Open Access
Add to BibSonomy
Abstract
The exploitation of satellite remote sensing is expected to be a critical asset in monitoring floating and submerged plastic litter in all aquatic environments. However, robust retrieval algorithms still havel to be developed based on a full understanding of light interaction with plastic litter and the other optically active constituents of the atmosphere-water system. To this end, we performed laboratory-based hyperspectral reflectance measurements of submerged macroplastics under varying water clarity conditions (clear – 0 mg/L, moderate – 75 mg/L, very turbid – 321.3 mg/L) and submersion depths. This comprehensive optical dataset was used (i) to relate the plastic-related signal to submersion depth and turbidity parameters, and (ii) to investigate the top-of-atmosphere signal through full radiative transfer calculations. Simulated TOA radiation was used to explore the nominal pixel and spectral requirements based on WorldView-3, Sentinel-2, and Sentinel-3 missions with very high to moderate geo-spatial resolutions. Results showed that plastics remained detectable when submerged in the top ∼1 m of the water column regardless of water clarity conditions. At TOA, uncertainties attached to atmospheric correction were shown to be reasonable and acceptable for plastic detection purposes in the infrared part of the spectrum (> 700 nm). The impact of aerosols on the TOA signal was found to be complex as (i) over large plastic patches. The aerosols produced little impact on satellite observations mostly due to adjacency effects and (ii) optical signature from isolated/small extent plastic patches was critically altered suggesting the atmospheric transmittance should be accurately corrected for in plastic detection algorithms. The sensitivity analyses also revealed that the narrow band widths of Sentinel-3 did not improve detection performance compared to the WorldView-3 coarser band widths. It is proposed that high spatial resolution wavebands such as the pan-chromatic could be advantageously explored for submerged plastic monitoring applications.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Remote sensing detection of plastic litter in the aquatic environment is expected to provide supplementary evidence to help monitor the plastic-related pollution and further investigation on the missing plastic budgets estimated from net trawls and numerical distribution solutions [1]. Furthermore, interest in monitoring strategies with a wide geo-spatial and temporal coverage offered by remote sensing approaches has been rising [1–3]. These direct and indirect observation strategies utilize optical and radar technologies integrated on unmanned aerial vehicles, aircraft and satellites [4–9]. Future operational applications of remote sensing are anticipated to support interdisciplinary demands for the detection, identification, tracking and quantification of plastic litter in the blue planet [10]. To this end, several field and laboratory experiments have revealed diagnostic absorption features in the infrared spectrum from hyperspectral reflectance observations of virgin, washed-ashore and marine-harvested ocean plastics [11–16]. However, knowledge on these spectral properties for submerged plastics at sea has not yet been well reported although it is gaining interest [14,17,18].
The upper layers of the oceans are nowadays monitored by a comprehensive network of scientific satellites [19]. On the other hand, it has been reported that levels of buoyant plastics decreased exponentially with depth and highest abundances were observed in the top ∼1 m to 5 m in open ocean or coastal waters [20–22]. Therefore, the exploitation of the satellite capabilities has to be rooted on the quantitative knowledge of the relationship between light propagation within the top water layer, water clarity and the submerged plastics to identify. Initial experiments involving submerged plastics pointed out that the strong absorption characteristics of pure water itself decreased the magnitude of the expected inherent signal of wet plastics [14,17,23]. Those results can be applied to open ocean waters where water clarity is usually high. Conversely, supplementary studies are needed to explore the optical signatures of submerged plastics for the different levels of water turbidity encountered in coastal areas or inland waters. Based on this knowledge, technical requirements for monitoring plastics might be integrated in future ocean color remote sensing missions thus supporting interdisciplinary environmental studies with a single sensor in addition other promising technologies like synthetic-aperture radar or [5,24,25].
For remote sensing purposes, the optical signature of submerged plastic as observed right above the water surface has to be converted into top-of-atmosphere (TOA) signal to fully characterize the satellite measurements. Thus, the sensor specifications and the requirements retrieval algorithm development could be rigorously established in view of operational monitoring of aquatic plastic litter. Moreover, the atmospheric radiance and transmittance strongly modify the water-leaving radiance from bottom to top of atmosphere [26]. As a result, the optical signature of submerged plastics will be significantly altered when acquired from the satellite device. Atmospheric correction is commonly adopted to correct for those effects with a particular attention paid to correct for the aerosol impacts [27]. Investigation on the interplay between water and atmospheric optical conditions is therefore thought to provide the optimal ways to monitor plastic debris from optical space sensors.
