Background and introduction
In 1973 the European Commission (EC), through Directorate General (DG)/XII (Research), began supporting research into the atmospheric physical and chemical processes that govern the dynamics of air pollutants in Europe. In particular, the six European Remote Sensing Campaigns (ERSC; Guillot et al. 1979; P. Guillot 1985, personal communication; Sandroni and de Groot 1980;Le Bras 1988) were instrumental in documenting that polluted air masses in central Europe—that is, in Belgium (Ghent), in France (Cordemais), and in the United Kingdom (Drax)—mostly were advected, but they underwent oscillations with marked diurnal cycles in southern Europe—that is, in France (Lacq, Fos-Berre) and in Italy (Turbigo). A similar approach had been used in 1968 and 1969 to track the dynamics of pollutants in the Los Angeles and San Francisco basins of the United States (Barringer Research, Ltd., 1969; Newcomb and Millán 1970).
Analysis of the data from the sixth ERSC (Fos-Berre, Marseille, France, June 1983) suggested that pollutants could be recirculated vertically in some regions of southern Europe in ways similar to those documented in the Great Lakes of North America by Lyons and Olsson (1973), Lyons and Cole (1973, 1976), and Portelli et al. (1982). This hypothesis was recalled in 1985 to explain pollution cycles in Spain and to integrate the available data within a coherent regional picture (Millán et al. 1991). It also formed the basis for the Mesometeorological Cycles of Air Pollution in the Iberian Peninsula (MECAPIP) project to study the relationship between observed pollutant behavior and the formation of the Iberian thermal low (Millán et al. 1992, 1996). Two other projects followed to document the continuity of the observed processes and to compile a mosaic of atmospheric circulations over the western Mediterranean basin (Regional Cycles of Air Pollution in the West-Central Mediterranean Area; RECAPMA) and across the whole Mediterranean basin (South European Cycles of Air Pollution; SECAP) (Millán et al. 1997; Kallos et al. 1998).
In these projects, air pollutants were regarded as tracers of opportunity of the atmospheric flows in midsummer (July). As a result, more knowledge is now available about specific interactions between meteorological processes in the Mediterranean from local to subcontinental scales. Other EC projects (Le Bras 1993; Angeletti and Restelli 1994; Le Bras and Angeletti 1995; Cox 1997; Seufert 1997) have documented further that the physical and chemical processes affecting ozone formation vary greatly across Europe and even within the Mediterranean basin. In summer, the western basin, which is surrounded by high mountains, is under the influence of weak levels of anticyclonic subsidence, low winds, and strong insolation. These conditions favor the development of mesoscale processes and recirculations within air masses (Millán et al. 1992, 1996). During the same period, the eastern basin is under conditions of weak ascent and strong advection, that is, the Etesian winds, and the development of recirculations is largely inhibited (Millán et al. 1997; Ziomas et al. 1998).
In the meantime, DG/XI (Environment) has been preparing air quality legislation, and the new European Daughter Directives on Ozone are expected in 1999. These directives will require data from regional and national monitoring networks, some of which are still to be designed and/or optimized. It is currently felt that the spatial and temporal coverage of ozone data in some European countries is not adequate (EMEP 1998), and questions have been raised regarding site representativeness. Statements also have been made to the effect that monitoring stations must be geographically and climatologically representative, that they should help to identify and to explain the formation and transport of ozone (O3) and its precursors, that they should monitor changes in O3 concentrations in areas affected by background pollution, that the concurrent meteorological measurements should be representative of the meteorological conditions that affect pollutant dispersion within the area to be monitored, and so on.
We feel that these statements are just too vague, and, because no procedures are suggested, they also hint at a problem in Europe as compared with North America, that is, the generalized lack of a subculture in applied air pollution meteorology, which has led to an excessive reliance on modeling as the answer to all these questions. Modeling, however, has not proved yet to be a substitute for a proper analysis of the meteorological processes involved (Noll and Miller 1977), and invoking the use of coupled meteorological and photochemical models eventually runs into the same problems for their validation (Salvador et al. 1999). We also feel that these statements tend to overshadow the real problems of (a) how to interpret the available data, and, in particular, (b) how to incorporate the new knowledge into legislative activity, that is, to the accepted know-how, in a timely way.
In an attempt to bridge this gap, available data have been combined with mesometeorological analysis and modeling results to explain the processes and typify the observed ozone cycles in a complex coastal area in the western Mediterranean in summer. The monitoring network and the numerical model used are introduced in sections 2 and 3. The numerical model has been used as a tool to fill in observational gaps and to structure the more relevant experimental results from the MECAPIP and RECAPMA projects, which are presented in section 4. In section 5, a conceptual model is introduced to support the interpretation of the monitoring data presented in section 6. Sections 7 and 8 present the typification of the diurnal cycles and the conclusions, respectively.
The monitoring network
Figure 1 locates the study area within southwestern Europe and shows the three modeling grids used in this work. The monitoring network, shown within grid 3 in Fig. 2, was designed in 1993–94 after the MECAPIP project had documented the recirculations in this region. Some of its design objectives were to compile a historical database of the processes along the same atmospheric paths used in the MECAPIP project, to support measurement campaigns for other EC projects in the western Mediterranean, for example, Biogenic Emissions in the Mediterranean Area (Seufert 1997), and to obtain data to validate coupled meteorological and photochemical models. The first four stations, at the northern edge of the domain, began operation in 1995, the four near Castellón started in 1996–97, and two more will begin operation in 1999.
The network (Fig. 2) includes fully automated stations (squares) and field sites (small diamonds). Measurements of sulfur dioxide (SO2), total suspended particulates (TSP), nitric oxide (NO), nitrogen dioxide (NO2), and O3 are made at the automated sites year-round, whereas the field sites have been equipped with automatic instruments only during brief intensive measurement campaigns (1–2 week) in July since 1989. During these campaigns, tethered balloon soundings (to ∼1000 m) were made at Grao, Onda, and Valbona, and upper-atmospheric soundings (including O3) also were made in Burriana and Valbona in 1991 and 1995–97. Meteorological measurements (temperature, wind speed, and wind direction) at 10 m are made continuously at all sites. Table 1 lists the sites, their coordinates, and the altitudes.
