Interactions between complicated flow-dispersion patterns and boundary layer evolution in a mountainous complex terrain during elevated SO₂ concentrations

Abstract

The dispersion of air pollutants from multiple industrial stacks located in complex topography is an interesting subject. An attractive case is that of the wider region of Western Macedonia in NW Greece, where the greater amount of electric power of Greece is being produced by lignite power plant stations (LPPS). Considerable amounts of atmospheric pollutants are emitted by those LPPS into the atmosphere due to the quantities of coal burned. The variability of the topographic features and the terrain complexity of the area may lead to the formation of local atmospheric circulations of various types, which affect pollutant’s transport and dispersion. In the present work, the dispersion conditions that favor the pollutants accumulation in the area are investigated. For this purpose, 1 year’s hourly SO₂ concentrations, surface wind measurements and a mesoscale meteorological and air pollution model (The Air Pollution Model, TAPM) were used. The SO₂ and wind measurements were collected in situ from monitoring stations located nearby and at a greater distance from the power plants. Yearly and daily variations of SO₂ concentrations are analyzed and discussed, and the period with the highest concentrations is selected. During this period, the evolution of the atmospheric boundary layer (ABL) in the area as well as the pollutants dispersion is examined. Statistical measures between modeled and observed meteorological data were in good agreement and a good correlation coefficient 0.68 and 0.98 was found in the SO₂ variations. The analysis of the wind fields indicated better ventilation in the center of the area due to topographic venturi effects, while the dispersion mechanism which resulted in the relatively high ground level concentrations was fumigation. Finally, the evolution of the ABL was affected by the complex interactions between topography and mesoscale flows as it was found by the turbulent kinetic energy cross sections.

1 Introduction

Air pollution from multiple industrial stacks is a well-known and recognized problem. Pollutants’ concentrations in areas related to industrial activity are not only changing from day to day, but also varying through daytime according to the nearby topographic characteristics. The latter cause marginally distorted surface flows, which may subsequently influence and change the dispersion conditions. As a matter of fact, the air pollution problem becomes more intense and interesting, when specific synoptic weather patterns in conjunction with topography’s complexity favor the development of local circulations and further augment pollutant concentrations (Triantafyllou et al. 1999; Deligiorgi et al. 2013). In such cases, critical role in the determination of the dispersion conditions have the topography, the valley’s geometry and the channeling of the synoptic/regional flow. Those are among the main reasons for the formation of local flow patterns, such as anabatic/katabatic flows, sea–lake effects and stagnant conditions (Kallos et al. 1993; Helmis et al. 1995, 1997; Triantafyllou 2001; Ding et al. 2004; Baumgardner et al. 2006). Nevertheless, topography is not the only reason which influences the fate of pollutants concentration in the atmosphere. An equal and perhaps more important parameter that affects the pollutants and also interacts with topography and land use is the evolution of the ABL. The boundary layer is that part of the atmosphere that responds directly to the flows of mass, energy and momentum from the earth’s surface, characteristically at timescales of an hour or less (Stull 1988). Most pollutants are emitted or chemically produced within this layer, while its diurnal evolution plays a critical role for the determination of the pollutants dispersion pathways (Salmond and McKendry 2005). In addition, the evolution of the ABL as reported in several studies is the main reason for the elevated ground level pollutant’s concentrations (Kraus and Ebel 1989; Zhang and Rao 1999; Triantafyllou and Kassomenos 2002; Matthaios et al. 2014).

An interesting case which combines large amounts of air pollutants’ emissions and complex topography is the Greater Western Macedonia Region (GWMR), an area in northwestern Greece. In this mountainous area, the greatest percentage of the total electrical energy production in Greece is produced by the local LPPS. Moreover, another LPPS also operates in FYROM, close to the borders of the investigated area, which is substantially affecting the air pollution levels in the region under particular meteorological conditions (Triantafyllou et al. 1999; Triantafyllou and Kassomenos 2002). These LPPSs use raw lignite as fuel that is mined in the nearby open-pit mines.

