2.1 Target area

The city of Bolzano ($\mathrm{258}\mathrm{m}\mathrm{a}.\mathrm{s}.\mathrm{l}.$, northeastern Italian Alps), where about 108 000 inhabitants live, is the most populated town of South Tyrol. As shown in Fig. 1, the city lies in a wide basin at the junction of the Adige Valley (S and NW) with the Sarentino Valley (N) and the Isarco Valley (E). The area is characterized by a highly complex mountainous topography, with very steep sidewalls and many surrounding crests exceeding $\mathrm{1200}\mathrm{m}\mathrm{a}.\mathrm{s}.\mathrm{l}.$ Such a morphology deeply affects the local meteorology. Indeed, the elevated ridges allow at lower levels the onset of wind regimes mainly dominated by thermally driven local circulations, sometimes associated with the development of ground-based temperature inversions. Moreover, especially under fair-weather conditions, the nocturnal wind regime inside the basin is mostly dominated by drainage winds from the Isarco Valley, accelerating at the outlet of the valley and spreading into the Bolzano basin like a valley exit jet, with peaks of intensity exceeding 12 m s⁻¹ (Tomasi et al., 2019). As it is well documented in the literature (e.g., Whiteman, 2000; Zardi and Whiteman, 2013), thermally driven valley exit jets develop when a narrow valley abruptly joins the adjacent plain. Indeed, the Isarco Valley displays these characteristics (see in Fig. 1 the stretch from Barbiano Colma (WS12) to the outlet).

The waste incinerator is 2 km southwest of Bolzano, in the Adige Valley (label 1 in Fig. 1), and became operative in July 2013. The plant has a maximum treatment capacity of 130 000 t yr⁻¹ and was designed to treat the exhaust smoke along a three-stage system, before it is released in the atmosphere at $\mathrm{60}\mathrm{m}\mathrm{a}.\mathrm{g}.\mathrm{l}.$, with a temperature of 413 K and at a rate of 10⁵ Nm³ h⁻¹.

2.2 Experimental setup

2.2.4 Sampling strategy

Figure 2Vacuum-filled glass bottle (a) and polyvinyl fluoride (PVF) bag (b) used to collect samples of ambient air. These pictures were taken during the tracer releases of BTEX. On the right of the vacuum bottle, the incinerator of Bolzano can be recognized.

Download

During each release of SF₆, 14 sampling teams were located in the field. A crucial issue in the design of the sampling strategy was represented by the definition of the sampling-point positions and their scheduling, i.e., starting time and duration. Given the complex orography of the study area and its related heterogeneous meteorological fields, the samplers distribution could be neither geometrically designed (as for example with concentric circles around the incinerator stack, e.g.,  Hanna et al., 1986) nor fixed on the basis of the mean wind flow. For these reasons, seven sampling teams were located in the main residential areas, where most of the receptors are concentrated, while the remaining teams were placed in the surroundings of the incinerator, in agreement with the ground concentrations predicted by an operational modeling chain composed of both meteorological and air quality models, specifically set up for the present field campaign. A detailed description of this modeling chain is beyond the purposes of the current paper and will be presented in future related works. Here, its structure and operational use during the experimental campaigns is briefly reported.

The meteorological field inside the Bolzano basin was simulated by means of the Weather Research and Forecasting (WRF) model (Skamarock and Klemp, 2008) through four nested domains, with an increasing horizontal resolution of 13.5 km (northern Italy), 4.5 km (northeastern Italian Alps), 1.5 km (Trentino-Alto Adige) and 0.5 km (Bolzano basin). Moreover, in the innermost domain, observational nudging was adopted, by assimilating all the available data described in Sect. 3.1. Then, the CALMET model (California Meteorological model; Scire et al., 1999) was used to increase the horizontal resolution of the meteorological field simulated by the WRF model up to 200 m, whereas the CALPUFF model (California Puff model; Scire et al., 1999) was adopted to simulate the dispersion processes and the ground concentrations within the study area.