The three stations near the city of Castellón (Ermita, Grao, Penyeta) are intended to provide links between O3 concentrations measured near the coast and those inland, and this linking now can be done along several possible transects. This group includes a station (Penyeta) on the side of a steep mountain at some 250 m above the coastal plain. Because the ground-based nocturnal inversion over this coastal plain in summer was known to be on the order of 100–150 m or less (Millán et al. 1992), the siting for this station was chosen specifically so that it would remain above the stable surface layer during the night.
The variations in O3 cycles with orography for July are also shown in Fig. 2 for the stations underlined. Graph (a) illustrates variations along a coastal valley with data from the MECAPIP project; graph (b) uses data from the network to illustrate the height dependence. The most notable features are the changes in shape from the sites at the coast to those at the mountain tops. At the coast, the profiles look like square waves. A sharp morning rise is followed by nearly constant O3 values during the afternoon and a drop to low values at night. The mountain sites show a nearly flat profile with high concentrations both day and night.
Figure 3 shows the ensemble averages of the O3 diurnal cycles for each month of 1997, and is intended to compare the concurrence between certain processes (section 6a) and to emphasize the absolute values recorded and their annual evolution. These averages document the existence of chronic-type episodes with persistent medium-to-high O3 concentrations for several months (i.e., March–August), as compared with shorter (i.e., 1 week), peak-type episodes in central Europe. They further confirm the marked differences among the cycles at stations on the coastline, coastal plains, midvalley, and mountain tops, all located within a 100-km square, and between stations (Ermita/Grao and Penyeta) located just a few (i.e., 9) km apart. For people in charge of air quality networks, these cycles can only lead to perplexity and/or exotic conclusions if interpreted in terms of “typical O3 cycles” such as those seen in Los Angeles or the “averaged” outputs from models with a 150 km by 150 km horizontal gridcell size.
Numerical modeling
Modeling of the observed processes was initiated in 1993. The Regional Atmospheric Modeling System (RAMS version 3b; Pielke et al. 1992) has been employed to reproduce the observed mesometeorological processes. The model is used in a test bench mode in which the selection of the initial parameters is performed through an iterative procedure of comparison with experimental data from MECAPIP and following projects. Current simulations use a 3D, nonhydrostatic atmosphere in a terrain-following vertical coordinate system with polar stereographic coordinates.
For this work, a nonhomogeneous initialization with nonstationary boundary conditions was used. The model was run for 27 July 1989, on which date extensive aircraft data were available from MECAPIP (section 4). It was initialized with the European Centre for Medium-Range Weather Forecasts global analysis data with 1° gridcell size in longitude and latitude. These data were ingested into the isentropic analysis package of RAMS as part of the model initialization procedure.
Three nested grids were used—a coarse grid (grid 1) with a 1440 km by 1184 km domain (Fig. 1) was run with a 16-km gridcell size, a medium grid (grid 2) with a 360 km by 296 km domain was run with a 4-km gridcell size, and a fine grid (grid 3) with a 180 km by 148 km domain that focused on the monitoring-network area of Fig. 2 was run with a 2-km gridcell size. In the vertical, 24 levels were used with geometric grid stretching that increases with height, and the model top was located at 15 km. Nesting was not applied in the vertical coordinate, and thus the three grids use the same vertical spacing.
The model was run for 30 h, starting at 1800 UTC 26 July 1989, to allow for numerical spinup. Climatological values of sea surface temperatures were used for initialization. The U.S. Geological Survey dataset for defining the underlying terrain with a 30 arc-s resolution for land use and a homogeneous soil texture, set equal to loamy sand, also were used. Both land use and soil properties datasets are being upgraded for Spain and Portugal and will be incorporated into the model as they become available; in the meantime, improvements to the model have been focused on selecting the optimum gridcell size (Salvador et al. 1999).
Coastal circulations
Figure 4 shows four stages in the development of the combined sea-breeze and upslope winds on the Spanish east coast as obtained with the model for 27 July 1989. These stages are substantiated by the data obtained with the instrumented aircraft in MECAPIP (Millán et al. 1992, 1996, 1997). The streamlines are shown for the first model layer at 57.3 m above the surface. The vertical cross section of the wind field was prepared from the computed 3D wind field by multiplying the vertical component by 10 and projecting the resulting field onto the vertical plane, which follows the Mijares River valley, that is, the southeast–northwest-oriented dashed line in Fig. 2, which passes over four monitoring stations.
The first set of plots, at 0400 UTC, shows the drainage flow along the Mijares valley and other drainage flows initiated near the mountain tops. When they reach the sea they encounter southerly winds along the coast. By 1200 UTC, the leading edge of the breeze reaches some 60 km inland along the Mijares valley. By 1600 UTC, the streamlines show that the breeze, at almost 90 km inland, has overcome the ridges at the head of the valley northwest of Valbona (Fig. 2) and is in the process of joining with the flow from the valleys to the South (Palancia, Turia, Fig. 2). By 1800 UTC, the breeze has penetrated almost 100 km inland.
The upper plots in Fig. 4 also show the development of a chimneylike effect at the leading edge of the combined sea-breeze and upslope winds, the evolution of the return flows (with several layers), and the compensatory subsidence above the breeze layer. The injection of pollutants into the return flows and subsidence along their path toward the sea are both instrumental in creating (reservoir) layers of secondary pollutants aloft and have been documented by Millán et al. (1992, 1996, 1997).