From the aforementioned heavy industrial activity, huge amounts of atmospheric pollutants are emitted in the area. In particular, the most significant emissions arising from the combustion of lignite and from mining of the lignite coal, as well as, from the transport of lignite ash, which cause local air pollution problems in the nearby areas (Triantafyllou 2003; Samara 2005), are mainly particles (fly ash and fugitive dust), sulfur dioxide (SO₂) and NOx (Triantafyllou and Kassomenos 2002). For the year 2001, the SO₂ and NOx emission of GWMR according to European Pollutant Release and Transfer Register (E-PRTR) are 81,600 and 43,700 t, respectively, while the PM emissions are mainly depended on the electrostatic precipitators efficiency (Triantafyllou et al. 2013). It should be mentioned, however, that aspects of distant secondary pollutants’ transfer as a result of the primary emissions from industrial sources were also found in the area (Aidaoui et al. 2015). Regarding the particulate matter (PM), this is a difficult pollutant to investigate, since its emissions and dispersion pathways come from different type of sources. PM includes not only emissions from the LPPS (point sources), where fly ash is emitted in the atmosphere, but also from the nearby open pit mines (area and line sources), where uncontrolled fugitive dust is generated by mining operating procedures (Chaulya et al. 2003; Chaulya 2005).

Despite the fact that the air pollution problem in the area is related only to PM concentrations which exceed the air quality standards (Triantafyllou 2001, 2003) the problem regarding the PM is more complicated, due to the large number of emission sources in the area (lignite power plants, open mines, quarries, agricultural activities, biomass burning and even Saharan dust). Thus, in this work SO₂ concentrations are analyzed and are actually used in the frame of the present study as a kind of “tracer gas” emitted from the stacks. This is because of the purpose of the work, which is not a study of the air pollution in a specific area, but the investigation of the dispersion mechanisms in complex terrain with variable characteristics. In spite of the apparent simplicity of the problem, it should be further taken into consideration that the LPPSs’ stacks were constructed with different heights, ranging from 115 to 200 m and they are located in different parts of the GWMR with different topographic characteristics. It is important to investigate these mechanisms because the work of Triantafyllou et al. (2013) indicated that the monthly LPPS’s relative contribution to the PM levels of the area, within the studied basin, ranged from 27 to 84 % when the background concentration was removed. Regarding the LPPS’s NOx emissions, in the work of Aidaoui et al. (2015), it is indicated that they are evident even in the distant urban places of the GWMR and they are mainly dependant on the wind regime effects of the basin.

Considering, therefore, SO₂ as a proxy of other more important pollutants, in this paper, the SO₂ concentrations for 1-year period (2011) from monitoring stations located nearby and at a greater distance from the industrial activity are analyzed and discussed. Particular focus on the paper is given in the period when the highest concentrations were recorded, which is analyzed separately as a worst case, combining the SO₂ observations with model simulations. Emphasis is also given in the atmospheric dispersion conditions and mechanisms that favor the accumulation of the pollutants in the lower levels of the atmosphere in a complex terrain which are investigated with the implementation of a mesoscale prognostic meteorological and air pollution model.

2 Site description and data used

The GWMR is a mountainous area with complex topography located in the northwestern part of Greece. The GWMR can be described as broad, relatively flat bottom basin, surrounded by high mountains and lakes (Fig. 1). The area is approximately 50 km in length and the width ranges from 25 km in the sides to 10 km in the center, while it has also a sloping geometry of 2:1000 m from southeast to northwest. In the eastern sides of the basin, there are located the high mountains of Voras (2525 m) and Vermio (2100 m), while in the western sides there are situated the high mountains of Askio (2111 m) and Burinos (1866 m). Moreover, south of the basin is located the lake of Polifitos which is about 25 km long and 3 km wide, while in the north and northeastern there are the lakes of Petron and Vegoritida. The climate is continental Mediterranean with high temperatures during summer of 29.3 °C (mean monthly maximum temperature) and low ones during winter −1.2 °C (mean monthly minimum temperature). The dominant winds are mostly low to moderate, blowing in the NW/SE axis due to channeling of the synoptic wind, since NW/SE axis matches the geographical location of the basin (Fig. 1). The major industrial activity of the area is from the four lignite power plants (total power 3860 MW) and their open pit mines, which operate in the area by the Greek Power Public Corporation (GPPC) (see Fig. 1 for the exact location), while the LPPS5 from the nearby area of FYROM has 675 MW and two stacks.

