Sampling was conducted in Yoff, located at 14.75389˚N and 17.46778˚W. This area, with a population of 120,000 covers approximately 15 km² and includes a hospital, a military airport, hotels, and printing and paint factories. The sampler was installed 2 km from the coast, beside a road, directly opposite printing works and paint factories. Each sampling session lasted approximately 24 hours. The sampling site is shown in Figure 1 .

The collected particle samples effectively represented a larger urban area, as the air masses, regardless of direction, reached the collection points undisturbed. A low-volume sampler from the University of GAND was used for sample collection. Maenhaut [20] provide a detailed description of this sampler. Briefly, it operates at a flow rate of 16 L/min and collects PM_2.5 and PM₁₀ using two successive 47 mm diameter filters. Particles larger than 10 μm were removed using a PM₁₀ pre-impactor stage, positioned upstream of the stacked filter cassette. The sampler uses a diaphragm vacuum pump, enclosed in a special casing, and equipped with a needle valve, vacuum gauge, flow meter, volumetric counter, time switch (for interrupted sampling), and hour meter to regulate air intake.

Nucleopore polycarbonate filters (47 mm diameter, 0.4 μm pore size) were used to collect PM_2.5 [20] . To prevent hydration and contamination, the filters were placed in a desiccator after collection. Gravimetric quantification was performed using Sartorius microbalances (Secura and Quintix 26) with a precision of 10⁻⁵ g. The average of three measurements was used to determine the filter weights, provided that variations were below 0.5%.

A yearly measurement campaign was organised in 2018 and 2019 to gather PM_2.5 samples. Sampling took place twice a week, including both weekdays and weekends. This sampling protocol was chosen to maximise the diversity of contributions from various sources, a crucial factor for optimising the performance of source attribution models. During the 2018-2019 period, numerous samples were collected, and 69 PM_2.5 samples were selected for this study. Wind direction and speed data were obtained from the Air Quality Management Centre of the Directorate of Environment and Reserved Buildings [17] .

The Gaussian model follows the laws of normal statistical distribution. Resolutions are represented in the vertical, crosswind, and downwind directions using a system of three-dimensional axes. Weighted by the wind speed at the emission site, the concentration of a pollutant is proportional to the emission rate from the source. The standard deviations of the Gaussian distribution function, primarily determined by air stability, local turbulence, and the distance travelled downwind, define the dispersion of a pollutant. In general, the model’s axis is oriented to correspond to the predominant wind direction. The Gaussian distribution equation is presented as follows [21] :

$C_{x,y,z}=\frac{Q}{2\pi u{\sigma }_y{\sigma }_z}\left\{\exp \left[-\frac{{\left(h-z\right)}^2}{2{\sigma }_z^2}\right]+\exp \left[-\frac{{\left(h+z\right)}^2}{2{\sigma }_z^2}\right]\right\}\left\{\exp -\left[-\frac{{\left(y\right)}^2}{2{\sigma }_y^2}\right]\right\}$ (1)

where:

$C_{x,y,z}$ = Pollutant concentration as a function of downwind position $\left(x,y,z\right)$;

Q = mass emission rate in g.s^-1;

u = wind speed in m.s^-1;

${\sigma }_y$ = standard deviation of pollutant concentration in the y (horizontal) direction;

${\sigma }_z$ = standard deviation of pollutant concentration in the z (vertical) direction

y = distance in horizontal direction;

z = distance in vertical direction;

h = effective stack height.

The coefficients ${\sigma }_y$ and ${\sigma }_y$ are functions of the downwind distance x. These coefficients are determined using Equations (2)-(4):

${\sigma }_y=465.11628x(\tan \theta )$ (2)

where:

$\theta =0.017453293\left(c-d.\ln \left(x\right)\right)$ (3)

${\sigma }_z=a{x}^b$ (4)

x is in kilometers ${\sigma }_y$ and ${\sigma }_z$ are in meters; a and b depend on x.