A useful result of the modeling effort has been to fill in some observational gaps in the evolution of the sea-breeze front. Figures 4a–c clearly show that the position of the convergence line that defines the (frontal?) edge of the combined sea-breeze and upslope winds varies with the orography and time of day. It extends further inland along the coastal valleys and is closer to the sea wherever it encounters coastal mountain ranges. Likewise, as part of the same process, the chimneylike injection into the return flows becomes locked and nearly stationary over the mountains by the coast while it still is progressing inland along the coastal valleys.
The Iberian thermal low (ITL)
In Fig. 1a the ITL for 1800 UTC 27 July 1989 appears as a closed loop of low pressure. However, the available observational data and modeling results (Fig. 5) show that its structure consists of a combination of several thermal circulatory cells that grow during the day and merge into an organized circulation system at the peninsular scale by noon or soon afterward. Modeling also has shown that the surface flows coalesce into several major convergence lines that become locked to some of the main orographic features by midafternoon. The edge of the Mediterranean coastal cell reaches approximately to the tops of the coastal mountains, and the central cell is supported by a line of strong convergence along the mountains north of Madrid (Fig 1b). The results also show that the finer structure can change from day to day (Millán et al. 1997).
One of the most important effects of the ITL is to produce dynamic compensation over the Atlantic Ocean and Mediterranean Sea, which intensifies the anticyclonic subsidence over these waters (Fig. 5a). The change in depth of the sea-breeze layer near the sea, as a result of the overall development of the ITL, albeit with low vertical resolution, can also be observed by comparing the model results at 1200 and 1600 UTC in Fig 4. This process was documented experimentally with simultaneous soundings at Sines, Portugal, and Grao, Castellón, in 1991 and 1995 during the RECAPMA and SECAP projects (Millán et al. 1997), and it affects the evolution of the boundary layers along most of the coastal areas surrounding the Iberian Peninsula (Alonso et al. 2000).
The soundings at Grao have shown that the top of the boundary layer within the sea breeze rises during the morning, begins to sink as soon as the ITL becomes consolidated around noon, and continues sinking during the afternoon (Fig. 11). Along with this behavior, weak subsidence inversions that appear in the upper reaches of the soundings sink during the afternoon, and, by early evening, several inversions can be observed (Millán et al. 1992, pp. 58–59) and remain until the following morning, as shown in Figs. 6, 7, and 12.
The terrain-related evolution of the breeze is the most obvious difference in relation to results obtained in flat, coastal regions (Lyons and Olsson 1973; Lyons and Cole 1973, 1976; Portelli et al. 1982), over which the lake-breeze front progresses inland in a more or less uniform way. Another difference is the evolution of the coastal boundary layers, caused by the amplified subsidence attributable to the combined sea-breeze and upslope flows plus the formation of the ITL. As a result, this system becomes stratified strongly in comparison with the residual layer(s) in classic boundary layer theory (Stull 1988). A third, and perhaps most important, difference is that these residual-return-reservoir layers, under the influence of weak synoptic winds during the night, that is, on the order of 5 m s−1 or less (Fig. 19), do not migrate away from the region.
Supporting experimental data
The O3 reservoir layers over the sea
The result of the return flows of the previous day(s) plus their compensatory subsidence over the sea for 27 July 1989 appears in Figs. 6 and 7. These figures show the vertical ozone distribution along 450 km of the Spanish east coast over the flight tracks in Fig. 1; the soundings along two legs in each flight (points F1, F2, M1, and M2 in Fig. 1) illustrate the stratification. Figure 6 shows the results from the outgoing flight (0447–0553 UTC), 20–40 km offshore, and Fig. 7 shows the data from the return flight (0600–0710 UTC) over the coastal plains. The average separation between the two flight tracks was 50 km (Fig. 1b).
Figure 6 shows two holes in the O3 distribution, produced by 1) the drainage of the Ebro valley, which appears between 160 and 270 km with lower O3 concentrations, and 2) titration by NO in the plume from Tarragona at 360 km. In Fig. 7, the vertical scale extends to 2500 m, and the drainage of the Ebro valley is centered at 250 km. Other data from MECAPIP and RECAPMA (Millán et al. 1997) indicate that along more than 600 km of the Spanish Mediterranean coast the layers reach 2–3 km in height (e.g., O3 profile for leg 1 in Fig. 7), they have variable widths over land, that is, up to 100+ km, depending on the orography and the processes of the previous day(s), and they extend more than 300 km out to sea.
Recirculations
On the Spanish east coast, two main boundary layers form within the advected marine air mass. A thermal boundary layer (TBL) develops offshore as a result of the strong gradient in the water temperature from the shore (at 25°–28°C in July) to a few km out to sea (at 20°C or less). This kind of process was documented by Portelli et al. (1982) in Lake Ontario and has been observed in all the soundings at Grao (Castellón). The profiles on the right side of Fig. 11, from Millán et al. (1997), show the sinking of the top of this layer on 23 July 1991.
The TBL is also responsible for the thorough mixing of the O3 from the reservoir layers above the sea before the air mass arrives at the shore and, thus, for the flattop shape of the O3 cycle at those sites (section 6). Once the breeze flow reaches the shoreline, a second, thermomechanical boundary layer (MBL) develops over land. When the sea breeze is fully established over the coastal plains, the two layers merge approximately 10–15 km inland.
Recirculations are created when the lower layers fumigate the surface the next morning and those layers located just above the sea are carried inland by the sea breeze the next day(s). Experiments on the Spanish east coast have shown that tracers released in the morning, after the onset of the sea breeze, return with the breeze of two days later (Millán et al. 1992, pp. 149–155). Similar processes involving either the vertical recirculation or the horizontal oscillation of air masses also have been documented in other Mediterranean regions (Ciccioli et al. 1987; Fortezza et al. 1993; Georgiadis et al. 1994; Orciari et al. 1998).