Fig. 1

Map of the Western Macedonia Greater Region showing the location of the two monitoring sites (green color) as well as the location of the five lignite power plants (PPS, red color)

Average hourly values of SO₂ concentration and meteorological parameters of wind speed and direction were collected in situ and analyzed for 1-year period (2011), from the monitoring network of the GPPC. For this work, we used average hourly values from the station of Pontokomi which is located near the industrial activity (see Fig. 1) as well as from the station of Petrana, which is located in the southern part at a greater distance from the industrial activity. The selection of the above two monitoring stations was made due to the high percentage of the completeness of the hourly values which was above 98 %. Valid were considered the days in which the hourly value percentage was equal or above 75 %. The measurements of SO₂ were made by automatic in situ online monitors (TELEDYNE) with their general principle to be UV fluorescence. The minimum range of the monitors is from 0 to 5 ppb (0–13 μg/m³) while the maximum is from 0 to 20,000 ppb (0–52,350 μg/m³). The measurement units are ppb, ppm, μg/m³, mg/m³ (selectable) according to the preference of the user, while the detection limit of the instrument is 50 ppt (0.13 μg/m³).

3 Modeling details and verification

3.1 Model description and configuration

The implemented model in this study is The Air Pollution Model (TAPM), which is a mesoscale prognostic meteorological and air pollution model. In particular, TAPM is a PC-based, 3D, coordinate terrain following, non-hydrostatic, nestable, prognostic meteorological and air pollution model that solves fundamental fluid dynamics and scalar transport equations to predict meteorology and pollutant concentrations for a range of air quality applications (Hurley 2008). Furthermore, it is consisted by a coupled inline approach whereby the meteorological component is used in the air pollution component. The coupled approach taken by the model, whereby mean meteorological and turbulence fields are passed to the air pollution module every 5 min, allows pollution modeling to be done more accurately. Additionally, TAPM’s nested approach for meteorology and air pollution allows the user to zoom into a local region of interest quite rapidly, with the pollution grids optionally being able to be configured for a sub-region and/or at finer grid spacing than the meteorological grid. The meteorological component of TAPM predicts the local-scale flow against a background of larger scale meteorology provided by input synoptic analyses (or forecasts). The boundary-layer height (Zmix) in convective conditions is defined through the strength of convective updraft. During daytime, the ABL top is reached at the first vertical level where convective updraft decreases to zero and during nighttime when the vertical heat flux has decreased to 5 % or less from the surface value (Derbyshire 1990). The air pollution component of TAPM uses the predicted meteorology and turbulence from the meteorological component, and consists of an Eulerian grid-based set of prognostic equations for pollutant concentration and an optional Lagrangian particle mode. The Lagrangian mode uses a part puff approach, whereby the released mass at a given time is represented by a circular puff in a horizontal plane and the vertical motion of this puff is defined via the Lagrangian approach (Hurley 1994). Gas- and aqueous-phase reactions of sulfur dioxide and particles are also included, with the aqueous-phase reactions based on Seinfeld and Pandis (1998). More information about the model’s equations and parameterizations including the numerical methods and a summary of some verification studies can be found in the study by Hurley et al. (2001, 2005). In the present work, TAPM was configured and performed for 7 days, during the period where the relatively high SO₂ concentrations were measured, from 28/11/11 until 4/12/11. The configuration included three nested domains both for meteorology and pollution. 31 × 31 horizontal points and spacing 30 km, 10 and 3 km for meteorology (Fig. 2) and 33 × 53 grid points for pollution with spacing 7.5, 2.5 and 0.75 km were used [note that pollution grids can be subsets of the meteorological grid at finer grid spacing (Hurley et al. 2001)]. To capture the synoptic circulation, 25 vertical levels were used; the lowest ten model’s levels were at heights of 10, 25, 50, 100, 150, 200, 250, 300, 400 and 500 m, with the model top at 8 km. The input synoptic fields of the horizontal wind components, temperature and moisture were derived from the US NCEP (National Centers for Environmental Prediction) reanalysis, with a resolution of 2.5° longitude × 2.5° latitude on 17 pressure levels, every 6 h (Luhar and Hurley 2012). For the topography, global terrain height data on a longitude/latitude grid at 30-s grid spacing (approximately 1 km) based on public domain data available from the US Geological Survey; Earth Resources Observation Systems (EROS) was used. Soil characterization of the area was derived from the global soil texture types on a longitude/latitude grid at 2° grid spacing (approximately 4 km) based on FAO/UNESCO soil classes dataset. As pollution sources, eleven stacks of the aforementioned LPPSs from the GPPC were selected and simulated in a chemistry mode including the pollutants of PM₁₀, PM_2.5, NOx, NO₂, O₃ and SO₂. Their characteristics are presented in Table 1. The available emission rates in our study were average hourly values, which were calculated according to the US EPA AP-42, after the conversion of the SO₂ measurements made in stacks (mg/Nm3dry 6 % O₂) to g/s. Those measurements were made by the GPPC and they are reported for each power plant factory in the annual reports of the Greek Public Power Corporation 2011. The model has two options for the simulation of the emission rates: constant and cycled (up to 216 h). In this study, the second option was preferred since our study includes 168 h for the air pollution simulations and practically we used average hourly values for 168 h based on the measurements of the GPPC without completing a full cycle. The model run in Lagrangian mode to capture the near LPPSs source dispersion more accurately during the first 900 s and then changed to Eulerian mode, where Lagrangian particles were converted to grid concentration and from then on represented by the Eulerian transport equation.

