Georgios Grivas

Neural Network and Multiple Regression Models for PM10 Prediction in Athens: A Comparative Assessment

Abstract Particulate atmospheric pollution in urban areas is considered to have significant impact on human health. Therefore, the ability to make accurate predictions of particulate ambient concentrations is important to improve public awareness and air quality management. This study examines the possibility of using neural network methods as tools for daily average particulate matter with aerodynamic diameter <10 µm (PM10) concentration forecasting, providing an alternative to statistical models widely used up to this day. Based on a data inventory, in a fixed central site in Athens, Greece, ranging over a two-year period, and using mainly meteorological variables as inputs, neural network models and multiple linear regression models were developed and evaluated. Comparison statistics used indicate that the neural network approach has an edge over regression models, expressed both in terms of prediction error (root mean square error values lower by 8.2–9.4%) and of episodic prediction ability (false alarm rate values lower by 7–13%). The results demonstrate that artificial neural networks (ANNs), if properly trained and formed, can provide adequate solutions to particulate pollution prognostic demands.

Spatial and Temporal Variation of PM10 Mass Concentrations within the Greater Area of Athens, Greece

This study presents the results of a yearlong PM10 measurement campaign, conducted at four sampling sites, within the Greater Athens Area, between June 2001 and May 2002. Daily average PM10 concentrations were determined using reference samplers installed in two urban locations (Maroussi and Aristotelous), a mixed urban–industrial location (Elefsina) and a background location (Thrakomacedones). The 24-h and annual PM10 average concentrations were calculated and compared to the corresponding air quality standards, promulgated by the European Union, revealing severe exceedances of the limit values. The mean concentrations for the sampling periods were 73.8, 83.2, 32.9 and 54.9 μg/m3 for Maroussi (MAR), Aristotelous (ARI), Thrakomacedones (THR) and Elefsina (ELE), respectively. The spatial variation of PM10, was investigated resulting at a coefficient of variation of 0.36 within the study area and correlation coefficients ranging from 0.57 to 0.84. In addition, strong associations between PM10 and other primary gaseous pollutants were found. Regression analysis of PM10 against NOx (used as an indicator for road traffic emissions) revealed significant vehicular contributions to the measured PM10 concentrations. Higher PM10 levels were recorded during prevailing winds of the S–SW sectors, while lower levels were observed during strong northerly flows. The overall results allowed a first assessment of the severity of PM10 air pollution in the Athens basin, and of the potential sources responsible for it. The control of traffic-related particle emissions appears to be a principal objective, for the confrontation of the PM10 pollution problem affecting the area.

Elemental and organic carbon in the urban environment of Athens. Seasonal and diurnal variations and estimates of secondary organic carbon.

Elemental and organic carbon (EC and OC) hourly concentrations were measured continuously, at an urban location in central Athens, Greece, for an 8-month period (January-August). Average concentrations of 2.2 μgC m(-3) and 6.8 μgC m(-3) were observed, for EC and OC, respectively. The combined contribution of carbonaceous compounds (EC plus organic matter) to PM(10) was calculated at 26%. The seasonal variability of EC was limited, while OC mean concentrations were significantly higher (by 23%), during the warm months (May-August). The weekly variation followed a different pattern, with the weekend decrease of EC levels (25%) being more pronounced than of OC (14%). EC produced a bimodal diurnal cycle, with the morning rush hour traffic mode prevailing. The OC mean circadian variation displayed those peaks as well. However, midday-to-afternoon presence of secondary organic aerosol (SOA) was strongly indicated. The conditional probability function was used to assess the impact of wind direction. High EC, OC levels were linked to southern flows, which during summer are mainly related to the appearance of sea breeze circulation. The temporal variation of EC, OC and their correlation patterns with primary and secondary gaseous pollutants, suggested that, although primary emissions affected both fractions, SOA formation is an important factor to be accounted for, especially during the photochemical season. Secondary organic carbon was estimated using the EC tracer method and orthogonal regression on OC, EC hourly concentration data. The average contributions of secondary organic carbon (SOC) to OC were calculated at 20.9% for the cold period and 30.3% for the warm period. Maximum values of 58% and 91% were estimated for daily and hourly contributions, respectively. The SOC diurnal variations suggested photochemical formation throughout the year, intensified during summer months, with the correlation coefficient between SOC and the sum of oxidants (NO(2+)O(3)) reaching up to 0.84.