Chemistry
In the MECAPIP project, two instrumented flights (Fig. 2) documented the effects of chemical transformations along the path of the coastal recirculations and also were used to track the tracers injected into the plume of the 150-m Castellón power plant stack. In these flights, a nearly constant height above the surface was maintained (300 m for the coastal flight and 400–500 m for the inland flight). For the second flight, the objective was to reach the edge of the combined upslope and sea breezes.
Figure 8, from Millán et al. (1996), shows the relationship between the NO and O3 plumes directly downwind of Castellón approximately 3 h after the onset of the sea breeze. The NO urban plume is reacting with the O3 background that enters with the sea breeze, resulting in a distinct O3 depletion. Figure 9 shows the data from the late-afternoon flight that reached the tops of the coastal mountain ranges; a similar relationship between NO and O3 also is evident.
Fig. 10 complements Figs. 4 and 9 by showing the measured wind field at aircraft altitude (some 400 m above ground) and the modeled streamlines at the nearest grid height (346.5 m above ground). The observed wind field indicates that the edge of the anabatic winds, marked by the line of converging winds, approximately follows the ridges of the mountain tops. The modeled and real data show reasonable agreement on the location of this line north and east of the Valbona site. Some other effects, for example, the details of the wind merging southeast of Valbona, are not replicated as well yet.
The conceptual model
Figure 11 summarizes the main aspects in the development of the combined sea-breeze and upslope winds and their return flows during the day. Letters a–d indicate, generically, successive stages in the breeze development and the formation of stratified reservoir layers aloft. As stated in section 3a, however, they cannot be assigned specific times or distances because these two things depend on the orography.
In this region, most of the NO emissions occur on the coastal plains, and the flows essentially are perpendicular to the coastline. Thus, the Derwent and Davies model (1994) that relates NO emissions to the production of O3 and other oxidized products is considered to apply conceptually along the path of the air masses from the coast to the mountain tops as shown in the lower half of Fig. 11. A point to consider again is that, during the development of the combined sea-breeze and upslope winds, the injection into the return flows occurs at various distances from the coast and heights above the ground. This fact also means that the degree of aging of the pollutants injected aloft varies during the day, and photochemical reactions can take place within the return flows and within the reservoir layers. Modeling of these chemical processes is currently under way.
During the evening and night the solar-driven circulations die out. Air masses in transit along the surface and in varying degrees of aging come to a stop and become part of a developing stable surface layer. This process may include, in approximate sequence: stagnation with the formation of a ground-based inversion, accumulation of drainage air from the valley slopes, deepening of the ground-based inversion, and reversal of the surface flows. The formation of the stable surface layer (or surface layers, since each valley develops its own) also results in its progressive decoupling from the air and pollutants located just above, which then become part of the (residual?) reservoir layer system aloft.
In a stably stratified system like this one, the layers are decoupled from each other vertically and are free to move horizontally. The mobility of the layers aloft, however, decreases as the extent of the topographic confinement increases. The layers located above the coastal plains are more mobile than those within the valleys, and the most mobile layers, that is, those over the sea, can drift along the coasts during the night and contribute to interregional transport of aged pollutants. All these layers act as reservoirs of pollutants for processes that occur during the night (upper mountain sites) and the following day(s).
At night (Fig. 12), the mountain tops and ridges remain inside the layers formed the previous afternoon and are at the origin of the developing drainage winds, as shown in Fig. 4a at 0400 UTC, with the result that some of the air aloft may flow toward them. These stations also can be exposed to layers formed elsewhere and transported during the night under decoupled conditions. Either one of these, or a combination of both, may ensure a continuous supply of O3 from the reservoir layers to the stations located on the mountain tops or high on valley slopes. Another example of the stable structure over the sea and the O3 layers recorded by an instrumented aircraft on 20 July 1989 appears at the right of the figure.
By dawn, drainage flows are established. Depending on the valley aspect (Whiteman 1990), the cold air can accumulate on the bottom or keep flowing toward the coastal plains and the sea. Ozone is consumed along the path of the drainage flows by contact with the surface and/or is titrated by fresh surface NO emissions. Surface ozone concentrations thus can become very low on the valley floor and coastal plains, while, at a certain height above the valley floor and over the sea, the O3 levels remain closer to those of the previous afternoon.
Interpretation of the O3 cycles
To synthesize the daily variation and to emphasize the shape of the cycle, the ensemble average of the July diurnal cycles at each site, for the years shown in each figure, has been normalized for the years shown in each figure, by dividing each value by the maximum and multiplying by 100. In the normalized cycles the vertical scales are nondimensional. For the same periods, the averaged wind parameters, that is, direction (vectorial average) and speed, have been included in the same figures to relate their variation to the O3 evolution.
Transect along a coastal valley
Figure 2a shows that the O3 concentrations drop to very low values in Burriana during the night. In Cirat and Valbona, located on the valley floor at some 38 and 75 km inland and at 420- and 950-m altitude, respectively, ozone still is available from the reservoir layers;that is, it has not been depleted fully from the drainage flows, and the concentrations do not drop below 50–60 μg m−3 during the night. Between 0700 and 0930 UTC, the O3 increases and reaches nearly the same concentrations at the three sites simultaneously, and the same behavior has been observed at Onda (operating since 1996). Their normalized cycles (Figs. 13–16) show that this increase occurs with a minimum of wind speed at about the time it changes direction. This timing is consistent with a “classical Hewson-type fumigation” (Munn 1966) of ozone from the reservoir layers aloft and is associated with the development of convective activity just before the onset of the sea breeze.
After the initial rise, the O3 remains nearly constant at the coast during the breeze period (Fig. 13) and can be regarded as the background level entering with the sea breeze. It originates within the marine TBL from fumigation of O3 in the reservoir layers above the sea (sections 4a,b). In Onda, at 20 km from the coast (i.e., midway to Cirat), the O3 maximum occurs around 1200 UTC, before the maximum wind speed is reached (Fig. 14). In Cirat (Fig. 15), the O3 keeps increasing during the morning as a result of transport and new photochemical production from precursors emitted along the coast. It reaches a maximum around 1300 UTC with a well-developed sea breeze.
