The measurements were conducted in Al Samha area (Figure 1), which is located approximately 40 km northeast of Abu Dhabi City at about the midway to

Dubai. The area is surrounded by various contributors of particulate matters, from sources such as desert sand, power plants, an aluminium smelter, and construction activities, in addition to road and sea traffic. Furthermore, sandstorms are a common occurrence across the entire region, especially during the summer months.

The UAE generally has a subtropical and arid climate, being hot, humid and very dry during summer (April to September), and becoming cooler with occasional rainfall during the winter season (October to March) [22] .

In this study, TSP , PM₁₀, PM_2.5 mass concentrations and Particle Number (PN) were simultaneously measured using a Grimm aerosol spectrometer (Grimm Aerosol Technik GmbH, Germany, model EMD 365), which is also equipped with weather sensors (LufftGmbH, Germany, model WS600) to jointly monitor meteorological parameters such as wind speed, wind direction, relative humidity and temperature. The mass concentrations of the coarse particles (PM_2.5-10) were calculated as the difference between PM₁₀ and PM_2.5 concentrations. Concurrently, PM₁₀ mass concentrations were measured continuously using a beta attenuation monitor (Environment S.A., France, model MP101M). For comparison and verification purposes, TSP and PM_2.5 daily levels were also gravimetrically determined by collecting of some random samples.

The spectrometer (EDM 365) is designed on the principle of orthogonal light scattering, where air containing multiple particle sizes passes through a flat laser beam. The scattered signal is collected at approximately 90˚ to the beam by a mirror and is detected by a high speed photodiode. Each signal is then counted and classified into different size channels by an integrated pulse height analyzer. Eventually, these counts are converted to a mass distribution using the density factor established for urban environments. The EDM 365 utilizes a diffusion dryer to avoid condensation during measurement, which is activated when the relative humidity exceeds 70%. In the beta attenuation monitor, the sampling stream is slightly heated to avoid water condensation, and the air sample is sucked at a constant flow rate (16.7 L/min) from PM₁₀ size-selective inlet and pulled through a filter to deposit particles. At the end of a predefined hourly sampling cycle, the loaded filter is positioned between a carbon 14 beta source and a Geiger-Mueller detector to determine attenuation of the beta ray signal which is directly proportional to the mass of dust accumulated on the filter.

Finally, a particulate sampler (Environment S.A., France, model MP162) was used to collect daily random samples of TSP and PM_2.5, where an air sample is drawn for 24 hours at a constant flow rate of 16.7 L/min through a size-selective inlet (TSP or PM_2.5) and then collected on a 47 mm filter membrane. The filters were conditioned and weighted prior and after sampling to determine net weight gain due to the collection of sample and eventually estimate the concentration.

Air quality standards for suspended PM have been established by different entities in order to protect public health and the environment (Table 1). These standards identify the maximum acceptable concentrations in ambient air, which should not be exceeded during a specified time interval. In this study, the UAE standards were used to assess the daily concentrations of TSP and PM₁₀, while the PM_2.5 daily limit of 35 µg/m³ was also consulted since it is widely applied in many countries such as the Kingdom of Saudi Arabia (KSA), United States (USA) and others.

*Based on a percentile value.

All statistical analyses were performed using Microsoft Excel and SPSS Statistics. Pearson’s correlation analysis was used to determine the linear correlations between the measured parameters, where the existence and strength of the relationship is assessed based on the correlation coefficient (r) as follows: negligible if r < 0.19, weak if r is between 0.2 and 0.39, moderate if r is between 0.4 and 0.59, strong if between 0.6 and 0.79, and very strong if r > 0.8 [23] .

Descriptive statistics of the hourly concentrations obtained throughout the study period are summarized in Table 2. Based on mean value, TSP was approximately 2.2, 3.8 and 5.3 times greater than PM₁₀, PM_2.5-10 and PM_2.5, respectively; while PM₁₀ was higher than PM_2.5-10 by a factor of 1.7 and PM_2.5 by a factor of 2.4. Hourly concentrations of the particulate number varied widely from 34,035 cm⁻³ to 2,085,556 cm⁻³ with a median of 247,431 cm⁻³.

As shown in Figure 2, elevated daily concentrations were observed during the study period for TSP, PM₁₀ and PM_2.5 reaching values of up to 1160 µg/m³, 657 µg/m³ and 252 µg/m³, respectively. Furthermore, about 34%, 15% and 56% of the monitored days had 24-hour average concentrations above the maximum allowable limits of TSP, PM₁₀ and PM_2.5, respectively. These elevated levels might

be attributed to various factors including; increased human activities (e.g. industries and traffic), frequent natural events (e.g. dust storms) and the significant influence of climate conditions (e.g. enhanced formation conditions of secondary particles with high temperatures and intense sunlight in addition to re-suspension of surface dusts in dry conditions).