Performance of Multi-City Land Use Regression Models for Nitrogen Dioxide and Fine Particles

Background: Land use regression (LUR) models have been developed mostly to explain intraurban variations in air pollution based on often small local monitoring campaigns. Transferability of LUR models from city to city has been investigated, but little is known about the performance of models based on large numbers of monitoring sites covering a large area. Objectives: We aimed to develop European and regional LUR models and to examine their transferability to areas not used for model development. Methods: We evaluated LUR models for nitrogen dioxide (NO2) and particulate matter (PM; PM2.5, PM2.5 absorbance) by combining standardized measurement data from 17 (PM) and 23 (NO2) ESCAPE (European Study of Cohorts for Air Pollution Effects) study areas across 14 European countries for PM and NO2. Models were evaluated with cross-validation (CV) and hold-out validation (HV). We investigated the transferability of the models by successively excluding each study area from model building. Results: The European model explained 56% of the concentration variability across all sites for NO2, 86% for PM2.5, and 70% for PM2.5 absorbance. The HV R2s were only slightly lower than the model R2 (NO2, 54%; PM2.5, 80%; PM2.5 absorbance, 70%). The European NO2, PM2.5, and PM2.5 absorbance models explained a median of 59%, 48%, and 70% of within-area variability in individual areas. The transferred models predicted a modest-to-large fraction of variability in areas that were excluded from model building (median R2: NO2, 59%; PM2.5, 42%; PM2.5 absorbance, 67%). Conclusions: Using a large data set from 23 European study areas, we were able to develop LUR models for NO2 and PM metrics that predicted measurements made at independent sites and areas reasonably well. This finding is useful for assessing exposure in health studies conducted in areas where no measurements were conducted. Citation: Wang M, Beelen R, Bellander T, Birk M, Cesaroni G, Cirach M, Cyrys J, de Hoogh K, Declercq C, Dimakopoulou K, Eeftens M, Eriksen KT, Forastiere F, Galassi C, Grivas G, Heinrich J, Hoffmann B, Ineichen A, Korek M, Lanki T, Lindley S, Modig L, Mölter A, Nafstad P, Nieuwenhuijsen MJ, Nystad W, Olsson D, Raaschou-Nielsen O, Ragettli M, Ranzi A, Stempfelet M, Sugiri D, Tsai MY, Udvardy O, Varró MJ, Vienneau D, Weinmayr G, Wolf K, Yli-Tuomi T, Hoek G, Brunekreef B. 2014. Performance of multi-city land use regression models for nitrogen dioxide and fine particles. Environ Health Perspect 122:843–849; http://dx.doi.org/10.1289/ehp.1307271

Spatial variation of PM elemental composition between and within 20 European study areas - Results of the ESCAPE project

An increasing number of epidemiological studies suggest that adverse health effects of air pollution may be related to particulate matter (PM) composition, particularly trace metals. However, we lack comprehensive data on the spatial distribution of these elements. We measured PM2.5 and PM10 in twenty study areas across Europe in three seasonal two-week periods over a year using Harvard impactors and standardized protocols. In each area, we selected street (ST), urban (UB) and regional background (RB) sites (totaling 20) to characterize local spatial variability. Elemental composition was determined by energy-dispersive X-ray fluorescence analysis of all PM2.5 and PM10 filters. We selected a priori eight (Cu, Fe, K, Ni, S, Si, V, Zn) well-detected elements of health interest, which also roughly represented different sources including traffic, industry, ports, and wood burning. PM elemental composition varied greatly across Europe, indicating different regional influences. Average street to urban background ratios ranged from 0.90 (V) to 1.60 (Cu) for PM2.5 and from 0.93 (V) to 2.28 (Cu) for PM10. Our selected PM elements were variably correlated with the main pollutants (PM2.5, PM10, PM2.5 absorbance, NO2 and NOx) across Europe: in general, Cu and Fe in all size fractions were highly correlated (Pearson correlations above 0.75); Si and Zn in the coarse fractions were modestly correlated (between 0.5 and 0.75); and the remaining elements in the various size fractions had lower correlations (around 0.5 or below). This variability in correlation demonstrated the distinctly different spatial distributions of most of the elements. Variability of PM10_Cu and Fe was mostly due to within-study area differences (67% and 64% of overall variance, respectively) versus between-study area and exceeded that of most other traffic-related pollutants, including NO2 and soot, signaling the importance of non-tailpipe (e.g., brake wear) emissions in PM.

Is daily exposure to ozone associated with respiratory morbidity and lung function in a representative sample of schoolchildren? Results from a panel study in Greece

Previous time series or panel studies of asthmatics have reported respiratory health effects following short-term exposure to ozone (O3). We followed 186 children aged 10 years old in Athens and Thessaloniki, Greece for 5 weeks during the academic year 2013–2014 and recorded daily their respiratory symptoms, absenteeism and peak expiratory flow (PEF). We applied mixed models controlling for various possible confounders to investigate the daily associations between O3 exposure — derived from weekly personal and fixed school site measurements calibrated using daily values of the fixed monitoring site nearest to the child’s school location — and PEF, presence of any symptom, cough and stuffy nose, as well as absenteeism. We tested the robustness of our findings to varying modeling assumptions and confounders and investigated effect modification patterns by medication use, time spent outdoors and prevalence of asthma. A 10 μg/m3 increase in O3 personal exposure was associated with increased odds of any symptom (odds ratio (OR): 1.19, 95% confidence interval (CI): 0.98, 1.44), largely attributed to the increase in the odds of stuffy nose (OR: 1.23, 95% CI: 1.00, 1.51). PEF and absenteeism were not related to O3 exposure. Our results were robust to several sensitivity analyses. Effects were modified by medication use as presence of symptoms but also decreases in PEF were mainly reported among non-users, while our effect estimates were not driven by the asthmatic subgroup of children. Our findings indicate that short-term O3 exposure may be associated with respiratory symptoms extending previously reported results for asthmatics to the general population.

