Application of statistical methods to diagnose causes of poor air-quality model performance

The application of statistical tests has been too often restricted to defining performance measures that are a necessary part of the model validation process. Toward the goal of model improvement, statistics can also be applied in a diagnostic mode that relates the differences in observed and predicted concentrations to other parameters that influence the model output. The examples presented described how diagnostic tests were used to isolate problems in an air quality simulation model using data from the St. Louis-based Regional Air Pollution Study.     

Validation of trajectory statistical methods

A number of works have been trying to validate various trajectory statistical methods (TSMs), mostly through subjective comparison with known sources. Here in a more comprehensive and quantitative approach three trajectory statistical methods (potential source contribution function (PSCF), concentration field method (CF), and redistributed concentration field method (RCF)) were subjected to two validation approaches: validation with virtual and real sources under idealised conditions, where the effects of dispersion and removal of the trace substance are excluded, and comparisons with the EMEP SO₂ emission inventory under realistic conditions.The best performance was achieved in an idealised situation with about 78% common spatial variance between the EMEP emission inventory and the trajectory statistical reconstruction of the EMEP emission inventory with the RCF method, whereas the real world experiments for SO₂ on an European scale resulted in a much lower performance with 33% common spatial variance between the EMEP SO₂ emission inventory and the trajectory statistical reconstruction with the PSCF method.The experiments suggest that the limitation of the accuracy and spatial range of TSMs are rooted in the simplified transport process described just by trajectory paths. If one links these limitations with the concept of the mean residence time of the considered trace substance, a temporal and spatial scope can be deduced, within which the effect of the simplification of the transport process is restricted and useful information can be expected from TSMs. The lower values of the mean residence time for SO₂ range from 9 to 17 h, which were deduced from the decay approach, where an exponential decay, respectively, removal of SO₂ was built into the trajectory statistical procedure. The values derived from the optimum real world validation experiment place the upper range of the mean residence time to about 60 h or 2.5 days. Both figures are within the range of mean residence times for SO₂ cited in literature. Through the validation experiments of this work the rule of thumb, not to trust TSMs beyond the mean residence time of the substance, has become palpable. Nevertheless TSMs and related methods are computationally fast procedures, which deliver first hints on potential source areas, if applied within the frame of the mean residence time of the considered substance.     

Comparisons of statistical characteristic of air pollutants in Taiwan by frequency distribution

The lognormal, Weibull, and type V Pearson distributions were selected to fit the concentration frequency distributions of particulate matter with an aerodynamic diameter of < or = 10 microm (PM10) and SO2 in the Taiwan area. Air quality data from three stations, Hsin-Chu, Shalu, and Gain-Jin, were fitted with three distributions and compared with the measured data. The parameters of unimodal and bimodal fitted distributions were obtained by the methods of maximum likelihood and nonlinear least squares, respectively. Moreover, the root mean square error (RMSE), index of agreement (d), and Kolmogorov-Smirnov (K-S) test were used as criteria to judge the goodness-of-fit of these three distributions. These results show that the frequency distributions of PM10 concentration at the Hsin-Chu and Shalu stations are unimodal, but the distribution at Gain-Jin is bimodal. The distribution type of PM10 concentration varied greatly in different areas and could be influenced by local meteorological conditions. For SO2 concentration distribution, the distributions were all unimodal. The results also show that the lognormal distribution is the more appropriate to represent the PM10 distribution, while the Weibull and lognormal distributions are more suitable to represent the SO2 distribution. Moreover, the days exceeding the air quality standard (AQS) (PM10 > 125 microg/ m3) for the Hsin-Chu, Shalu, and Gain-Jin stations in the coming year are successfully predicted by the theoretic distributions.     

