Water samples were collected for this study in two field campaigns on 15 January 2019 and 10 September 2020 along the coast of Dubai (Figure 1a) on the Arabian Gulf. In the first campaign, coastal water samples were collected from 80 locations on the coast, and in the second campaign, samples were collected from 10 locations. Samples were collected from various beaches along the Dubai coast: Jumeirah Beach, Mamzar Park Beach, Salt Beach, Kite Beach, FS Beach, JBPMTS Beach, JBR Beach, Mercato Beach, Soufouh Beach, and Black Palace Beach. Figure 1b shows a sample Sentinel-2 image of the study area, acquired on 13 January 2022.

Figure 1.

Coastal waters MPs sampling locations along the coast of Dubai (a) and Sentinel-2 image of the study area acquired 13 January 2022 (b)

One litre of coastal water was collected from each location using glass beakers. The samples were collected during high tide and labelled according to the sample location's coordinates, date, and sampling time. Sampling was carried out at the surf (or swash) zone, which is believed to be where the MPs are generated [25]. The surface water grab samples were collected and transferred to the lab for further analysis.

The samples were then analyzed to determine the levels of MPs in water using the NOAA method for MPs enumeration in the marine environment [26]. Pretreatment of the samples was performed to eliminate organic and inorganic materials. The samples were rinsed with distilled water to remove salts and then passed through stainless steel mesh sieves to separate and discard any plastic particles greater than 5 mm in size. In the present study, there were no solids greater than 5 mm. Solids collected in the 0.3 mm sieve were transferred to a tared beaker using a spatula and were placed in a drying oven for 24 hours at 60 °C to ensure sample dryness, followed by measurement of the mass of the dried solids.

Wet Peroxide Solution (WPO) was then added to the solids, which were collected in the beaker, followed by adding 20 mL of hydrogen peroxide and heating the mixture to 75 °C on a hotplate. The resulting mixture was filtered using a vacuum filter and 0.45 µm pore size standard filter paper. The filter paper was transferred to a known mass of Petri dish using metal forceps for examination under the microscope with 40 X magnification. The final mass of the MPs on the filter paper was then measured. Subsample of the particles were subjected to final FTIR confirmation as previously described in another study [10].

The reference spectral signatures of the MPs produced in the laboratory were identified through an experimental setup prepared in the laboratory, as shown in Figure 2. A correlation was established by analyzing the spectral reflectance of water samples with synthetic MPs added to them using a Stellar-RAD® handheld spectrometer. The experimental setup included a spectrometer, glass beaker, and ultrapure water with synthetic MPs added to it. Six types of MPs were used in the experiments, including polyethylene (PE), polypropylene (PP), polyvinyl chloride (PVC), polystyrene (PS), polyurethane (PUR), and polyethylene terephthalate (PET), which are the most common types found in the marine environment [27].

Figure 2.

Experimental setup for developing a correlation

The Stellar-RAD® handheld spectrometer used in this experiment has a wavelength range of 350-1150 nm. Using a spectrometer with such a wavelength range was based on the outcomes of previous research studies. Moroni et al. [28] investigated spectral signatures and the absorption feature characteristics for two polymers, PET and PVC, acquired with a two-linear-spectrometer apparatus. They calculated wavelengths associated with the maxima of reflectance values that have been optimal for both polymers. These points are located at wavelengths of 970 nm, 1090 nm, 1320 nm, and 1580 nm.

The Stellar-RAD® handheld spectrometer is a research-grade spectrometer with a spectral resolution smaller than 1 nm. It was used to measure the spectral reflectance of the MPs in water by measuring the reflectance/emittance of the water with MPs mixed in it and placed in the glass beaker with water. The six types of MPs used in the experiment were created artificially by cutting the plastic sources into smaller than 5 mm pieces. The plastic sources for these six types of common MPs are the following: shampoo bottles from PE, drinking straws from PP, wire and cable sheathing from PVC, egg boxes from PS, shoe soles from PU, and water bottles from PET. All of these MPs are commonly found in coastal water [29].

In the present study, different sizes of plastic materials (5 mm, 2.5 mm, and 1.25 mm) were also used. To simulate real coastal water, which normally includes suspended sediments, different proportions of sand were added to the MPs in the experiment. The experiment was conducted for the six types of MPs mentioned above for three sizes for each type (Figure 3). The spectral reflectance of the clear water was first measured for reference. Literature shows that the TSS value at the surface of deeper water of the Arabian Gulf is about 260 mg/L [30]. Accordingly, for each size and type of the tested MPs, three different amounts of sand were added: 50, 150, and 260 mg to represent TSS in the study area. The used sand was a sediment commonly found in the Dubai coastal waters. For every part of the experiment, three trials were administered to ensure the robustness of the measured spectral reflectance. Therefore, a total number of 234 trials were conducted. These spectral responses were carefully examined and analyzed to develop correlations that will aid in identifying the spectral signatures of the tested types of MPs; however, these specific spectral responses were not directly utilized in the subsequent model development. Having different variables aside from the fixed factors (known volume of water in a glass beaker and a dark room) helped to consider all possibilities and develop a correlation that can help detect MPs. These spectral responses were carefully examined and analyzed to develop correlations that will aid in identifying the spectral signatures of the tested types of MPs; however, these specific spectral responses were not directly utilized in the subsequent model development.