Fig. 2

Simulation domains in TAPM. a Outer domain 30 km, middle domain 10 km, inner domain 3 km; b inner domain of the simulation depicting the topography of the area as well as the two monitoring sites (PNT, PETR). Elevations are in meters. The box 1–4 indicates the dispersion domain. The lines AB, CD indicate the positions of the vertical cross sections which are depicted in Fig. 12

Table 1 Characteristics of the stacks included for the model simulations (annual reports from the Greek Power Public Corporation 2011)

3.2 Verification indices

The model’s adequacy for regulatory applications is examined by the evaluation procedure and must first be well defined because it could lead to different variables to evaluate and different performance measures to use (Chang and Hanna 2004; Hanna and Chang 2012). To validate the results obtained by the model TAPM, hourly averaged meteorological values of wind speed, wind direction and temperature were used from both monitoring sites (nearby and at a greater distance from the industrial activity). In particular, the statistical evaluation of the model during the simulated period is made according to Willmott (1981) and Pielke (1984), while the statistical measures applied here, were Pearson’s correlation coefficient (Cor) and index of agreement (IOA) defined as: IOA = $1 - \left[ {{{\overline{{(P - O)^{2} }} } \mathord{\left/ {\vphantom {{\overline{{(P - O)^{2} }} } {\overline{{(|P - \bar{O}| + |O - \bar{O}|)^{2} }} }}} \right. \kern-0pt} {\overline{{(|P - \bar{O}| + |O - \bar{O}|)^{2} }} }}} \right]$ and root mean square error (RMSE) defined as: RMSE = $\left[ {\overline{{(P - O)^{2} }} } \right]^{1/2}$. The IOA is a measure of how well the predicted variations around the mean observations are represented and ranges from 0 to 1, with a value higher than 0.5 to represent a good model’s prediction (Hurley et al. 2005), while RMSE values were considered good when they are approaching zero.

4 Results and discussion

4.1 SO₂ observation analyses

The diurnal variation of SO₂ concentrations during the year of 2011 from the two monitoring sites are presented in Fig. 3. The concentrations given in these figures resulted from the average of all concentration values of every hour during the studied period. According to these figures, the diurnal variation of the SO₂ concentrations shows a typical maximum during noon hours in both monitoring stations. In particular, a characteristic sharp peak appears during the noon hours, of about 1–2 h duration, in the station near the industrial stacks (Pontokomi). The ratio highest hourly concentration/lowest hourly concentration (hhc/lhc) for the station is equal to 2.5 (Fig. 3a). On the contrary, a rather different variation in the diurnal concentrations pattern is presented in the distant station (Petrana). A smoother variation appears, while the ration of hhc/lhc for the station is 1.5 (Fig. 3b). Additionally, in this distant monitoring station, the smoother variation also illustrates a wider maximum, which in fact presents a very interesting case as it is discussed later on.

Fig. 3

Diurnal variations of SO₂ concentrations for the stations of a Pontokomi and b Petrana for the year 2011

To investigate in detail the diurnal variation of SO₂ concentrations, knowledge of the evolution of the ABL is necessary. However, there are not enough systematic observations of the vertical structure of the atmosphere in GWMR. In a representative study made by Triantafyllou et al. (1995), it can be summarized that the inversion’s depths in the basin ranged from 150 to 250 m for summer and autumn, and from 250 to 650 m for winter, while the average inversion’s strength was 9 and 10.5°K, respectively. From the aim of air pollution, the ABL evolution is interesting enough to be studied, when the vertical structure of the atmosphere favors the accumulation of pollutants in the lower layers near the ground. Such case is attempted and discussed later on in this paper.