Prediction of ozone concentration in ambient air using multivariate methods

Multivariate statistical methods including pattern recognition (Principal Component Analysis--PCA) and modeling (Multiple Linear Regression--MLR, Partial Least Squares--PLS, as well as Principal Component Regression--PCR) methods were carried out to evaluate the state of ambient air in Miskolc (second largest city in Hungary). Samples were taken from near the ground at a place with an extremely heavy traffic. Although PCA is not able to determine the significance of variables, it can uncover their similarities and classify the cases. PCA revealed that it is worth to separate day and night data because different factors influence the ozone concentrations during all day. Ozone concentration was modeled by MLR and PCR with the same efficiency if the conditions of meteorological parameters were not changed (i.e. morning and afternoon). Without night data, PCR and PLS suggest that the main process is not a photochemical but a chemical one.     

Relationship among parameters of lake polluted sediments evaluated by multivariate statistical analysis

In the attempt to assess the relationship and interdependency among sediment toxic pollutants, in particular heavy metals, polycyclic aromatic hydrocarbons (PAH), and linear alkyl sulfonates (LAS) and some of the sediment typical components: inorganic carbon (IC), organic material (OM) and acid volatile sulphides (AVS), multivariate techniques of statistical analysis have been applied to a set of chemical data obtained by the analysis of the sediments of the Trasimeno Lake, a central Italy lake characterized by a large surface (128 km(2)) and a low mean depth (about 4.5 m). The results of principal component analysis (PCA) show interrelationships between: OM content and PAH, Pb, and Cu concentrations of the sediments, LAS and AVS, and AVS and IC. The effect of the different sampling periods on sediment composition and contamination level, and the clustering of the sampling sites as a consequence of pollutant load are also shown. The principal component bi-plot of the variables and samples indicates that PAH have the greatest influence on the separation of samples in the different sampling periods.     

Tools based on multivariate statistical analysis for classification of soil and groundwater in Apulian agricultural sites

In this paper, the results obtained from multivariate statistical techniques such as PCA (Principal component analysis) and LDA (Linear discriminant analysis) applied to a wide soil data set are presented. The results have been compared with those obtained on a groundwater data set, whose samples were collected together with soil ones, within the project “Improvement of the Regional Agro-meteorological Monitoring Network (2004–2007)”. LDA, applied to soil data, has allowed to distinguish the geographical origin of the sample from either one of the two macroaeras: Bari and Foggia provinces vs Brindisi, Lecce e Taranto provinces, with a percentage of correct prediction in cross validation of 87%. In the case of the groundwater data set, the best classification was obtained when the samples were grouped into three macroareas: Foggia province, Bari province and Brindisi, Lecce and Taranto provinces, by reaching a percentage of correct predictions in cross validation of 84%. The obtained information can be very useful in supporting soil and water resource management, such as the reduction of water consumption and the reduction of energy and chemical (nutrients and pesticides) inputs in agriculture.     

Performance evaluation of ETP from pesticide manufacturing industry by using WWQI and multivariate statistical analysis

Environmental Science and Pollution Research - The appropriate operation of wastewater treatment plants is essential to maintain the quality of treated water. The aim of the present study is to...     

Groundwater quality in Imphal West district, Manipur, India, with multivariate statistical analysis of data

The aim of this paper was to analyze the groundwater quality of Imphal West district, Manipur, India, and assess its suitability for drinking, domestic, and agricultural use. Eighteen physico-chemical variables were analyzed in groundwater from 30 different hand-operated tube wells in urban, suburban, and rural areas in two seasons. The data were subjected to uni-, bi-, and multivariate statistical analysis, the latter comprising cluster analysis (CA), principal component analysis (PCA), and factor analysis (FA). Arsenic concentrations exceed the Indian standard in 23.3 % and the WHO limit in 73.3 % of the groundwater sources with only 26.7 % in the acceptable range. Several variables like iron, chloride, sodium, sulfate, total dissolved solids, and turbidity are also beyond their desirable limits for drinking water in a number of sites. Sodium concentrations and sodium absorption ratio (SAR) are both high to render the water from the majority of the sources unsuitable for agricultural use. Multivariate statistical techniques, especially varimax rotation of PCA data helped to bring to focus the hidden yet important variables and understand their roles in influencing groundwater quality. Widespread arsenic contamination and high sodium concentration of groundwater pose formidable constraints towards its exploitation for drinking and other domestic and agricultural use in the study area, although urban anthropogenic impacts are not yet pronounced.     