As it is obvious in Figure 3, the diurnal variation of PM with different size fractions did not follow a similar pattern because of their divergence characteristics as well as the variance of their behaviors in the atmosphere. The lowest levels of TSP and PM₁₀ were observed during the early morning hours between 01:00 - 02:00 am, where the human activities are minimal and the climate is relatively cool, damp, with an eastern low-speed wind. After sunrise, the concentrations remarkably started to rise in conjunction with increased temperature and wind speed and reduced humidity, reaching primary and secondary peaks at 10:00 am and at 13:00 pm for PM₁₀ and one hour later for TSP, and then began to decline. The rise in TSP and PM₁₀ levels might be justified by a longer lifetime of particles at low humidity conditions, re-suspension of surface dusts by higher wind speed, and formation of secondary aerosols at high temperatures. The observed time lag between PM₁₀ and TSP can be explained by the longer time required to transport larger and heavier particles by the wind, in addition to the contribution of the small particles that are agglomerated and coalesced to form greater ones over time. On the other hand, the least level of PM_2.5 occurred at 12:00 noon associated with high temperature, low humidity, and moderate-speed western wind, and then PM_2.5 level increased gradually to reach its peak at 04:00 am. The humid conditions are associated with high levels of PM_2.5 which might be attributed to the role of moisture in forming secondary fine aerosol such as ammonium nitrate through the gas-to-particle conversion. Changes in prevailing wind directions have no noticeable effect on the average diurnal concentrations.

As illustrated in Table 3, the highest mean concentrations were observed on Saturdays for TSP and PM₁₀ and on Sundays for PM_2.5. On the other hand, the lowest levels were recorded on Thursdays for TSP and on Wednesdays for PM₁₀ and PM_2.5. The elevated PM levels during Saturday and Sunday might be attributed to the increased human and industrial activities during the free-time

weekend (Saturday) and the first working day of the week (Sunday). As shown in Figure 4, relatively elevated concentrations were observed over extended time for the days of maximum records (Saturday for TSP and PM₁₀ and Sunday for PM_2.5) as compared with other days.

Monthly variations of PM mass concentrations are given in Table 4. The highest mass mean concentrations were observed for all size fractions in July primarily due to the frequent occurrence of dust storms during this period of time. The pattern of TSP during April to July is consonant with the wind speed pattern, which indicates that there is a notable influence of wind speed on large particulate levels. As expected, the lowest PM levels were recorded during the cool winter season as a result of the humid and occasionally rainy conditions. As presented in Figure 5, non-identical pattern of higher diurnal concentrations of particulate matters was observed during the summer months (April-August) as compared to the winter months (September, November and December).

*Data is not available from September 26, 20:00 to October 24, 14:00 due to power supply failure.

Based on the mean ratio values shown in Table 5, TSP contains nearly 39% of particles with an aerodynamic diameter greater than 10 µm (PM_>10), and the rest (61%) is PM₁₀, which consists of 47% PM_2.5 and 53% PM_2.5-10. These results are inconsistent with the results of Engelbrecht et al. [24] for daily samples collected in the UAE, where the reported ratios of PM₁₀/TSP and PM_2.5/PM₁₀ were 0.71 and 0.37, respectively. The deviations between the obtained results and the above mentioned reported results by Engelbrecht et al. are mainly due to the influence of temporal and spatial variation in PM ambient levels. However, our results are closer to the typical PM_2.5/PM₁₀ ratio of 0.5 that have been documented for urban areas in developing countries [25] , and reported for urban sites in Iran [26] ‎.

As shown in Table 6, very strong to moderate inter-correlations are found between PM of different size fractions. The weak correlations between total particle number and mass concentrations of particulate matter with different sizes indicate that the number of particles is an inadequate indicator of the mass levels and vice versa. The moderate correlation between TSP and wind speed is noticeable by the influence of wind on the diurnal variations of TSP. Figure 6 indicates that the highest average concentrations of airborne particulate are associated with wind coming from the south and south-southwest directions, where heavy highway traffic flow exists (re-suspension of surface dusts).

Measurement results of PM₁₀ mass concentrations obtained by the light scattering spectrometer were compared with the concentrations measured by the beta attenuation monitor, as shown in Figure 7. Consequently, a correlation coefficient of r = 0.73 (coefficient of determination r² = 0.539) indicates a strong linear relationship between the two measurement techniques. However, the PM₁₀ mean value of the spectrometer results was 18% higher than its counterpart obtained by the beta attenuation monitor with the presence a statistically significant

difference between the two data sets. This difference can be explained by the fact that both techniques may misestimate the actual concentrations [27] [28] , and therefore their results need to be corrected by applying site specific and seasonal correction factors developed in line with the standard reference methods [29] ‎ which is beyond the scope of this study. However, TSP and PM_2.5 results measured by the spectrometer are perfectly correlated (r > 0.995) with its counterparts obtained by gravimetric analysis of randomly collected samples as shown in Figure 8, noting that the spectrometer overestimated the TSP and underestimated the PM_2.5 concentrations of the collected sample, especially at the high levels.