The mean daily concentration levels for the year of 2011 are presented in Fig. 4. The SO₂ concentration levels of the area are below the daily legislative limits of 125 μg/m³ (EC 1999) and varied between 7 and 53 μg/m³ in the monitoring site nearby the industrial activity (Pontokomi). For the monitoring site at a greater distance from the industrial activity (Petrana) the concentrations ranged from 2 to 11 μg/m³. Despite some observed SO₂ increment during January in both monitoring sites probably due to the Etna’s activity (http://earthobservatory.nasa.gov/IOTD/view.php?id=48612, http://www.volcanolive.com/etna2.html), the biggest noticeable simultaneous increment in the concentration levels to both monitoring stations is during November and December and more specifically from 28/11/11 to 4/12/11. Those specific days with enhanced SO₂ concentrations in both stations are analyzed later on as a worst case.

Fig. 4

Mean daily concentrations of SO₂ for the year of 2011 a for the station of Pontokomi, b for the station of Petrana

The most important weather features that affect the dispersion condition and consequently either increase or reduce the concentration levels are the wind speed, the wind direction, the air temperature, the atmospheric stability and the solar radiation. In this section, the role of wind is examined. Particularly, we examined the role of wind in conjunction with SO₂ concentrations to investigate any potential source influence. For that reason, pollution roses were constructed for the entire year (2011) from hourly average data (Fig. 5). From the pollution roses it is evident for the monitoring station of Pontokomi (Fig. 5a) that the greater concentration values are correlated to the N or NE sector winds, where the main industrial activities take place. In contrast, Petrana monitoring station presents concentrations correlated to N and S sector winds (Fig. 5b). This case is very interesting because on the one hand the station’s concentrations point out the main industrial activity from the N, but, on the other hand, the station presents almost equal concentration levels from the opposite direction (south). The southerly concentrations can be attributed to the local thermal south circulation, which was detected at the Aliakmon valley coming from the artificial lake of Polifitos (see Fig. 1). This local thermal circulation develops during late noon and early afternoon hours and it is observable during both cold and warm periods of the year (Triantafyllou 2001). Additionally, it may contribute to the formation of high concentrations due to the recirculation of pollutants from the lake of Polifitos, where the pollutants were transported southerly during the day or during the previous day. This fact also explains the wider concentration peak in the diurnal variation for the station in Fig. 3b.

Fig. 5

Pollution roses a from the station of Pontokomi, b from the station of Petranafor the year of 2011. Down to the right is the percentage of calms and the concentration

4.2 Worst case simulations

The highest concentrations were recorded in both monitoring sites during the period of 28/11/11–4/12/11. Regarding the synoptic conditions of this high-concentration period, a high-pressure system generated over the central Europe moved toward the east on 28th of November. This anticyclone practically converted into stationary above Balkans during the remaining days (Fig. 6) until 4th December and covered most of the Balkan Peninsula and Greece area. Such synoptic circulation is a common weather pattern during winter period in Greece (Kallos et al. 1993; Triantafyllou 2001) and usually lasts almost 2 weeks. In our case, the Balkan stationary anticyclone occurred in conjunction with the omega atmospheric blocking at 500 hPa. This synoptic mechanism in the middle and lower troposphere dominated the area almost for 1 week and resulted in the degradation of the surface wind field and the absence of significant upward motions.

Fig. 6

Synoptic surface pressure map in 30/11/11 at 00:00 UTC

For the aforementioned period, a simulation for the meteorological and dispersion conditions of the area was made by TAPM model. Firstly, the predicted meteorological results by the model were validated against observations from both monitoring sites. Secondly, we used them to interpret the mesoscale mechanisms responsible for the recorded relatively high concentrations, based on the investigation of the wind fields and the ABL evolution. The diurnal pattern of the SO₂ concentrations was investigated and TAPM results were correlated against observations, while modeling dispersions patterns were also used to explain the diurnal evolution of concentrations, nearby and at distant places from the industrial activity.