An integrated multivariate statistical approach for the evaluation of spatial variations in groundwater quality near an unlined landfill

Environmental Science and Pollution Research - Groundwater is a major resource for water supply in Canada, and 43 of 68 Saskatchewan municipalities rely on groundwater or combined groundwater and...     

Assessment of metal pollution based on multivariate statistical modeling of 'hot spot' sediments from the Black Sea

The paper deals with application of different statistical methods like cluster and principal components analysis (PCA), partial least squares (PLSs) modeling. These approaches are an efficient tool in achieving better understanding about the contamination of two gulf regions in Black Sea. As objects of the study, a collection of marine sediment samples from Varna and Bourgas "hot spots" gulf areas are used. In the present case the use of cluster and PCA make it possible to separate three zones of the marine environment with different levels of pollution by interpretation of the sediment analysis (Bourgas gulf, Varna gulf and lake buffer zone). Further, the extraction of four latent factors offers a specific interpretation of the possible pollution sources and separates natural from anthropogenic factors, the latter originating from contamination by chemical, oil refinery and steel-work enterprises. Finally, the PLSs modeling gives a better opportunity in predicting contaminant concentration on tracer (or tracers) element as compared to the one-dimensional approach of the baseline models. The results of the study are important not only in local aspect as they allow quick response in finding solutions and decision making but also in broader sense as a useful environmetrical methodology.     

Establishment of a Box-Jenkins multivariate time-series model to simulate ground-level peak daily one-hour ozone concentrations at Ta-Liao in Taiwan

Box-Jenkins univariate autoregressive integrated moving average (ARIMA) and regression with time-series error (RTSE) models were established to simulate historical peak daily 1-hr ozone concentrations at Ta-Liao, Taiwan, 1997-2001. During 1995-2003, the 600 days of Pollution Standard Index (PSI) more than 100 (peak daily 1-hr ozone concentrations detected by greater than 120 ppm) at Tao-Liao showed the highest ozone exceedances among the six monitoring stations in Kaohsiung County. To improve the predictability of extremely high ozone, two different principal components, PC1 and PC(1 + 2), were introduced in the RTSE model. Four typical predictors (particular matter with an aerodynamic diameter less than or equal to 10 microm, temperature, wind speed, and wind direction) plus a PC trigger remained significant in the RTSE model. The model performance statistics concluded that the RTSE model with PC1 was optimal, compared with the univariate ARIMA, the RTSE model without PC, and RTSE model with PC(1 + 2). The contingency table shows that the successful predictions of the univariate model were only 12.9% of that of the RTSE model with PC1. Also, the POD value was improved approximately 5-fold when the univariate model was replaced by the RTSE model, and almost 8-fold when it was replaced by the RTSE model with PC1. Moreover, introducing the PC trigger indeed enhanced the ozone predictability. After the PC trigger was introduced in the RTSE model, the POD was increased 69.9%, and the FAR was reduced 8.3%. The overall correlation between the observed and simulated ozone was improved 9.6%. Also, the first principal component was more useful than the first two components in playing the "trigger" role, though it counted only for 58.62% of the environmental variance during the high ozone days.     

A review of statistical methods for the meteorological adjustment of tropospheric ozone

A variety of statistical methods for meteorological adjustment of ozone have been proposed in the literature over the last decade for purposes of forecasting, estimating ozone time trends, or investigating underlying mechanisms from an empirical perspective. The methods can be broadly classified into regression, extreme value, and space–time methods. We present a critical review of these methods, beginning with a summary of what meteorological and ozone monitoring data have been considered and how they have been used for statistical analysis. We give particular attention to the question of trend estimation, and compare selected methods in an application to ozone time series from the Chicago area. We conclude that a number of approaches make useful contributions to the field, but that no one method is most appropriate for all purposes and all meteorological scenarios. Methodological issues such as the need for regional-scale analysis, the nonlinear dependence of ozone on meteorology, and extreme value analysis for trends are addressed. A comprehensive and reliable methodology for space–time extreme value analysis is attractive but lacking.     