The modeling chain was used

1. 48 h before the experiment to pinpoint a first-guess distribution of the samplers on the basis of the predicted meteorological conditions and fallout areas and

2. right before each release, in nowcasting mode, to adjust (if needed) the distribution of the samplers with the updated model output (driven by the most recent assimilated meteorological measurements) and to define the starting time and duration of each sampling.

The resolution adopted for the numerical weather prediction simulations (i.e., 500 m) can fall, in principle, into the so-called gray zone or terra incognita, i.e., those scales where turbulence is neither subgrid nor fully resolved but rather is partially resolved. However, this rather high resolution was necessary to adequately capture the spatial variability of meteorological fields in the Bolzano basin. The CALMET model was used as a physically based interpolator to further increase the horizontal resolution from 500 to 200 m.

3.1 Meteorological measurements

During BTEX, the meteorological field within the study area was monitored by means of 15 surface weather stations (WS in Fig. 1), 1 microwave temperature profiler (MTP; label 3 in Fig. 1), 1 sodar (label 2 in Fig. 1) and 1 Doppler wind lidar (label 4 in Fig. 1).

3.1.1 Surface weather stations

Table 1Surface weather stations (WSs) adopted to monitor the meteorological field during BTEX along with their position (V: valley floor station; S: valley sidewall station), coordinates, altitude and available measurements: air temperature (T); relative humidity (RH); rain (R); atmospheric pressure (P); wind direction (dir); wind intensity (U); wind gust (U_g); global radiation (GR); and sunshine duration (SD). The first column indicates the label of the station shown in Fig. 1.

Download Print Version | Download XLSX

The 15 surface weather stations are part of the network operated by the Weather Service of the Autonomous Province of Bolzano and are partly located on the valley floors and partly on the sidewalls. These stations, listed in Table 1 and reported in Fig. 1, record every 10 min the following meteorological quantities: air temperature and humidity at $\mathrm{2}\mathrm{m}\mathrm{a}.\mathrm{g}.\mathrm{l}.$, average wind intensity and direction and wind gusts at $\mathrm{10}\mathrm{m}\mathrm{a}.\mathrm{g}.\mathrm{l}.$, rainfall, atmospheric pressure, global radiation, and sunshine duration.

3.1.2 Microwave temperature profiler

Table 2Technical characteristics of the microwave temperature profiler (MTP-5HE) installed at the airfield of Bolzano.

Download Print Version | Download XLSX

The atmospheric thermal structure inside the basin is constantly monitored by means of an MTP-5HE passive microwave radiometer (manufactured by ATTEX, Russia). The instrument determines the vertical temperature profile at 11 elevation angles (from 0 to 90^∘), by measuring the brightness temperature of the molecular oxygen at a frequency of 60 GHz. The device retrieves the air temperature from the measured brightness temperature by solving a Fredholm equation of the first kind (Gaikovich et al., 1992; Troitskii et al., 1993) and interpolates the obtained profile along 21 vertical levels, 50 m spaced. Further information on the functioning of this device can be found in Westwater et al. (1999), Scanzani (2010), Ezau et al. (2013) and Wolf et al. (2014).

The MTP-5HE used during BTEX (technical specifications in Table 2; label 3 in Fig. 1) is operated by the Physical Chemistry Laboratory of the Environmental Protection Agency of the Autonomous Province of Bolzano and is installed at the airfield of Bolzano (46.46^∘ N, 11.32^∘ E; 238 m a.s.l.), on the roof of a 12 m high building. This device provides the vertical temperature profile from 250 to $\mathrm{1250}\mathrm{m}\mathrm{a}.\mathrm{s}.\mathrm{l}.$ every 10 min.

Figure 3A view of the roof of the incinerator of Bolzano during the experimental campaign with the MFAS mini-sodar (a) and the detail of the antenna (b).

Download

3.1.4 Doppler wind lidar

The interaction between the valley exit jet of the Isarco Valley and the smoke of the incinerator can strongly affect the pollutant dispersion scenarios within the Bolzano basin. Therefore, a dedicated measurement campaign was performed to monitor this flow, and a Doppler wind lidar (label 4 in Fig. 1) was installed, from 9 January to 5 April 2017 on the roof of a public building at $\mathrm{18}\mathrm{m}\mathrm{a}.\mathrm{g}.\mathrm{l}.$, at the outlet of the Isarco Valley (Fig. 4).

