal air masses, but higher 1 h-max O₃ concentrations (by 8–16 ppbv) at the center PRD site under each of air-mass category, suggesting that the ozone pollution in the PRD region exhibits both regional and super-regional characteristics.     

Sulfate in precipitation as observed by the European atmospheric chemistry network

The temporal variation in sulfur concentration/deposition involves long term fluctuations which show striking similarities within certain areas but are different between areas in different positions relative to major sulfur emission areas in Europe. As an example, the deposition has been constant or decreasing during the last ten years in most of the area covered by the network, indicating that an increasing amount of the sulfur emitted in Europe is transported and deposited elsewhere, possibly in an eastward direction.The seasonal variation in deposition resembles that for amount of precipitation and shows a maximum during the summer or autumn for most areas. The concentration shows a maximum in the spring and a minimum in the autumn at most stations.The concentration field has a maximum approximately over Belgium and Holland with decreasing concentrations towards SW to NE (over W) but with comparatively high levels extending up over Finland. (Areas to the east and south are not covered by the network.)The deposition field was, based on a brief discussion of the relation between concentration and amount of precipitation, obtained as the product of these latter fields. High deposition rates are found in the south of Norway and in the middle of Britain in addition to the areas with high concentration.Sulfur and hydrogen ion are the dominating ions in precipitation followed by ammonium and nitrate except in coastal areas and in places where the soil is bare.Sources of errors in the data base are discussed and the results from a large number of additional sampling sites around and between the regular network sampling sites are most helpful in this regard. These latter measurements also permit an estimate of the uncertainty in areal concentration averages which are due to local and mesoscale variability.Finally, past and future importance of continuous measurements (as in the EACN) is discussed briefly.     

Estimating contribution of precipitation scavenging of atmospheric particulate mercury to mercury wet deposition in Japan

Mercury wet deposition is dependent on both the scavenging of divalent reactive gaseous mercury (RGM) and atmospheric particulate mercury (Hg(p)) by precipitation. Estimating the contribution of precipitation scavenging of RGM and Hg(p) is important for better understanding the causes of the regional and seasonal variations in mercury wet deposition. In this study, the contribution of Hg(p) scavenging was estimated on the basis of the scavenging ratios of other trace elements (i.e., Cd, Cu, Mn, Ni, Pb and V) existing entirely in particulate form. Their wet deposition fluxes and concentrations in air, which were measured concurrently from April 2004 to March 2005 at 10 sites in Japan, were used in this estimation. The monthly wet deposition flux of mercury at each site correlated with the amount of monthly precipitation, whereas the Hg(p) concentrations in air tended to decrease during summer. There was a significant correlation (P<0.001) among the calculated monthly average scavenging ratios of trace elements, and the values in each month at each site were similar. Therefore, it is assumed the monthly scavenging ratio of Hg(p) is equivalent to the mean value of other trace elements. Using this scavenging ratio (W), the wet deposition flux (F) due to Hg(p) scavenging in each month was calculated by F=WKP, where K and P are the Hg(p) concentration and amount of precipitation, respectively. Relatively large fluxes due to Hg(p) scavenging were observed at a highly industrial site and at sites on the Japan Sea coast, which are strongly affected by the local sources and the long-range transport from the Asian continent, respectively. However, on average, at the 10 sites, the contribution of Hg(p) scavenging to the annual mercury deposition flux was 26%, suggesting that mercury wet deposition in Japan is dominated by RGM scavenging. This RGM should originate mainly from the in situ oxidation of Hg⁰ in the atmosphere.     

An exploratory look at hydrocarbon data from the Photochemical Assessment Monitoring Stations network

This paper describes some characteristics of speciated nonmethane organic compound (NMOC) data collected in 1994 at five Photochemical Assessment Monitoring Stations (PAMS) and archived in the U.S. Environmental Protection Agency's Aerometric Information Retrieval System (AIRS). Topics include data completeness, distribution of individual NMOCs in concentration categories relative to minimum detectable levels, percentage of total NMOC associated with the sum of the 55 PAMS target compounds, and use of scatterplots to diagnose chromatographic misidentification of compounds. This is an early examination of a database that is expanding rapidly, and the insights presented here may be useful to both the producers and future users of the data for establishing consistency and quality control.     