Models and statistical methods for gaseous emission testing of finite sources in well-mixed chambers

Two families of mathematical models are proposed to represent either the concentration of a gaseous emission in or the accumulated amount exiting from a well-mixed, environmentally controlled test chamber. A thin film model, which seems applicable to such sources as carpet adhesive, etc., has the capability of isolating the true emission rate constant from chamber effects. It has successfully modeled emissions of methyl ethyl ketone, a C₈ alcohol, and butyl propionate from latex caulk. Chamber effects in the form of temporary wall retention were identified for the latter two compounds. An analogous, deep source, diffusionlimited model for plywood, etc., once fitted to a data set, can be used to generalize to other combinations of source surface area, chamber volume, and air exchange rate.     

Regulating colored textile wastewater by 3/31 wavelength ADMI methods in Taiwan

The wastewater from textile dyeing facilities is difficult to treat satisfactorily because of high compositional variability and high color intensity. To reduce colored effluents discharged into watercourses, the government of Taiwan adopted the Effluent True Color Standard in 1998. The true color discharge limit is 400 American Dye Manufactures Institute (ADMI) units. The adopted analytical method is the ADMI Tristimulus Filter Method (3 wavelength (WL) method), and the 31 WL ADMI method might be also adopted as an alternative for color value measurement. The refractory nature of textile dyes and the introduction of this new regulation present an environmental challenge to the Taiwanese textile industry. The main objectives of this study were to (1) evaluate the efficacy of current wastewater treatment systems for controlling the colored textile wastewater discharges, and (2) evaluate the correlations between 3 and 31 WL ADMI methods. Ten representative textile wastewater treatment facilities employing biological and chemical coagulation treatment technologies were selected to perform a 10-consecutive-day effluent sampling and analysis. Results show that a significant difference between 3 and 31 ADMI methods was observed. These two ADMI methods cannot be substituted for each other, and the discharge standard should be determined based on the selected testing method. Investigation results also suggest that the commonly used wastewater treatment technology (biological + chemical coagulation) fails to effectively remove dye from the colored textile wastewater. Sodium hypochlorite (NaOCl) addition was applied by most facilities as the temporary post-polishment step to comply with the color discharge standard.     

Chernoff-Flury faces: a statistical means for representing multivariate response parameters of air pollution induced injury on plants

To evaluate the effects of dry and wet deposition on forest trees (Picea abies [L.] Karst.), the LIS-Essen is operating an Open-Top Chamber Field Station within an area where novel forest decline has been prevalent since 1982. Chambers are ventilated with either ambient or charcoal-filtered air and receive either natural or artificial rain, the latter being prepared by natural rain and distilled water in ratio 1:10. Besides deposition data, acquired above and below the tree crowns as well as via lysimeters of soil percolates, various parameters describing vitality of trees are measured. To obtain a persuading representation of total parameters and their interdependencies, a multivariate graphical cluster analysis has been performed by use of Chernoff-Flury faces. Interdependencies of vitality parameters are more easily recognizable in this multivariate picture than in usually applied binary correlation diagrams.     