In Valbona (Fig. 2a), the initial rise is followed by nearly constant values or very small growth until the sea-breeze front arrives after 1400 UTC along with an increase in wind speed (Millán et al. 1996) and produces a second rise and the diurnal maximum by 1700 UTC. The normalized average (Fig. 16) shows a pronounced inflexion (or shoulder) between 1000 and 1300 UTC, which represents the delay between the morning fumigation and the arrival of the coastal air mass with more O3. The maxima for both O3 and the wind speed appear near 1600 UTC. A similar situation, but with much shorter inflexion, can be noticed on the Cirat profile at about 0900–1000 UTC but is not observed at Onda.
Transect with altitude dependence
Figure 17 shows data from 1997 to complement the graphs in Fig. 2b. Ozone evolution at the coastal site (Grao) follows a trend similar to that of the Burriana site (Figs. 2a and 13). Figure 18 shows that the Villafranca site, located at a saddle-shaped valley head, behaves similarly to Valbona (Fig. 16), including a shoulder between 0900 and 1200 UTC. At this site, the sea breeze arrives around 1200 UTC and the maxima for O3 and wind speed occur around 1500 UTC.
During the night, the mountaintop stations at Morella and Corachar (Fig. 17) remain inside the reservoir layers, and the concentrations do not drop below 80–90 μg m−3. Winds on mountaintops sometimes can be considered as valid indicators (or surrogates) of the synoptic winds: for example, during the night and under stable conditions. Thus, it is important to notice that the average winds at these mountain sites on summer nights do not exceed 4–5 m s−1 in Corachar (Fig. 19) and are even lower in Morella (below 3 m s−1, not shown). These low wind speeds are an indicator of the degree of stagnation of the air masses in this region during the summer.
At these mountaintop stations, the diurnal maximum also is reached in the midafternoon (around 1300 UTC) when O3 from new photochemical production arrives at the summit with a fully developed sea breeze. After this time, the O3 decreases slowly through the evening and night. This pattern can be seen in the normalized cycle for Corachar shown in Fig. 19. The O3 daily minimum occurs in the morning, and it is significant that it coincides with the steep morning rise at the other stations. Figure 3 also can be checked for these concurrences.
We feel that the morning minimum is, essentially, the “complement” to, or the upper side of, the surface fumigation process. Increased solar heating in the morning results in convective mixing and the onset of upslope winds. This process brings O3 down from the upper (reservoir) layers (i.e., the classic fumigation for a surface station) while lifting up O3-depleted and/or NO-enriched air from the surface. Thus, for a station located inside the reservoir layers during the night, the mixing with surface air from the valleys below can only decrease the observed O3 levels.
The effect of mixing O3-depleted or NO-enriched air from the surface with O3 from the reservoir layers aloft is also very evident in the data from stations located high above the coastal plains near the urbanized coast. Figure 20 shows the monthly averaged O3 cycles in Ermita and Penyeta (located less than 9 km apart) for July 1996 and 1997. The coastal site (Ermita) follows the expected square-wave behavior, and at Penyeta the cycle is not as pronounced, it has higher average values, and the morning minimum coincides with the sharp rise in concentrations at the lower site(s). This behavior is more evident in the individual daily cycles shown in Fig. 21.
High coastal sites
The observed cycles at the Penyeta station or at other similar sites could be especially difficult to interpret in comparison with sites on the coastal plains, for both night and day, unless the effect of the evolution of the boundary layer is taken into account. During the night, the complexity arises because the O3 values are affected by the depth of the stable surface layer. If the top of the stable layer is lower than the height of the station, the site remains within the lower reservoir layers and the expected behavior is that shown in Fig. 21, that is, the O3 fluctuates around a nearly constant value during the night and shows a narrow and sharp drop at the onset of convective mixing at the station, just before the onset of the sea breeze, as indicated by a minimum in wind speed and a sharp change in direction. If the stable surface layer deepens during the night and rises to engulf the station, the O3 concentrations start to decrease and can drop to low values by dawn, that is, almost as low as at stations on the coast, as shown in Fig. 22, and, in this latter case, they also show the morning O3 rise.
During the day, this kind of station also can remain within the marine TBL, that is, below the upper boundary of the developing sea breeze but above the MBL developing over the coastal plain (section 4b). As long as the top of the TBL stays below the elevation of a high coastal site, the station is within the return flow, or the residual air aloft. If the top of the TBL should rise above the elevation of the site, the site is then in the freshly polluted sea-breeze flow. This situation also emphasizes how important it is to have complementary meteorological measurements at these stations to help to diagnose the right situation, for example, measurements of wind direction and speed. For these situations, we feel that relative humidity can be used to discriminate the origin of the air mass.
A typology of observed O3 cycles
In relation to the stations in Fig. 2, and the numbers shown in Figs. 11 and 12, the typified cycles are as follows.
Type 1: Coastline stations (Burriana, Grao, Ermita, Benicarló)
During the day they are immersed within the sea breeze, which brings premixed O3 from the reservoir layers over the sea. In the evening or night the wind reverses, and these stations fall inside the drainage flows, which are enriched in NO emissions from the coastal strip and depleted in O3. The O3 cycles assume nearly the form of a square wave, as shown in Figs. 13, 17, and 20.
Type 2: Valley floor or coastal plains at varying distances inland within the O3-producing region (Onda, Cirat, S. Domingo)
Their cycle shows a morning rise, due to fumigation of old O3 from the reservoir layers, that merges almost directly with another rise and the diurnal maximum from new O3 produced photochemically within and advected by the developing sea breeze. During the night they sample drainage air partially depleted of O3 (Figs. 14 and 15).