Figure 4The WINDCUBE 100S Doppler wind lidar from Leosphere (France) installed on a roof at $\mathrm{18}\mathrm{m}\mathrm{a}.\mathrm{g}.\mathrm{l}.$ In the background is the outlet of the Isarco Valley.

Download

The Doppler wind lidar was a WINDCUBE 100S manufactured by Leosphere (France). This device probes the atmosphere with coherent beams of pulsed, eye-safe electromagnetic waves in the near-infrared domain (wavelength 1.54 µm) and returns measurements of the wind speed component along each line of sight with physical resolutions of 25, 50, 75 or 100 m, up to 319 gates. In the framework of BTEX, the vertical profile of the wind was measured every 5 s by means of the Doppler beam swinging (DBS) technique along 110 vertical levels, 10 m spaced (from 335 to $\mathrm{1385}\mathrm{m}\mathrm{a}.\mathrm{s}.\mathrm{l}.$) and with a vertical resolution of 25 m. These profiles were hourly averaged and assimilated by the modeling system to nowcast both the meteorological variables and the tracer dispersion scenarios.

3.2 Tracer concentration measurements

After each release, samples of ambient air were collected by means of two different technologies: vacuum-filled glass bottles and polyvinyl fluoride bags (hereafter PVF bags). Bottles with a capacity of 1 L were filled by means of a valve, whose calibrated nozzle allowed a constant inflow, thus determining the time required to fill the bottle. In the framework of BTEX, nozzles were calibrated to fill a bottle in 20 or 60 min. The filling of the PVF bags was instead regulated by a suitable air pump controlling the inflow speed.

Different from the online monitoring of SF₆ at the incinerator stack, a much higher sensitivity was required for the detection of the tracer concentrations in the samples of ambient air collected in the surroundings of the plant. Accordingly, a triple quadrupole gas chromatography mass spectrometer (TSQ 8000 EVO from Thermo Scientific) with chemical ionization with methane as the reagent gas was used in the laboratory analyses. The obtained sensitivity allowed us to measure down to background levels. Nevertheless, to avoid the influence of the background (see Sect. 2.2.1), SF₆ concentrations in the collected samples were quantified at levels above 30 pptv.

5.1 Weather conditions

Releases of SF₆ were performed on 14 February 2017, when over northern Italy a high-pressure system was present with very weak synoptic winds. The evolution of the synoptic conditions from 13 February 00:00 UTC to 15 February 00:00 UTC is reported in Fig. 5, in terms of maps of geopotential height at 500 hPa as simulated by the reanalyses of the Climate Forecast System (CFS) model (source: https://www.wetterzentrale.de/). On 13 February, the weakening of a low-pressure system above the Iberian Peninsula induced southwestern synoptic winds blowing over the study area, which channeled into the valleys. Especially during the afternoon and the evening of 13 February, these synoptic winds interacted with the local circulations by strengthening the up-valley winds, as observed by the surface weather stations (Fig. 6a) and by both the sodar and the Doppler wind lidar (Fig. 7b–e). As a further effect, the synoptic winds transported moist air into the valleys. Indeed, the distributions of absolute humidity shown in Fig. 6c display an increase of the moisture content in the afternoon of 13 February at all the weather stations used in BTEX (Table 1).