4.2.1 Meteorological footprints modeling

The hourly averaged statistical performance of TAPM for the meteorological variables of temperature, wind speed and wind direction is presented in Table 2, while Fig. 7 compares the observed and simulated wind direction for the two monitoring sites. More specifically, the statistical measures in Table 2 indicate a good model’s performance for air temperature with values greater than 0.84 and 0.78 for the IOA and Cor, correspondingly. The RMSE for the same variable is 2 °C, a fact which indicates that TAPM can simulate the air temperature at both monitoring sites. Despite the bias in wind speed for the Petrana monitoring site, the prediction by TAPM is acceptable, with respect to the complex topography of the area and the synoptic weather pattern that the simulation was attempted. The IOA is above the threshold of 0.5 in both monitoring sites (0.58 and 0.51 for Pontokomi and Petrana), and an acceptable correlation coefficient is evident Cor = 0.49 and 0.44 for Pontokomi and Petrana, respectively. Regarding the wind direction (Fig. 7) in both monitoring sites it is rather encouraging to note that the shifts in the modeled direction occur at the correct time, indicating a satisfactory prediction. However, some differences appear between the observed and modeled values in Petrana monitoring station. Specifically, the model simulates a southwesterly flow instead of the observed southerly wind. Similar results for the wind direction with those presented in our study were also found by Zawar-Reza et al. (2005), who applied the model in a coastal complex terrain. Zawar-Reza et al. (2005) found that TAPM instead of calculating calms, it produces weak westerly (drainage) winds by gentle sloping. The statistical performance of the model shows an underprediction of the component U with IOA 0.49 (Pontokomi) and 0.39 (Petrana) which is also supported by the low correlation coefficient 0.37 and 0.29, respectively. The component V has a better agreement with IOA 0.72 and 0.68 and a better correlation coefficient of 0.64 and 0.65 for the two sites correspondingly. Taking all the above reasons into consideration, it can be said that the overall performance of TAPM indicates a satisfactory agreement with the observed surface air temperature and winds.

Table 2 Model’s performance statistics for temperature and wind speed

Fig. 7

Comparison between modeled and observed wind direction from 28/11/2011 to 4/12/2011

Figure 8a–c illustrates the 2D diurnal variation of the surface level wind field calculated in the inner grid domain, during morning (08:00), noon (13:00) and night hours (22:00) on 1st December 2011, the day on which the maximum concentrations were observed. As it is depicted in the morning and night hours, the wind field near the surface is more heterogeneous and it is mainly influenced by the topography’s complexity, while during noon and late noon hours the winds are more homogenous. It can be further observed that during the whole day the ventilation conditions of the area are poor, since there is a wind inflow from both valley openings (NE and SE), a fact which is more clear during the warm hours. More specifically, during the noon and early afternoon hours, the incoming from the northeastern opening light wind flow travels southward and it is reinforced due to topographic venturi effects in the center of the basin, where the width of the valley is narrower. Then, it weakens again as it continues to travel southward, since the basin’s width grows again. Simultaneously from the SE opening, the breeze of the local thermal circulation from the artificial lake (Polifitos) starts to develop (see Fig. 8b wind field at 13:00). This opposing flow near the ground in the SE part, that is also evident during morning and night hours due to complex topographic effects (Fig. 8a, c), practically blocks the weak NW flow and results in calm conditions: a case which is observed better in the homogenous wind fields during the warm hours. Such wind fields indicate that under stagnant anti-cyclonic systems, the atmospheric motions in the first atmospheric layers (10 m) in GWMR are almost exclusively dominated by complex mesoscale meteorological phenomena, which are expected to influence the diurnal variation of pollutants.

Fig. 8

Variation of the surface wind field at GWMR as it was calculated by TAPM on 1st December 2011. a–c Wind fields at 10 m above ground level. Time format in the right top of each figure presents day:hour:minute

4.2.2 SO₂ variations and dispersion modeling

The diurnal variation from the predicted and the measured SO₂ concentrations for the aforementioned period is depicted in Fig. 9. The time series between observed and modeled SO₂ concentrations (Fig. 9a, b) show that TAPM follows the daily variation in both stations. Furthermore, the latter predicts the mechanism that leads to the noon peak at both monitoring sites (Fig. 9c, d). The modeled values are following the general variation of the measured SO₂ concentrations at both stations with good correlation coefficient, while the calculated ABL evolution by the model (described later on) can explain the daily variation. The ratio hhc/lhc is 3 and 1.9 times higher in the nearby and at a greater distance monitoring sites, respectively, with the model to follow this ratio. In the diurnal variation, Pearson’s correlation coefficient between modeled and measured SO₂ concentrations for the nearby industrial activity monitoring site (Pontokomi) is 0.68. This fact indicates that the monitoring station is probably affected by other sources (e.g., trucks from the nearby open pit mines), mainly due to its location in the center of the basin (note that as pollution sources were considered only the LPPSs, see Sect. 3.1). For the distant monitoring station (Petrana) the correlation coefficient is 0.92, suggesting in that way that LPPS are the monitoring station’s primary source of attribution. Furthermore, from Fig. 9 it can be observed that the peak in the concentrations during the noon hours is narrower in the center and wider in the southern part, with TAPM to follow this variation.