The effect of improved nowcasting of precipitation on air quality modeling

The predictive potential of air quality models and thus their value in emergency management and public health support are critically dependent on the quality of their meteorological inputs. The atmospheric flow is the primary cause of the dispersion of airborne substances. The scavenging of pollutants by cloud particles and precipitation is an important sink of atmospheric pollution and subsequently determines the spatial distribution of the deposition of pollutants. The long-standing problem of the spin-up of clouds and precipitation in numerical weather prediction models limits the accuracy of the prediction of short-range dispersion and deposition from local sources. The resulting errors in the atmospheric concentration of pollutants also affect the initial conditions for the calculation of the long-range transport of these pollutants. Customary the spin-up problem is avoided by only using NWP (Numerical Weather Prediction) forecasts with a lead time greater than the spin-up time of the model. Due to the increase of uncertainty with forecast range this reduces the quality of the associated forecasts of the atmospheric flow.In this article recent improvements through diabatic initialization in the spin-up of large-scale precipitation in the Hirlam NWP model are discussed. In a synthetic example using a puff dispersion model the effect is demonstrated of these improvements on the deposition and dispersion of pollutants with a high scavenging coefficient, such as sulphur, and a low scavenging coefficient, such as cesium-137. The analysis presented in this article leads to the conclusion that, at least for situations where large-scale precipitation dominates, the improved model has a limited spin-up so that its full forecast range can be used. The implication for dispersion modeling is that the improved model is particularly useful for short-range forecasts and the calculation of local deposition. The sensitivity of the hydrological processes to proper initialization implies that the spin-up problem may reoccur with changes in the model and increased model resolution. Spin-up should be an ongoing concern for atmospheric modelers.     

Source extraction information from air quality data monitored in an Argentinean steel mill

A statistical analysis of a series of ambient air concentrations of suspended particulate matter (SPM) and NO2 is presented. Measurements were taken at four sites that belong to an Argentinean steel mill and in another site located in its vicinity. The air pollutants were measured during a three-week exploratory sampling. The monitoring sites were selected on the basis of relevant characteristics of the emission sources and the corresponding climatological statistics of the last decade. Suspended particulate matter with aerodynamic diameter of less than 10 microm (PM10) and NO2 were continuously measured at only one site, while 1-hr samples of NO2 and 24-hr samples of total SPM and SO2 were collected at the other sites. The registered concentrations show that SPM was the pollutant of major concern. A first estimate about the nature of the contribution of the different sources of particles and NO2 present in the area was obtained through the statistical analysis of measured concentration data coupled with prevalent meteorological variables.     

Estimating catchment nutrient flow with the HBV-NP model: sensitivity to input data

The dynamic catchment model HBV-N has been further developed by adding routines for phosphorus transport and is now called the HBV-NP model. The model was shown to satisfactorily simulate nutrient dynamics in the R?nne? catchment (1,900 km2). Its sensitivity to input data was tested, and results demonstrated the increased sensitivity to the selection of input data on a subcatchment scale when compared with the catchment scale. Selection of soil and land use databases was found to be critical in some subcatchments but did not have a significant impact on a catchment scale. Although acceptable on a catchment scale, using templates and generalization, with regards to emissions from point sources and rural households, significantly decreased model performance in certain subcatchments when compared with using more detailed local information. A division into 64 subcatchments resulted in similar model performance at the catchment outlet when compared with a lumped approach. Adjusting the imported matrixes of the regional leaching of nitrogen, from agricultural land, against mean subcatchment water percolation did not have a significant impact on the model performance.     

Neuro-fuzzy and neural network systems for air quality control

In order to define efficient air quality plans, Regional Authorities need suitable tools to evaluate both the impact of emission reduction strategies on pollution indexes and the costs of such emission reductions. The air quality control can be formalized as a two-objective nonlinear mathematical problem, integrating source–receptor models and the estimate of emission reduction costs. Both aspects present several complex elements. In particular the source–receptor models cannot be implemented through deterministic modelling systems, that would bring to a computationally unfeasible mathematical problem. In this paper we suggest to identify source–receptor statistical models (neural network and neuro-fuzzy) processing the simulations of a deterministic multi-phase modelling system (GAMES). The methodology has been applied to ozone and PM10 concentrations in Northern Italy. The results show that, despite a large advantage in terms of computational costs, the selected source–receptor models are able to accurately reproduce the simulation of the 3D modelling system.     

National crop loss assessment network: quality assurance program

A quality assurance program was incorporated into the National Crop Loss Assessment Network (NCLAN) program, designed to assess the economic impacts of gaseous air pollutants on major agricultural crops in the United States. To satisfy US EPA requirements that all environmental data collected be of known and documented quality, adequate for the intended use, the quality assurance program developed standardized research and monitoring protocols among sites, and included a range of audit and review procedures. The goal of the quality assurance program was to quantitatively describe the overall quality of data collected in terms of precision, accuracy, completeness, representativeness, and comparability. From this program, it can be concluded that (1) project data quality objectives were valuable for determining the acceptability of data from diverse sites, (2) standardized protocols ensured data comparability among research sites, (3) independent on-site audits served to evaluate protocol adherence, and (4) precision and accuracy measurements provided a way to assess data quality, determine data acceptability, and indicate the need for instrument adjustment or repair.     

Estimating chemical biotransformation rates from food web concentrations

Biotransformation is widely recognized as the most important and most uncertain determinant of bioaccumulation. A step-wise method for estimating organism-specific biotransformation half-lives from field observations and using established food web modeling is developed. As a proof of concept, the method is applied to the case of nine polycyclic aromatic hydrocarbons (PAHs) in a well-studied food web in Bohai Bay, China. The estimated half-lives are in good agreement with the existing literature. The proposed biotransformation estimation method, through data mining, for sufficiently defined ecosystems, may greatly reduce the necessary animal testing involved in chemical assessments by providing useful guidance to experimentalists and regulators.     