Figure 3.

The artificially created microplastics in the lab experiment

Each type of MP was added to 1 litre of water in the glass beaker. The artificially created MPs were mixed with deionized water in the laboratory inside a glass beaker. The purpose of using glass beakers and rod stirrers instead of plastic is to prevent them from being a source of contamination. The experiment was conducted in a dark place to prevent any interference from any other light source and ensure the accuracy of the results. Spectral responses were measured by the spectrometer for each type and size of MPs used in the experiment. These spectral responses were analyzed to develop the correlation, which was used to identify the spectral signatures of the tested types of MPs. The resulting spectra in these experiments were first smoothed using a 20 nm span average filter followed by derivative spectroscopy analysis to identify absorption features in the mixes created in the laboratory [31]. Then, the spectral reflectance curves for the mixes (water + MPs + TSS) were created.

Sentinel-2 Level-1C imagery was obtained on the two sampling days to develop an empirical spectral model for MPs in coastal water in the study area. Atmospheric correction (considering ozone, water vapour, and cirrus corrections) was applied to the two Sentinel-2 images used in the present study with the sen2cor tool provided by the European Space Agency (ESA) for processing of Sentinel-2 imagery [32]. The two bands used in this study (B5 & B9) were resampled using the sen2cor tool at 20 m resolution.

The laboratory results of the analysis of the coastal water samples showed that all water samples were contaminated with MP particles. However, the contamination varies along different locations and at different beaches. The results have been reported in mg/L and items/L to quantify the extent of MP contamination in the study area. An average of 0.633 and 0.324 mg/L of MPs and 7.85 and 5.30 items/L along the coast of Dubai were found in the two campaigns, respectively. A study in Australia showed that the abundance of MPs in coastal waters is around 2.5 items/L; the higher contamination rate in the study area could be due to the factors discussed earlier [33]. Table 1, Table 2 and Table 3 show the levels of MPs in the samples in mg/L in the two field campaigns. Spectral reflectance analysis in the laboratory of the water samples of the two field campaigns showed absorption peaks in the spectral range 700-875 nm and reflectance peaks in the range 900-1150 nm.

Polymer composition of the subsample was performed using Fourier Transform Infrared (FT-IR) analysis. Analysis showed that different beaches have different types of MPs present. In Jumeirah Beach, PE and PP were present; in Kite Beach, PE and PP were present; Black Palace Beach was mostly contaminated with PP, PE, and PVC, whereas in JBR Beach, PE, PET, and PS were present. As for the Mercato Beach, PS, PET, PP, and PE were present. The densities of the MPs verify the types identified in the samples that correspond to each beach. PE and PP have lower densities (PE = 0.93 g/cm³, PP = 0.83 g/cm³), which is true for Mercato Beach and JBR Beach, which shows low contamination in mg/L as low-density MPs have low weight compared to higher-density MPs. As for the remaining beaches, higher-density MPs are present, reflected in the higher contamination in mg/L as the PVC and PS have higher densities (PVC = 1.16 g/cm³, PS = 1.04 g/cm³). The main source of this contamination could be coastal tourism since all these beaches receive thousands of people every day all year long. Also, all these beaches are close to fisheries and harbours, which is a possible source of these MPs.

Spectral analysis of various MP samples was conducted, taking into account their size, type, and total suspended solids (TSS). In the wavelength range of 700-850 nm, there is an evident relative increase in spectral reflectance with the increase in the concentration of the TSS; however, there is no clear difference in reflectance with the increase in TSS in the mixes beyond 850 nm as the reflectance peaks to about 50%.

The results indicate that varying the size and type of MPs and adding sand in varying amounts did not significantly alter the resulting spectral reflectance. Specifically, it was observed that adding sand did not affect the peak wavelength. The highest spectral reflectance is within the 1000-1200 nm wavelength range for all readings. It indicates that the spectrometer could detect the MPs easily at this wavelength range, which is not affected by the presence of suspended material (Appendix Figures A1, A2, A3, and A4). Figure 4 shows the reference spectral reflectance curves for the considered types of MPs and the specified sand mixes.

Figure 4.

The laboratory-based (reference) spectral reflectance curve for the considered types of microplastics and the specified sand mixes

The sizes of MPs did not affect the outcomes, as the highest spectral reflectance remains within the identified wavelength range for all types and sizes of MPs. The results suggest that the MPs can be detected in the wavelength range 900-1200 nm (Appendix Figures A10, A11, and A12).