Characterizing and locating air pollution sources in a complex industrial district using optical remote sensing technology and multivariate statistical modeling

Emissions of volatile organic compounds (VOCs) are most frequent environmental nuisance complaints in urban areas, especially where industrial districts are nearby. Unfortunately, identifying the responsible emission sources of VOCs is essentially a difficult task. In this study, we proposed a dynamic approach to gradually confine the location of potential VOC emission sources in an industrial complex, by combining multi-path open-path Fourier transform infrared spectrometry (OP-FTIR) measurement and the statistical method of principal component analysis (PCA). Close-cell FTIR was further used to verify the VOC emission source by measuring emitted VOCs from selected exhaust stacks at factories in the confined areas. Multiple open-path monitoring lines were deployed during a 3-month monitoring campaign in a complex industrial district. The emission patterns were identified and locations of emissions were confined by the wind data collected simultaneously. N,N-Dimethyl formamide (DMF), 2-butanone, toluene, and ethyl acetate with mean concentrations of 80.0?±?1.8, 34.5?±?0.8, 103.7?±?2.8, and 26.6?±?0.7 ppbv, respectively, were identified as the major VOC mixture at all times of the day around the receptor site. As the toxic air pollutant, the concentrations of DMF in air samples were found exceeding the ambient standard despite the path-average effect of OP-FTIR upon concentration levels. The PCA data identified three major emission sources, including PU coating, chemical packaging, and lithographic printing industries. Applying instrumental measurement and statistical modeling, this study has established a systematic approach for locating emission sources. Statistical modeling (PCA) plays an important role in reducing dimensionality of a large measured dataset and identifying underlying emission sources. Instrumental measurement, however, helps verify the outcomes of the statistical modeling. The field study has demonstrated the feasibility of using multi-path OP-FTIR measurement. The wind data incorporating with the statistical modeling (PCA) may successfully identify the major emission source in a complex industrial district.     

Chemistry of surface water at a volcanic summit area, Norikura, central Japan: multivariate statistical approach.

Many hydrochemical studies on chemical formation of shallow ground water have been reported as results of water-rock interaction, and contamination of paleo-brine or human activities, whereas the preliminary formation of precipitation source in the recharged region has not been established yet. The purpose of this research work is to clarify the geochemical process of water formation from a water source unpolluted by seawater or human activity. Norikura volcano, located in western part of central Japan provided a suitable source for this research purpose, and hence chemical compositions of water samples from the summit and the mountainside area of Norikura volcano were determined. Most samples in the summit area showed very low electrical conductivity, and lower than 12 microS/cm. On the basis of the chemical compositions, principal component analysis (PCA) and factor analysis (FA), such as kinds of multivariate statistical techniques were used to extract geochemical factors affecting hydrochemical process. As a result, three factors were extracted. The first factor showed high loading on K+, Ca2+, SO2 and SiO2, and this factor was interpreted due to influence of the chemical interaction between acidic precipitated water and rocks. The second factor showed high loading on Na+ and Cl-, and it was assumed to be an influence of seawater salt. The third factor showed loading on NO3-, and it was interpreted to be caused by biochemical effect of vegetation. The proportionate contributions of these factors to the evolution of water chemical composition were found to be 45%, 20%, and 10% for factors 1, 2 and 3, respectively. The same exploration at the mountainside of Norikura volcano revealed that the chemical variances of the non-geothermal water samples were highly influenced by water-rock interactions. The silicate dissolution showed 45% contribution for all chemical variances, while the adsorption of Ca2+ and Mg2+ by precipitation or ion exchange showed 20% contribution. The seawater salt influence or biochemical effect was statistically negligible in this area. The clear differentiation of geochemical process on water formation was found between the summit area and the mountainside area.     

Assessment and rationalization of water quality monitoring network: a multivariate statistical approach to the Kabbini River (India)