The stability classes provided the foundation for establishing the values a, b, c, and d [22] . The Davidson-Bryant plume rise formula, given in equation 5 [23] , was used to calculate the effective emission height of the source:

$\Delta H=d{\left(\frac{V_s}{u}\right)}^{\frac{1}{4}}\left(1+\frac{\Delta T}{T_s}\right)$ (5)

where:

$\Delta H$ = the rise of the plume above the stack;

d = the inside stack diameter;

$V_s$ = stack gas velocity;

u = wind speed;

$\Delta T$ = the stack gas temperature minus the ambient air temperature (°K);

$T_s$ = the stack gas temperature (˚K).

The Python software was used to create the program. The dispersion of PM_2.5 released by stationary sources under various wind and atmospheric stability conditions was simulated using the Gaussian model. Basic parameter setup, source and aerosol customisation, Gaussian model-based pollutant concentration calculations, and result visualisation are all included in the code structure. We used a distance of 300 metres to study the local dispersion of PM_2.5 contaminants in Yoff. Based on the values given in Table 1 , the simulation of PM_2.5 dispersion at Yoff is carried out, taking into account an emission rate of 2603 Kg.s⁻¹ and a wind speed of 16.31 m.s⁻¹. The situation, therefore, corresponds to a neutral condition (D) in terms of Pasquill’s stability category. The following stability values were used to compute ${\sigma }_y$ and ${\sigma }_z$: 𝑎 = 34.459, 𝑏 = 0.86974, 𝑐 = 8.3330, and 𝑑 = 0.72382 [24] .

The PM_2.5 concentration levels, wind speed, and wind direction at Yoff throughout the 2018-2019 study period are summarised in Table 1 .

Table 1 shows that PM_2.5 concentrations ranged between 148 and 495 µg.m⁻³ in Yoff. Similar PM_2.5 concentrations were found when our study was compared with earlier research [25] . The high PM_2.5 values obtained in Yoff are due to the contribution of traffic and industrial emission sources [26] .

Our Gent stacked filter unit sampler was positioned two meters above the ground during our campaign. Equation (5) was used to determine the pollution plume’s effective height, which was a very low 0.000035 m. For this reason, the dispersion of PM_2.5 pollutants was simulated at a height of two meters.

To analyse the wind direction distribution in Yoff during the 2018-2019 period, Figure 2 shows the wind rose.

The wind rose was created using R Openair software. Figure 2 shows the distribution of wind directions and speeds. The predominant wind direction is South-South-East (SSE), with wind speeds ranging from 18 to 20 m/s recorded 15% of the time. The West-Northwest (WNW) direction follows, with similar wind speeds, but only for around 13% of the time. Wind speeds in the westerly direction range from 20 to 22 m/s. Wind speeds exceeding 22 m/s are recorded, particularly in the N, NNW, and WSW directions.

The aforementioned criteria (see section 2.3) were taken into account when simulating the dispersion of PM_2.5 pollutants at Yoff. Additionally, the dispersion of PM_2.5 from the Yoff source (14˚45′14″N, 17˚28′04″W) was accurately visualised using OpenStreetMap. Figure 3 illustrates this scenario.

A black circle, symbolising the limit of the PM_2.5 dispersion area, surrounds a red dot on the map, indicating the pollution source along the road. While concentrations in the more polluted red and orange zones are higher, those in the yellow zones are closer to the average (312.37 µg.m⁻³).

Due to sea salts and secondary sulphur compounds, which are recognised as key sources of pollution [25] , the N and NNW directions, including the sea, Yoff Bay, and the ‘Yoff Tangor’ market, are characterised by high levels of PM_2.5 [25] . In this scenario, customers and traders at the fish market are particularly exposed to these PM_2.5 pollutants.

High concentrations are also found in the W and WSW directions due to heavy traffic. The Philippe Madelaine Senghor Hospital, the L’Océan Hotel, and the military airport are all located in this area, exposing medical personnel, patients, military personnel, and hotel guests to these PM_2.5 pollutants.

Beyond Yoff, the pollutant dispersion continues south (S) towards the North clearance, SSE towards the commune of Grand Yoff, and finally reaches the Parcelles Assainies in an East-Northeast (ENE) direction, with winds of 14 to 16 m/s.