The spectral reflectance of the different types of plastic materials is measured, keeping the size and sand amounts constant. The different types of MPs are detected at a wavelength range of 900-1200 nm, which was identified as the spectral signature of these MPs in the study area. This spectral signature was used to select band B9 of the Sentinel-2 satellite to determine MP concentrations in the coastal water of Dubai (Appendix Figures A5, A6, A7, A8, and A9).

The summary of the laboratory results is presented in Table 4. PET showed a high reflectance in the range 900-1100 nm, PS in 900-1100 nm, PVC in 900-1150 nm, PP in 900-1100 nm and PE in 900-1100 nm. Accordingly, it was determined that the different types of MPs used in the laboratory experiments exhibited the highest reflectance in the wavelength range of 900-1150 nm. On the other hand, the three types have shown the highest absorption in the wavelength range of 700-850 nm. Different types of MPs were detected at the wavelength range 900-1150 nm, identified as the spectral signature of these MPs in the study area. The highest absorption at 700, 800, and 850 nm, and reflectance peaks at around 900, 985, and 1080 nm can be mapped to Sentinel-2 bands B5 (central wavelength: 705 nm) and B9 (central wavelength: 945 nm). Based on this, the highest absorption can be identified in band B5, while the highest reflectance can be detected in band B9. This spectral signature was used to select bands B5 and B9 of the Sentinel-2 satellite to develop spectral models for determining the concentrations of MPs in the study area.

Regression analysis was used to obtain the best-fit model correlating the sampled MP concentrations using the field data and the spectral reflectance of Sentinel-2 image bands B5 and B9 on 29 January 2019 at the sampling locations. Regression analysis was used to establish the correlation between the measured MP concentrations (N=90) and the spectral reflectance at bands B5, B9, B5×B9, and B5/B9. Sixty-five (N=65) data values from field campaign 1 were used to develop the empirical spectral model, and fifteen (N=15) values were used for model calibration. Table 5 shows the investigated Sentinel-2 band combinations and their corresponding R² values.

The best-fit spectral MPs model can convert atmospherically corrected Sentinel-2's reflectance at bands B5 (705 nm) and B9 (945 nm) into MPs (mg/L) with an R² value of 87.30% and RMSE of ±0.015 mg/L. The resulting empirical model is shown in eq. (1) and Figure 5.

$MPS=1.3311(\frac{{\text{ρ}}_{705}}{{\text{ρ}}_{945}})1.803$ (1)

where symbol MPs denotes the concentration of particles by mass in the coastal water in mg/L, and ⍴₇₀₅ and ⍴₉₄₅ are the central wavelengths of bands B5 and B9 of the Sentinel-2 sensor.

Figure 5.

The best-fit spectral MPs model, which combines Sentinel-2's bands B5 and B9

The resulting empirical spectral model in eq. (1) was then applied to the Sentinel-2 image on 29 January 2019 (bands B5 and B9) to obtain the levels of MPs in the water on the coast of Dubai, as shown in Figure 6.

Figure 6.

The spatial distribution of MPs estimated with the model on 29 January 2019; the bars show the sampled and modelled MPs levels at the sampling locations

The figure shows that the concentrations of MPs are higher in the nearshore zone and inland water bodies than in offshore and open gulf water areas. This spatial distribution pattern strongly suggests that human activities in coastal regions play a significant role in the elevated presence of MPs. These findings are consistent with previous studies highlighting the impact of urbanization, industrial activities, and improper waste management practices on coastal MP pollution [34]. This type of pollution monitoring model can be useful to identify the specific source of MPs and the potential management plan of the plastic disposal system.

Identifying hotspots with higher MP concentrations provides crucial information for targeted intervention and mitigation strategies. Coastal management authorities can utilize this information to prioritize and implement measures to reduce the input of MPs into nearshore waters. For instance, implementing stricter regulations on waste disposal and effective waste management practices can help minimize the release of plastic debris into water bodies [35]. Additionally, raising public awareness about the impact of plastic pollution and promoting responsible consumption and recycling habits can reduce the overall load of MPs entering coastal environments [36].

To demonstrate the usefulness and viability, the resulting empirical model was used to obtain the spatial distributions of MPs in the area for September 2018, 2019, 2020, and 2021 (September marks the beginning of the autumn season in the UAE). The spatial distributions of MPs generated with the model in 2018-2021 are shown in Figure 7.

Figure 7.

The spatial distribution of MPs estimated with the spectral model on 11 Sep 2018 (a), 11 Sep 2019 (b), 10 Sep 2020 (c), and 10 Sep 2021 (d)

The MP distributions show high concentrations on the west coast of the Emirate, consistent with the locations of many busy beaches in this part of Dubai and the location of the Port of Jebel Ali, the busiest port in the region. This part of the Arabian Gulf is characterized by a 1-16 m deep bathymetry, west-to-east surface currents, and circulation processes with a seasonal averaged velocity field of about 0.2 m/s [37], which can also explain the resulting spatial patterns of MPs generated by the model. MP concentrations appear lower in 2020 and 2021 than in previous years due to the lack of coastal activities (i.e., beaches) during the pandemic, which decreased the MP concentrations.