Type 3: Upper-valley floor in the nitrogen oxide (NOx)-limited region (Valbona, Villafranca)
During the day they are within the path of the pollutants coming from the coast, and during the night they sample drainage air partially depleted of O3. Their cycle shows a morning rise from fumigation of O3 from the reservoir layers. This rise is followed by a clearly marked shoulder and a second rise with a maximum when the sea-breeze front brings more O3 of new production from the coast (Figs. 16–18).
Type 4: Mountaintops (Corachar, Morella, Vallibona, Camarena)
During the day they form part of, or are within, the chimneylike effect that injects pollutants into the reservoir layers aloft. During the night they remain inside the residual-reservoir layers. Their cycle is not very pronounced and shows high O3 concentrations (Figs. 17 and 19). During the night these sites also could be sensitive to interregional transport of O3 layers, including those injected elsewhere. Their diurnal minima coincide with the maximum rise at the stations in the valleys and coastal plains below.
Type 5: High coastal sites (Penyeta)
Ozone cycles at stations located on elevated ground (mountaintops or on sloping terrain) near the sea depend on the relation between the elevation of the site above the coastal plains and the depths of both the stable surface layer during the night and the marine TBL during the day (Figs. 20–23). For example, the shape of the graph in Fig. 23 shows a minimum at the time of maximum convective activity, which occurs in the morning just prior to the arrival of the sea breeze, and a flat top during the breeze period. Given section 6c and Figs. 21–22, it could be concluded that, on the average, the Penyeta site in July remains above the stable surface layer during the night and within the TBL of the breeze during the day.
Conclusions
The main point to notice is that the shape of the observed cycles at the different types of sites is invariant in spite of the interannual fluctuations in the absolute O3 concentration. This consistency emphasizes the fact that mesometeorological circulations are consistent features of the Mediterranean climate system, whereas chemical production of O3 is dependent on emissions that may change from year to year. Furthermore, it appears that mesometeorological processes are the dominant cause of the repetitive shape of the observed diurnal cycles of O3 in summer.
Because the observed O3 cycles depend on the topographic location of the observing station, they can show large differences within tens of km. Thus, each O3 monitoring station shows a part of the whole picture and even could be considered to represent a specific area, providing that the relevant processes are understood for each site and surrounding area. No single station, however, can be considered to be representative of (average) regional processes, much less the whole situation.
Other aspects of this problem are 1) that the classic concept of upwind (background concentrations) and downwind of conurbations (polluted) is inappropriate in regions of complex coastal terrain where recirculation processes are important, 2) that ozone is generated at the regional scale from emissions in urban centers and other NOx source areas, and 3) that as much as 60%–100% of the daily observed O3 at any one place, for example, coastal or mountain ridge sites, may result from fumigation from reservoir layers and/or advection within the recirculating air masses. A final point is that the modeling of atmospheric dispersion, on any Mediterranean coastal site, can lead to questionable results if one does not take into account that the evolution of the boundary layer can be dominated by mesoscale processes acting at distances of from 100 to 400 km from the coast.
Perhaps the most important issue, we feel, is that, in some international programs (e.g., Cooperative Programme for the Monitoring and Evaluation of the Long-Range Transmission of Air Pollutants in Europe) and other extensive modeling efforts before 1986 or even today, few if any of the atmospheric processes described here were considered to exist or to be relevant in these regions despite available experimental results that show otherwise. Therefore, the use of these same models, with typical 150-km gridcell sizes (or even with 50-km gridcell sizes), for regulatory purposes, forecasting scenarios, policy decisions, or economic evaluations should be seriously and consistently questioned until they are made to reflect the true workings of the system.
Acknowledgments
The work presented here has been supported partially by the Commission of the European Communities, under projects EV4V-0097-E (A), STEP-0006-C, and EV5V-CT91-0050(L). The Comisión Interministerial de Ciencia y Tecnología of the Spanish Ministry for Science and Technology, ENRESA and BANCAIXA, also supported this work financially. We also acknowledge the support of the Consellería de Medio Ambiente of the Generalitat Valenciana and, in particular, to José Vicente Miró. The instrumented aircraft measurements were financed by the Joint Research Centre of the European Commission. We thank one of the referees for suggestions to improve the quality of the paper.
REFERENCES
Alonso, L., G. Gangoiti, M. Navazo, M. Millán, and E. Mantilla, 2000: Transport of tropospheric ozone over the Bay of Biscay and the eastern Cantabrian coast of Spain. J. Appl. Meteor.,39, 475–486.
Angeletti, G., and G. Restelli, Eds., 1994: Physico-chemical behaviour of atmospheric pollutants. Proc. Sixth European Symp., 2 vols., Varese, Italy, European Commission Directorate General XII, 1074 pp.
Barringer Research, Ltd., 1969: Optical measurements of sulphur dioxide and nitrogen dioxide air pollution using Barringer correlation spectrometers. Contractors Rep. [NTIS PB 193-485.].
Ciccioli, P., E. Brancaleoni, C. Di Paolo, A. Brachetti, and A. Cecinato, 1987: Daily trends of photochemical oxidants and their precursors in a suburban forested area. A useful approach for evaluating the relative contributions of natural and anthropogenic hydrocarbons to the photochemical smog formation in rural areas of Italy. Physico-Chemical Behaviour of Atmospheric Pollutants (1986). Fourth European Symposium held in Stresa, Italy, September 23–25, 1986, G. Angeletti and G. Restelli, Eds., D. Reidel, 551–559.
Cox, R. A., 1997: Overview of results of tropospheric chemistry projects supported within the third framework program. Air Pollution Research Rep. 62, EUR 17769 EN, 68 pp. [Available from the European Commission, DG XII/D, Rue de la Loi 200, B-1040 Brussels, Belgium.].
Derwent, R. G., and T. J. Davies, 1994: Modelling the impact of NOx or hydrocarbon control on photochemical ozone in Europe. Atmos. Environ.,28, 2039–2052.