Determinants of personal exposure to ozone in school children. Results from a panel study in Greece

Background: In the wider framework of the RESPOZE (ReSPiratory effects of OZone Exposure in Greek children) panel study, we investigated possible determinants of O3 exposure of school children, measured with personal passive samplers, in Athens and Thessaloniki, Greece. Methods: Personal exposure to O3 was measured for five weeks spread along the academic year 2013‐14, in 186 school children in Athens and Thessaloniki, Greece. At the same time, at‐school outdoor measurements were performed and ambient levels of 8‐h daily maximum O3 from fixed sites were collected. We also collected information on lifestyle and housing characteristics through an extended general questionnaire (GQ) and each participant completed daily time activity diaries (TADs) during the study period. Results: Mean outdoor concentrations were higher during the warmer months, in the suburbs of the cities and in Athens. Personal exposure concentrations were significantly lower compared to outdoor. Daily levels of at‐school outdoor and ambient levels of O3 from fixed sites were significant determinants of personal exposure to O3. For a 10 &mgr;g/m3 increase in at‐school outdoor O3 concentrations and PM10 measurements a 20.9% (95% CI: 13%, 28%) increase in personal exposure to O3 was found. For a half an hour more spent in transportation an average increase of 7% (95% CI: 0.3%, 14.6%) in personal exposure to O3 was observed. Among other possible determinants, time spent in transportation (TAD variable) and duration of open windows were the ones associated with personal O3 exposure levels. Conclusions: Our results support the use of outdoor and ambient measurements from fixed sites in epidemiological studies as a proxy of personal exposure to O3, but this has to be calibrated taking into account personal measurements and time‐activity patterns. HighlightsPossible determinants of personal ozone exposure of school children were investigated.Methods rely on personal passive samplers and simultaneously at‐school measurements.Epidemiological studies may use ambient measurements as a proxy for personal exposure.

Ozone exposure assessment for children in Greece - Results from the RESPOZE study

Ozone exposure of 179 children in Athens and Thessaloniki, Greece was assessed during 2013-2014, by repeated weekly personal measurements, using passive samplers. O3 was also monitored at school locations of participants to characterize community-level ambient exposure. Average personal concentrations in the two cities (5.0 and 2.8ppb in Athens and Thessaloniki, respectively) were considerably lower than ambient concentrations (with mean personal/ambient ratios of 0.13-0.15). The temporal variation of personal concentrations followed the -typical for low-latitude areas- pattern of cold-warm seasons. However, differences were detected between temporal distributions of personal and ambient concentrations, since personal exposures were affected by additional factors which present seasonal variability, such as outdoor activity and house ventilation. Significant spatial contrasts were observed between urban and suburban areas, for personal concentrations in Athens, with higher exposure for children residing in the N-NE part of the area. In Thessaloniki, spatial variations in personal concentrations were less pronounced, echoing the spatial pattern of ambient concentrations, a result of complex local meteorology and the smaller geographical expansion of the study area. Ambient concentration was identified as the most important factor influencing personal exposures (correlation coefficients between 0.36 and 0.67). Associations appeared to be stronger with ambient concentrations measured at school locations of children, than to those reported by the nearest site of the air quality monitoring network, indicating the importance of community-representative outdoor monitoring for characterization of personal-ambient relationships. Time spent outdoors by children was limited (>90% of the time they remained indoors), but -due to the lack of indoor sources- it was found to exert significant influence on personal concentrations, affecting inter-subject and spatiotemporal variability. Additional parameters that were identified as relevant for the determination of personal concentrations were indoor ventilation conditions (specifically indoor times with windows open) and the use of wood-burning in open fireplaces for heating as an ozone sink.

Effect of Ambient Ozone Exposure Assessed by Individual Monitors on Nasal Function and Exhaled NO Among School Children in the Area of Thessaloniki, Greece

Objectives: The study of short-term effects of environmental ozone exposure on nasal airflow, lung function, and airway inflammation. Methods: Ninety one children—47 underwent rhinomanometry—were included. The study was carried out during the 2013 to 2014 academic year. Activity questionnaires and personal O3 samplers were distributed and 1 week later, respiratory measurements were performed. Daily measurements of outdoor ozone were also considered. Results: A 10 &mgr;g/m3 increase in weekly personal ozone exposure concentrations was associated with a non-statistically significant 12.7% decrease in nasal inspiratory airflow (29.4% during the high ozone period). When the outdoor exposure of the same and the previous day were taken into account the corresponding figures were 13.48% and 43.58% (P = 0.02). Conclusions: There is an indication for increased risk of nasal obstruction during exposure to high ozone.