Fig. 9

Model and observed SO₂ concentrations for the stations Pontokomi and Petrana. a, b Time series modeled versus observed values; c, d mean daily variation between observed and modeled values from 28/11/11 until 4/12/11. Up to the right is indicated the Pearson correlation coefficient between measured and modeled values

For the examination of the mechanisms and the dispersion conditions in the area which lead to the noon peak of the SO₂ concentrations under the above-mentioned period, vertical profiles of potential temperature were calculated by TAPM. Figure 10 illustrates the vertical profile of the potential temperature in the center of the GWMR for 1 December 2011, while Fig. 11 indicates the 3D dispersion conditions of the LPPS’s stacks for the same day at 08:00, 13:00 and 22:00 local time (UTC + 2). In Fig. 10, a ground-based inversion is observed from the afternoon until the early morning hours and increases in strength. During that time a stable inversion layer is evident to existing up to the height of 600 m. After the rising of the sun and the gradually heating of the soil, an unstable layer is evident, which reaches its maximum height from the ground during the noon hours.

Fig. 10

Vertical profile of potential temperature on 1st December 2011, for selected hours during the day, as it was calculated by TAPM model

Fig. 11

3D Diurnal dispersion patterns in GWMR during a morning, b noon and c night hours on 1st December 2011. The time format is day:time:minute. The yellow color detonates the plumes behavior as it was produced by the Lagrangian particle model

The dispersion conditions in the area are almost the same during morning and evening hours, while during the warm hours of the day the dispersion conditions change (Fig. 11). More specifically, during the early morning hours there are emissions from elevated stacks, embedded within a stable deep inversion layer, indicating that there is a rather horizontal transport in a fanning form plume which is traveling for long distances (Fig. 11a). After the sunrise and the gradual heating of the soil, an existence of a shallow convective boundary layer is evident near the ground below 100 m, while higher up at about 500 m a height inversion is evident. Therefore, during the noon hours, the dispersion conditions change, mainly due to the updrafts near the ground coming from the heating of the soil. Hence, the imbedded within the surface inversion fanning plume breaks up by the rising turbulence, when day-time heating of the soil breaks up the inversion, resulting in a fumigation plume (Fig. 11b). The fumigation occurs when portion of the plume which is trapped in the lower part of the overlying inversion layer has small residual buoyancy and it can be transported to the ground by downdrafts. This process leads to relatively high ground-level concentrations during those hours. At night, the anew imbedded plume within a stable inversion layer indicates a fanning narrower form (Fig. 11c).

To examine in detail the SO₂ concentrations from the above dispersion patterns in the two monitoring sites, vertical distributions of SO₂ were calculated and depicted in Fig. 12. TAPM’s equations for mean plume rise of a point source emission are based on a simplified version of the model of Glendening et al. (1984), and a more detailed description which includes the initial conditions of these equations can be found in Hurley (2008). From this Figure, it is observed that the ground-level concentrations exist in both stations during the warm hours. More particularly, in the monitoring station near the industrial stacks (Pontokomi), the short duration relatively high concentrations are evident near the ground level at early noon hours (between 12:00 and 13:00). In the more distant monitoring station (Petrana), relatively high concentrations are evident for longer duration, until late noon/early afternoon hours due to the development of local complex flows and the stations idiomorphic location. During those hours in the southern part and consequently in the Petrana monitoring site, such flows are responsible not only for the wider concentration peak, but also for the development of an internal boundary layer, as it is discussed later on. Analogous dispersion aspects and temperature profiles were evidenced during the whole episodic period which lasted about 7 days (28/11/11–4/12/11). Particularly, the vertical profiles of potential temperature revealed strong ground base inversions as in the case presented here, while the base of the inversion was lifted only around the warm hours of the day and in some cases a shallow convective boundary layer was formed.