EMEP, 1998: Calculations of tropospheric ozone and comparison with observations. EMEP/MSC-W Status Rep. 2/98, 132 pp. [Available from the Norwegian Meteorological Institute, P.O. Box 43-Blindern, N-0313 Oslo, Norway.].
Fortezza, F., V. Strocchi, G. Giovanelli, P. Bonasoni, and T. Georgiadis, 1993: Transport of photochemical oxidants along the northwestern Adriatic coast. Atmos. Environ.,27A, 2393–2402.
Georgiadis, T., G. Giovanelli, and F. Fortezza, 1994: Vertical layering of photochemical ozone during land–sea breeze transport. Nuovo Cimento,17, 371–375.
Guillot, P., and Coauthors, 1979: First European Community campaign for remote sensing of air pollution, Lacq (France) 7–11 July 1975. Atmos. Environ.,13, 895–917.
Kallos, G., V. Kotroni, K. Lagouvardos, and A. Papadopoulos, 1998:On the long-range transport of air pollutants from Europe to Africa. Geophys. Res. Lett.,25, 619–622.
Le Bras, G., 1988: European Community research on air pollution: Atmospheric processes, measurement and transport. European Community Rep. EUR 11590, 62 pp. [Available from the Commission of the European Communities, DG XII/E-1, Rue de la Loi 200, B-1040 Brussels, Belgium.].
Le Bras, G., Ed., 1993: Air quality: Analysis, sources, transport, transformation and deposition of pollutants. European Community Rep. EUR 15016, 117 pp. [Available from the Commission of the European Communities, DG XII/D-1, Rue de la Loi 200, B-1040 Brussels, Belgium.].
Le Bras, G., and G. Angeletti, Eds., 1995: Tropospheric processes and air quality, overview of research and results within the fifth environmental R&D program (1989–92) STEP. European Community Rep. ISBN 92-827-4853-7, 94 pp. [Available from the Commission of the European Communities, DG XII/D-1, Rue de la Loi 200, B-1040 Brussels, Belgium.].
Lyons, W. A., and H. S. Cole, 1973: Fumigation and plume trapping on the shores of Lake Michigan during stable onshore flow. J. Appl. Meteor.,12, 494–510.
Lyons, W. A., and L. E. Olsson, 1973: Detailed mesometeorological studies of air pollution dispersion in the Chicago lake breeze. Mon. Wea. Rev.,101, 387–403.
Lyons, W. A., and H. S. Cole, 1976: Photochemical oxidant transport: Mesoscale lake-breeze and synoptic scale aspects. J. Appl. Meteor.,15, 733–743.
Millán, M. M., B. Artíñano, L. Alonso, M. Navazo, and M. Castro, 1991: The effect of meso-scale flows on regional and long-range atmospheric transport in the western Mediterranean area. Atmos. Environ.,25A, 949–963.
Millán, M. M., B. Artíñano, L. Alonso, M. Castro, R. Fernandez-Patier, and J. Goberna, 1992: Meso-meteorological cycles of air pollution in the Iberian Peninsula, (MECAPIP). Air Pollution Research Rep. 44, EUR 14834, 219 pp. [Available from the European Commission DG XII/E-1, Rue de la Loi 200, B-1040, Brussels, Belgium.].
Millán, M. M., R. Salvador, E. Mantilla, and B. Artíñano, 1996: Meteorology and photochemical air pollution in southern Europe: Experimental results from EC research projects. Atmos. Environ.,30, 1909–1924.
Millán, M. M., R. Salvador, E. Mantilla, and G. Kallos, 1997: Photooxidant dynamics in the Mediterranean basin in summer: Results from European research projects. J. Geophys. Res.,102, 8811–8823.
Munn, R. E., 1966: Descriptive Micrometeorology. Academic Press, 245 pp.
Newcomb, G., and M. M. Millán, 1970: Theory, applications and results of the long-line correlation spectrometer. IEEE Trans. Geosci. Electron.,8, 149–157.
Noll, K. E., and T. L. Miller, 1977: Air Monitoring Survey Design. Ann Arbor Science, 296 pp.
Orciari, R., and Coauthors, 1998: Vertical evolution of photochemical ozone over greater Ravenna. Ann. Chim.,88, 403–411.
Pielke, R. A., and Coauthors, 1992: A comprehensive meteorological modeling system—RAMS. Meteor. Atmos. Phys.,49, 69–91.
Portelli, R. V., and Coauthors, 1982: The Nanticoke shoreline diffusion experiment, June 1978. Atmos. Environ.,16, 413–466.
Salvador, R., J. Calbó, and M. M. Millán, 1999: Horizontal grid size selection and its influence on mesoscale model simulations. J. Appl. Meteor.,38, 1311–1329.
Sandroni, S., and M. de Groot, 1980: Intercomparison of remote sensors of sulphur dioxide at the 1979 European Community Campaign at Turbigo. Atmos. Environ.,14, 1331–1333.
Seufert, G., Ed., 1997: BEMA: A European Commission project on biogenic emissions in the Mediterranean area. Atmos. Environ.,31 (Special Issue), 1–255.
Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic, 666 pp.
Whiteman, C. D., 1990: Observations of thermally developed wind systems in mountainous terrain. Atmospheric Processes over Complex Terrain, Meteor. Monogr., No. 45, Amer. Meteor. Soc., 5–42.
Ziomas, I. C., S.-E. Gryning, and R. D. Borstein, Eds., 1998: The Mediterranean campaign of photochemical tracers–transport and chemical evolution (MEDCAPHOT–TRACE): Athens, Greece 1994–1995. Atmos. Environ.,32, 2043–2326.