The establishment of an efficient surface water quality monitoring (WQM) network is a critical component in the assessment, restoration and protection of river water quality. A periodic evaluation of monitoring network is mandatory to ensure effective data collection and possible redesigning of existing network in a river catchment. In this study, the efficacy and appropriateness of existing water quality monitoring network in the Kabbini River basin of Kerala, India is presented. Significant multivariate statistical techniques like principal component analysis (PCA) and principal factor analysis (PFA) have been employed to evaluate the efficiency of the surface water quality monitoring network with monitoring stations as the evaluated variables for the interpretation of complex data matrix of the river basin. The main objective is to identify significant monitoring stations that must essentially be included in assessing annual and seasonal variations of river water quality. Moreover, the significance of seasonal redesign of the monitoring network was also investigated to capture valuable information on water quality from the network. Results identified few monitoring stations as insignificant in explaining the annual variance of the dataset. Moreover, the seasonal redesign of the monitoring network through a multivariate statistical framework was found to capture valuable information from the system, thus making the network more efficient. Cluster analysis (CA) classified the sampling sites into different groups based on similarity in water quality characteristics. The PCA/PFA identified significant latent factors standing for different pollution sources such as organic pollution, industrial pollution, diffuse pollution and faecal contamination. Thus, the present study illustrates that various multivariate statistical techniques can be effectively employed in sustainable management of water resources. Highlights ? The effectiveness of existing river water quality monitoring network is assessed ? Significance of seasonal redesign of the monitoring network is demonstrated ? Rationalization of water quality parameters is performed in a statistical framework     

Evaluation of aerosol sources at European high altitude background sites with trajectory statistical methods

This study has investigated the influence of synoptic weather patterns and long-range transport episodes on the concentrations of several compounds related to different aerosol sources (EC, OC, SO₄^2?, Ca²⁺, Na⁺, K⁺, ²¹⁰Pb, levoglucosan and dicarboxylic acids) registered in PM10 or PM2.5 aerosol samples collected at three remote background sites in central Europe. Air mass back-trajectories arriving at these sites have been analysed by statistical methods. Firstly, air mass back-trajectories have been grouped into clusters. Each cluster corresponds to specific meteorological scenarios, which were extracted and discussed. Finally, redistributed concentration fields have been computed to identify the main potential source regions of the different key aerosol components. A marked seasonal pattern is observed in the occurrence of the different clusters, with fast westerly and northerly Atlantic flows during winter and weak circulation flows in summer. Spring and fall were characterised by advection of moderate flows from northeastern and eastern Europe. Significant inter-cluster differences were observed for concentrations of receptor aerosol components, with the highest concentrations of EC, OC, SO₄^2?, K⁺ and ²¹⁰Pb associated with local and mesoscale aerosol sources located over central Europe related to enhanced photochemical processes. Emissions produced by fossil fuel and biomass burning processes from the Baltic countries, Byelorussia, western regions of Russia and Kazakhstan in spring and fall also contribute to elevated levels of EC, OC, SO₄^2?, K⁺ and ²¹⁰Pb. In the summer period long-range transport episodes of mineral dust from North-African deserts were also frequently detected, which caused elevated concentrations of coarse Ca²⁺ at sites. The baseline aerosol concentrations in central Europe at the high altitude background sites were registered in winter, with the exception of coarse Na⁺. While the relatively high concentrations of Na⁺ can be explained by sea salt advected from the Atlantic, the low levels of other aerosol components are caused by efficient aerosol scavenging associated to advections of Atlantic air masses, as well as lower emissions of these species over the Atlantic compared to those over the European continent and very limited vertical air mass exchange over the continent.     

Monitoring birds of prey in Finland: a summary of methods, trends, and statistical power

In Finland, Comprehensive Surveys to monitor numbers and productivity of four endangered species of birds of prey were started in the early 1970s. In 1982, the Ringing Center launched the Raptor Grid, a nationwide monitoring program for all other bird-of-prey species based on 10 x 10 km study plots of the Finnish National Grid. The annual total of study plots surveyed by voluntary raptor ringers has averaged 120. Since 1986, additional information on breeding performance has been collected using the Raptor Questionnaire. In 2006, more than 44 262 potential nest sites of birds of prey were inspected, and 12 963 occupied territories, including 8149 active nests, were found and reported by ringers. The population trend during 1982-2006 has been significantly negative in six species and positive or neutral in 18 species. Statistical power of the time series of numbers and productivity has been adequate for all species except the microtine specialists.