Fig. 12

Vertical distribution of SO₂ concentrations during morning, noon and night hours on 1st December 2011

To support these dispersion conditions and to get an insight, turbulence fields from the two vertical cross sections of turbulent kinetic energy (TKE) are illustrated in Fig. 13. In the atmosphere, convective or thermal turbulence co-exists with mechanical turbulence. From Fig. 13, it is evident that during the morning and night hours some (mechanical) turbulence exists in the mountains ridges and tops, which is the result of the eddies that are created from the mesoscale flow and their slow dissipation. During the noon hours (14:00), turbulence is mostly thermal. The thermal turbulence is caused by the relatively strong updrafts from the lower (near ground) atmospheric layers due to day-time heating, whereby warm volumes of air carry thermal energy upwards (convective heat transport) and the turbulence fields are rather distorted. At late noon (16:00) some abatement is evident in the TKE, because of the reduction of the thermal energy updrafts, since almost neutral conditions exist in the atmosphere (see Fig. 10, corresponding temperature profile at 16:00). For that reason, mechanical turbulence is the one that remains, which in fact is the only mixing mechanism during those hours. This kind of turbulence is evident during those hours more clearly in the southern part of GWMR, due to the mesoscale-influenced flows near the mountains and hills (e.g., katabatic winds), different surface roughness and/or lake effects (see Fig. 13, AB and CD cross sections at 16:00 in the southern and eastern part, respectively). In general, the TKE had small values, ranging from 0.34 m²/s² (morning and night hours) to 1.28 m²/s² (noon hours). The mixing layer abruptly sunk when a sudden decrease in the TKE was evident, while when mechanical and/or convective turbulence existed, broadening and growth with distance of the mixing layer was evident (see Fig. 13 at 16:00). This behavior of the mixing layer is especially apparent in the southern part due to complex topographic features. Such case needs further investigation and will come as a part of a forthcoming work concerning the development of an internal boundary layer in this area.

Fig. 13

Turbulent kinetic energy vertical cross sections along AB (south–north, y–z at 15 × point) and CD (west–east, x–z at 6 y point) shown in Fig. 2, on 1st December 2011. The contour interval is 0.1, while the z is depicted in the right side of each cross section. The straight line indicates the boundary layer Zmix

5 Conclusions

In the present work, SO₂ hourly average concentrations and surface meteorological data in an area with complex topography and multiple power plants stacks were analyzed. The mean hourly values which were used for the analysis were collected from two in situ monitoring stations, located nearby and in a greater distance from the power plants stacks. The period with the highest concentrations (worst case) was selected and analyzed regarding the development of the atmospheric boundary layer and the evolution of dispersion conditions with the implementation of TAPM, a mesoscale prognostic meteorological and air pollution model.

Generally, meteorological and dispersion TAPM simulations showed substantial agreement with observations. TAPM’s statistical performance-validation measures (IOA, RMSE and Pearson’s correlation coefficient) for meteorology indicated good overall prediction for temperature and wind speed, while the comparison between the observed and modeled shifts of the wind direction was also agreeable, despite the evident underprediction in the statistical performance regarding the component U. TAPM was also capable of simulating the main processes that lead to the noon peak of concentrations. The model followed the observed diurnal variation of SO₂ in both monitoring sites with r = 0.68 (nearby the industrial activity monitoring site) and r = 0.98 (at a greater distance monitoring site).

From the analysis, a diurnal variation of SO₂ concentrations was observed which presents a different pattern depending on the distance from the stacks. A characteristic peak appeared during the noon hours in the station near the industrial stacks of about 1–2 h in duration, with the ratio highest hourly concentration/lowest hourly concentration (hhc/lhc) equal to 2.5. A rather different variation presented the more distant station, with a smoother variation and a ratio of hhc/lhc = 1.5. The above pattern was more characteristic during the worst case period.

Regarding the flow-dispersion interaction, the simulated wind fields were found to be more heterogeneous, driven by the complex topographic characteristics during the morning–early night hours, and more homogenous during the noon hours, while ventilation and dispersion conditions improved when topographic venturi effects occurred.

Considering the air pollution and dispersion mechanisms, during the diurnal pattern, the vertical profile of potential temperature revealed strong temperature inversion which evolves from late afternoon until early morning hours and increases in strength. Throughout noon hours, the existence of a shallow convective boundary layer and a height inversion aloft occur. During morning and night hours there is a horizontal transfer of pollutants, illustrating a fanning plume, while after sunrise, with the development of a shallow convective layer and a height inversion aloft, fumigation plume occurs, resulting in a relatively high ground-level concentrations. Turbulent kinetic energy fields had changes in the mixing layer caused by both mechanical and convective turbulence. The mechanical mixing is clearer in places induced by complex flows and influences the development of the atmospheric boundary layer. More in-depth analysis for this layer will come as a forthcoming work about the development of an internal boundary layer.