During the night of 13 and 14 February, the moist air produced low-level clouds that covered the sky above the study area and inhibited the radiative cooling of the ground. Indeed, as confirmed by the microwave temperature profiler (Fig. 7a), this night was warmer than the previous one, and no significant ground-based temperature inversions developed. The absence of synoptic forcing instead allowed the development of thermally driven circulations. However, the wind regimes captured by the surface weather stations on the floor of the Adige Valley at Bronzolo (WS4 in Fig. 1) and at Gargazzone (WS11, Fig. 1) appear to be different from the one observed in the Isarco Valley at Barbiano Colma (WS12 in Fig. 1). Indeed, as shown in Fig. 6a and b, the winds observed at Bronzolo (WS4) and Gargazzone (WS11) were weak and with intensities of less than 1 m s⁻¹, while at Barbiano Colma (WS12) the down-valley wind was much stronger, with intensities of around 3 m s⁻¹. In particular, the topographic constraints of the Isarco Valley and of its outlet into the Bolzano basin determined the acceleration of the winds blowing from this valley, thus allowing the development of a thermally driven valley exit jet, whose evolution was observed by the Doppler wind lidar. The time–height plots of the wind intensity (Fig. 7d) and of the wind direction (Fig. 7e) measured by the Doppler wind lidar reveal the valley exit jet blowing from the east after 21:00 LST (UTC+1) with intensities of around 5 m s⁻¹ in a layer about 400 m deep. During the night, the wind became stronger in an increasingly deep layer until the early morning (14 February 09:00 LST), when it reached its maximum intensity, exceeding 13 m s⁻¹. Then, as is typical of these phenomena (Banta et al., 1995; Chrust et al., 2013), the jet abruptly ceased around 11:00 LST. The valley exit jet was also observed at ground level by the weather stations at Bolzano Ospedale (WS8 in Fig. 1) and Caldaro (WS7 in Fig. 1). Indeed, in the morning of 14 February, the weather station at Bolzano Ospedale (WS8, Fig. 6a, b) recorded a strong easterly wind exceeding 5 m s⁻¹. The weather station at Caldaro (WS7, Fig. 6a, b) instead measured a northerly wind of around 3 m s⁻¹ that intensified from 06:00 to 09:00 LST (wind speed above 4 m s⁻¹), in agreement with the behavior of the valley exit jet. The footprint of the valley exit jet can also be observed in the time series of absolute humidity (Fig. 6c) measured by the previously mentioned surface weather stations, i.e., Barbiano Colma (WS12), Bolzano Ospedale (WS8) and Caldaro (WS7). Indeed, in these series, a significant reduction of absolute humidity is observed during the valley exit jet event. Therefore, it seems reasonable to assume that the valley exit jet advected drier air from the upper valleys into the basin. After noon on 14 February, both the sodar and the Doppler wind lidar observed a weak up-valley wind (around 2 m s⁻¹) rising from the south in the Adige Valley. Also at ground level, the surface weather stations (Fig. 6a and b) recorded very weak winds, with intensities of around 1 m s⁻¹.

5.2 Dispersion patterns

The two releases of SF₆ were carried out to investigate the fallout areas of the tracer under different conditions of atmospheric stability and blowing winds, according to the release strategy (Sect. 2.2.2) and the sampling strategy (Sect. 2.2.4) described above. These data can be used to evaluate the spatial patterns of the tracer in the target area and its evolution in 14 different points. Table 3 summarizes the main characteristics of these releases. In particular, Table 3 reports the timetables and the masses of the tracer released, the metadata concerning the incinerator smoke, the number of sampling teams and the number of collected samples. Furthermore, information regarding the coordinates of the sampling points and the collected samples (sampling devices and time resolution) is reported in Tables 4 and 5 for the first and the second release, respectively. During both the releases, instantaneous samples of ambient air were also collected outside the laboratory, located in the Bolzano city center, and immediately analyzed as described in Sect. 3.2. These measurements allowed for monitoring the presence of the tracer in the atmosphere in quasi-real-time, thereby enabling verification of the effectiveness of the releases.

Table 3Summary of the two releases performed during BTEX (VB: vacuum-filled glass bottles; TB: PVF bags).

Download Print Version | Download XLSX

Table 4Geographical coordinates of the sampling points (SP) along with number of samples of ambient air collected during the first release of BTEX (VB: vacuum-filled glass bottle; TB: PVF bags). The numbers after VB and TB indicate the duration in minutes of the sampling.

Download Print Version | Download XLSX

Table 5Geographical coordinates of the sampling points (SP) along with number of samples of ambient air collected during the second release of BTEX (VB: vacuum-filled glass bottle; TB: PVF bags). The numbers after VB and TB indicate the duration in minutes of the sampling.