PM2.5 speciation trends network: evaluation of whole-system uncertainties using data from sites with collocated samplers

The objectives of this paper are to contrast the relative variability of replicate laboratory measurements of selected chemical components of fine particulate matter (PM) with total variability from collocated measurements and to compare the magnitudes of the uncertainties determined from collocated sampler data with those currently being provided to U.S. Environmental Protection Agency (EPA)'s Air Quality System (AQS) database by RTI International (RTI). Pointwise uncertainty values are needed for modeling and data analysis and should include all the random errors affecting each data point. Total uncertainty can be decomposed into two primary components: analytical measurement uncertainty and sampling uncertainty. Analytical measurement uncertainties are relatively easy to calculate from routine quality control (QC) data. Sampling uncertainties, on the other hand, are comparatively difficult to measure. In this paper, the authors describe data from collocated samplers to provide a snapshot of whole-system uncertainty for several important chemical species. The components of uncertainty were evaluated for key species from each of the analytical methods employed by the PM2.5 Speciation Trends Network (STN) program: gravimetry, ion chromatography (IC), X-ray fluorescence (XRF), and thermal-optical analysis for organic carbon and elemental carbon. The results show that the laboratory measurement uncertainties are typically very small compared with uncertainties calculated from the differences between samples collected from collocated samplers. These differences are attributable to the "field" components uncertainty, which may include contamination and/or losses during shipping, handling, and sampling, as well as other distortions of the concentration level due to flow and sample volume variations. Uncertainties calculated from the collocation results were found to be generally similar to the uncertainties currently being loaded into EPA's AQS system, with some exceptions described below.     

interpretation of air quality data with respect to the national ambient air quality standards

San Diego Gas &; Electric has developed a quality assurance program for continuous emission monitors (CEM). Extractive, rather than in situ, monitors were selected as a result of an in-house evaluation program. Two extractive systems have been certified and a good operating and maintenance record has been established on these systems. A successful program requires the involvement and support of all affected personnel. It is desirable to have one or two key personnel coordinate the development of the program. It is also highly desirable to have good in-house source testing capabilities.     

Estimating Fugitive Emissions of Volatile Compounds from Equipment Leaks

This paper is directed to those concerned with estimating releases to the air of volatile compounds from equipment component leaks. We compared emission rates for equipment component leaks using EPA’s Average Emission Factor Method, Leak/No-Leak Emission Factor Method, and Stratified Emission Factor Method. The latter two methods may be used if there are concentration measurements about the components. An organic vapor analyzer was applied with the appropriate response factor to measure the concentrations around each equipment component using EPA Method 21. We analyzed this data to determine mass emission rates. Results show that the Average Emission Factor Method substantially overstates the emission rates calculated using these other two methods. We conclude that use of Method 21 concentration measurements with the Leak/No Leak or Stratified Emission Factor Method may increase the accuracy and reduce significantly the estimated releases to the air of volatile compounds from equipment component leaks for many facilities.     

基于改进型灰色神经网络组合模型的空气质量预测

基于空气质量数据不足及波动较大的情况,将灰色GM（1,1）模型与人工神经网络模型组合并改进,建立改进型灰色神经网络组合模型。利用天津市2001—2008年PM10、SO2和NO2年均值作为原始数据预测2009—2010年PM10、SO2和NO2的浓度以进行模型精度检验,最后利用该模型预测2011—2015年天津市空气质量状况。结果表明,与灰色GM（1,1）模型、传统灰色神经网络组合模型相比,所建立的改进型灰色神经网络组合模型相对模拟误差小,预测结果更为可靠,可以用于空气质量预测。     

EuroBionet: a pan-European biomonitoring network for urban air quality assessment

EuroBionet, the 'European Network for the Assessment of Air Quality by the Use of Bioindicator Plants', is an EU-funded cooperative project currently consisting of public authorities and scientific institutes from 12 cities in 8 countries. In 2000, the bioindicator plants tobacco (Nicotiana tabacum Bel W3), poplar (Populus nigra 'Brandaris'), spiderwort (Tradescantia sp. clone 4430), Italian rye grass (Lolium multiflorum italicum) and curly kale (Brassica oleracea acephala) were exposed to ambient air at 90 monitoring sites according to standardised methods. Visible injuries and growth parameters were assessed and the accumulation of toxic substances in leaves determined. The exposure of tobacco resulted in a gradient with low levels of ozone-induced foliar injury in N and NW Europe, and medium to high values in the southern and central regions. The results of heavy metal and sulphur analyses in rye grass samples generally showed low to very low sulphur and low to medium heavy metal concentrations in leaves. In some cities, however, local hot spots of heavy metal contamination were detected. Analyses of the PAH contents in curly kale leaves gave low to medium values, with locally elevated levels at traffic-exposed sites.