Maps of southern Europe and the Iberian Peninsula showing the modeling grids used and the major orographic features. The surface pressure map (hPa) at 1800 UTC 27 Jul 1989 is shown in (a). The lower part (b) shows the east–west section used for Fig. 5a and the flight track and reference soundings points F1, F2, M1, and M2 used for Figs. 6 and 7.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Air quality monitoring network in the Castellón province of Spain within grid 3 (dashed), showing the transect for Fig. 4a and the flight paths for Figs. 8 and 9. Graphs (a) and (b) show ensemble averages of the O3 daily cycles in Jul for stations (underlined) located at various altitudes and distances inland.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Ensemble averages of the O3 diurnal cycles for each month in 1997 at six stations located at various altitudes and distances from the Spanish Mediterranean coast (Fig. 2).
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
(Continued)
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Four stages in the development of the combined sea-breeze and upslope winds (anabatic winds) as modeled for 27 Jul 1989 and represented on grid 3. The upper graphs show the modified wind field (see text), and the lower graphs show the surface streamlines (at 57.3 m) and the transect for the upper figure. The topography, as used by the model with 2-km grid resolution, is represented at 500-m height intervals. In graphs b, c, and d, the convergence lines marking the edge of the combined upslope and sea breezes have been highlighted.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Modeling results on grid 1. (top) Vertical wind component (cm s−1) along the east–west transect shown in Fig. 1. Solid lines represent ascent, and the dashed lines subsidence. (bottom) Surface streamlines for 1600 UTC 27 Jul 1989. The main convergence lines have been highlighted.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Instrumented flight at 0447–0553 UTC 27 Jul 1989. (top) Vertical O3 structure above the sea along 450 km of the Spanish east coast. (bottom) Temperature profiles along (a) leg 9 and (b) leg 13 of the flight (highlighted) over points F1 and F2 in Fig. 1b.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Instrumented flight at 0600–0710 UTC 27 Jul 1989. (top) Vertical O3 structure above the coastal plains along 450 km of the Spanish east coast. (bottom) Temperature profiles along (a) leg 1 and (b) leg 10 of the flight (highlighted) over points M2 and M1 in Fig. 1b.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
The NO and O3 plumes above the coastal plain around the city of Castellón at 1142–1223 UTC 27 Jul 1989 at ∼300-m average aircraft height above the ground. The flight path is marked by a dotted line, shown also in Fig. 2. The length of the plumes is approximately 20 km.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
The NO and O3 plumes inland from the city of Castellón at 1509–1655 UTC 27 Jul 1989 at 400–500-m average aircraft height above the ground. The dotted line marks the flight paths, shown also in Fig. 2.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
(top) Measured wind field at aircraft height (400 m above ground) at 1509–1655 UTC 27 Jul 1989, and (bottom) modeled streamlines at 346.5 m for approximately the same area (grid 3). The convergence lines marking the edge of the combined upslope and sea breezes have been highlighted. The mountaintops are represented by black triangles. For comparison purposes, three of the monitoring sites shown in Fig. 2 (Grao, Cirat, Valbona) are marked by bullets.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Schematic of the circulations in the coastal regions of the western Mediterranean on a summer day. Letters a–d indicate successive stages in the entrance of the sea breeze and the formation of stratified reservoir layers aloft, and the numbers correspond to typical station sites (section 7). The lower graph, after Derwent and Davies (1994), represents the relationship between net ozone production and the amount of NOx oxidized along the airmass trajectories. The soundings at right illustrate the sinking of the marine boundary layer on 23 Jul 1991 (Grao).
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Schematic of the circulations in the coastal regions of the western Mediterranean on a summer night. The idealized O3 evolution along the path of the draining air is shown at the bottom. Stations located high above the coastal plains (No. 5) can remain within the reservoir layers during the night. Soundings at right, from the MECAPIP instrumented flight at 0541–0546 UTC 20 Jul 1989, illustrate the stratification over the sea 30 km offshore.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Normalized O3 cycle and averaged meteorological data at a type-1 coastal site (Burriana) for Jul. The data are from the intensive field campaigns (5–10 days) for the years 1989–91 and 1994–95.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Normalized O3 cycle and averaged meteorological data for Jul for a type-2 site on the coastal plain at the mouth of a valley (Onda). Data for 1996 and 1997 are from the automatic station in operation since 1996.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Normalized O3 cycle and averaged meteorological data at a type-2 valley floor site (Cirat, 38 km inland). The data used are from the intensive field campaigns (5–10 days) in Jul for the years 1989–91 and 1994–97.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Normalized O3 cycle and averaged meteorological data for Jul at a type-3 upper valley site (Valbona, 75 km inland). The graphs combine data from the 1989–91 intensive field campaigns and continuous data for 1994–95.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Ensemble averages of the O3 cycles for Jul 1997 for stations at various altitudes and distances inland from the Spanish east coast.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Normalized O3 cycle and averaged meteorological data for Jul at a type-3 upper valley site (Villafranca). The data are from the automatic station in operation since 1995.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Normalized O3 cycle and averaged meteorological data for Jul at a type-4 mountain ridge site (Corachar). The data are from the automatic station in operation since 1995.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Ensemble average of the O3 cycles for Jul 1996–1997 for a type-1 coastal site (Ermita) at 20-m altitude and a type-5 high coastal site (Penyeta) located at some 250 m above the coastal plain 8 km inland.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
(top) Ozone concentrations at the Penyeta and Ermita sites for 16–18 Jul 1996. On these days the Penyeta site remains within the lower reservoir layers during the night. (bottom) Meteorological data, wind direction, and speed are from the Ermita site.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Same as Fig. 21 but for 28–30 Jul 1997. On these days the surface layer reached the Penyeta site during the night.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Normalized O3 cycle and averaged meteorological data for Jul at a type-5 high coastal site (Penyeta). The data are from the automatic station in operation since late 1995.
Citation: Journal of Applied Meteorology 39, 4; 10.1175/1520-0450(2000)039<0487:OCITWM>2.0.CO;2
Monitoring sites used in this study. Coordinates are on the Universal Transverse Mercator (UTM) grid, Zone 30. MSL indicates above mean sea level.