Download Print Version | Download XLSX

Figure 8 provides a graphical overview of the dataset, by representing the time evolution of the concentrations measured by the sampling teams during each release (gray areas). In particular, in view of describing the spatial patterns of the tracer along the axis of the Adige Valley, i.e., in agreement with the observed wind regime, the sampling points are ordered according to their latitude, i.e., from south (below) to north (above), while the horizontal black line marks the latitude of the incinerator. Each box corresponds to one sample: the length of the box fits the timing and the duration of the sampling, whereas the color corresponds to the measured concentration. Specifically, blue boxes correspond to samples in which the tracer concentration was lower than the detectability limit of the laboratory analyses, i.e., 30 pptv. White areas, instead, correspond to periods without samplings.

The first release of BTEX was carried out in the early morning, in order to evaluate the dispersion of pollutants in a stable nocturnal atmosphere with a light down-valley wind (i.e., blowing from north to south; see for example the weather station at Bronzolo (WS4) in Fig. 6a and b). In this case, the valley exit jet from the Isarco Valley, being confined north of the incinerator within the Bolzano basin, did not interact with the released smoke. These atmospheric conditions suggested the envisaging of a weak dispersion of the tracer and a fallout area south of the incinerator, in the Adige Valley. The tracer release started at 07:00 LST and ended 1 h later, whereas the sampling activity started at 08:30 LST and continued until 10:45 LST. During this release, the 14 sampling teams collected a total of 28 samples of ambient air at the points represented in the maps in Fig. 9. Figure 9 shows the positions of the sampling teams and the tracer concentrations at four different times (see times with arrows in Fig. 8). At the beginning of the sampling period (45 min after the end of the release; Fig. 9a), only one team northeast of the incinerator observed a concentration of SF₆ significantly different from the background (0.03–0.1 ppb), whereas at 09:15 LST the tracer was observed south of the incinerator in the Adige Valley, with concentrations ranging between 0.1 and 0.5 ppb (Fig. 9b). These findings confirm that the transport of the tracer due to the advection by the mean wind dominated over turbulent dispersion, and therefore the fallout of the tracer at surface level was slower. After 09:30 LST, thanks to the heating of the ground and the onset of a weak up-valley wind (WS3 and WS4 in Fig. 6a and b), tracer concentrations increased also in the area surrounding the incinerator (Fig. 9c and d), where a maximum concentration of 1.19 ppb was observed (Fig. 9d). However, consistent with the observed wind speed and direction, the tracer never reached the town of Bolzano.

The second release started at 12:45 LST and ended at 14:15 LST, in order to investigate the dispersion processes under a weakly unstable atmosphere and an up-valley wind rising in the Adige Valley (see WS3, WS4, WS8 and WS11 in Fig. 6a and b). Under these atmospheric conditions, the mixing is usually more efficient, and a faster fallout of the tracer, north of the incinerator, was expected. Therefore, samples of ambient air were collected from 13:10 LST (still during the release) until 16:30 LST. Indeed, a concentration of 11.96 ppb was observed north of the incinerator at 13:15 LST (Fig. 10a), i.e., 30 min after the beginning of the release. At the end of the release (14:15 LST; Fig. 10b), the up-valley wind moved the smoke of the plant northward, and significant concentrations of SF₆ were measured west of Bolzano, inside the basin, while very low concentrations were observed south of the incinerator. After 30 min (Fig. 10c), the tracer spread into the basin, even if high concentrations were still observed west of Bolzano. Background concentrations of SF₆ ranging between 0.2 and 0.5 ppb, as shown in Fig. 10a, were also found, as a consequence of the persistence of the tracer released in the morning.

At the end of this second release a total of 51 samples were collected by means of both vacuum-filled glass bottles and PVF bags.

Figure 8Tracer concentrations measured on 14 February 2017 after each release. Air samples were collected by means of vacuum-filled glass bottles (large rectangles) and PVF bags (small rectangles). In addition, instantaneous samples of air (circles) were analyzed. The sampling points are listed from north to south, while the two horizontal black lines indicate the latitude of the incinerator. The arrows at the top and bottom of the graph indicate reference time stamps (LST) for Figs. 9 and 10.

Download

Figure 9Spatial distribution of the tracer concentrations measured during the release performed in the morning of 14 February 2017. The position of the city of Bolzano is indicated on the map with the label BZ.