The study herein was conducted with the objective to further scientific evidence-based knowledge about the variations in the spectral characteristics of virgin plastics submerged in water. Knowledge gained from groundwork [14, 15] on reflectance characteristic of submerged plastics was extended by implementing sensitivity analyses based on full radiative transfer solutions. Based on laboratory measurements, the effect of water clarity on the observed reflectances were analyzed with respect to submersion depth and plastic types. The top-of-atmosphere signal was simulated based on the quality-controlled measurements and full radiative transfer computations for different atmospheric conditions including oceanic to continental aerosol types. Based on those TOA hyperspectral reflectances, potentialities of plastic detection were analyzed accounting for the spaceborne sensor responses. Three satellite missions (Sentinel-2, Sentinel-3 and Worldview-3) were simulated as representative sensors with high to moderate pixel resolution relevant to observing small to large spread plastic patches.
2. Material and method
2.1. Samples and spectral measurements
Targets used for the spectral measurements consisted of dry and wet virgin plastic materials. These included an orange placemat and polypropylene (PP) ropes tightly wound on clear sp plates. The PP rope targets ranged in color from blue, white to orange (Fig. 1(a)).
Fig. 1. (a) Example orange polypropylene rope wound on a clear Plexiglas plate and (b) a ∼13 × 13 cm Labsphere Spectralon 99% diffuse reflectance plaque.
A Spectral Evolution (SEV) SR-3501 hyperspectral spectroradiometer with an 8° foreoptic lens was used to measured reflectance from the ultraviolet (UV, 280 nm) to shortwave infrared (SWIR, 2500 nm) spectrum. Spectral resolution of the SEV is 4 nm (280-1000 nm), ≤ 10 nm (1000-1900 nm) and ≤7 nm (2100-2500 nm). The SEV was operated in reflectance mode and relative reflectance was automatically derived by white referencing using a Labsphere Spectralon 99% diffuse Lambertian plaque of size ∼13 × 13 cm (Fig. 1(b)). The light source was 12 V 50 W GY9.5 Original Gilway L9389 halogen lamps in Lowel Omni-Light casing. A total of five pseudo measurements were collected for each target and each measurement was an average of 30 scans.
Measurements of the dry samples were collected above water whilst the wet samples were submerged to fixed depths of ∼0.025, 0.05, 0.09, 0.12, 0.16, 0.32 m followed by a final measurement right above water. We assume that the dry and wet targets observed above water were representative of floating plastics whilst the shallow to deepest targets depict the submerged or suspended plastics in the water column. Water clarity was inferred from the measurements of total suspended particulate matter (SPM), clay collected from Deurganckdok, Port of Antwerp in Belgium. Three water clarity scenarios were low SPM = 0 mg/L, medium SPM = 75 mg/L and high SPM = 321.3 mg/L. Further details on the experimental tank setup, SPM and spectral measurements are reported in open-access [14,17]. The only experimental setup difference from the related study [14] was the primary use of the SEV instrument that extended spectral measurements into the UV spectrum (280-350 nm).
2.2. Spectral data processing
Measured spectra were smoothed using a third order polynomial Savitzky-Golay least-square algorithm at frame length of 31. A simplified splice correction [12] was applied to the data to compensate for typical steps in spectra due the transition between UV- near infrared (∼280-1000 nm) to SWIR-1 (∼1000-1900 nm) and SWIR-1 to SWIR-2 (∼1900-2500 nm). Blank measurements, Rblank, were obtained in absence of any plastic target for the three water clarity conditions (Fig. 2). Note that the state of water conditions was defined by the SPM concentrations, nevertheless, all water conditions were slightly impacted by absorption of dissolved material. It was determined that with the setup there was no interaction or influence of the targeted plastic measurements with the tank walls optical properties. Black material was placed on the tank walls to reduce unwanted signal from background contributing the measured bulk reflectance. The field of view at nadir was at the center of the target measuring less than 0.02 m2 at 0.32 m depth of submersion, suggesting the bulk signal was from the water and the plastic target observed.
Fig. 2. Reflectance measurements of the background scenarios of water clarity as indicated by suspended particulate matter (SPM). Reflectance beyond 1200 nm was negligible.
The above-water measurements of the reflectance, R, were converted into subsurface (z = 0- m) reflectance following the formula [28,29]:
where 0- and 0+ stand for the subsurface (just beneath the air-water interface) and above surface levels, respectively. Following basics in contrast propagation, the relationship between (sub)surface reflectance and upward radiance at the target depth z can be written [30,31]:2.3. Full radiative transfer computations
The measured reflectances were assimilated into radiative transfer computations to derive the top-of-atmosphere (TOA) spectra of each plastic type, submersion depth and water clarity condition. Computations were based on the assumptions that (i) the reflected light by the water surface was assumed negligible in the measured spectrum and (ii) the corresponding water-leaving radiance was Lambertian, namely, the water reflectance was independent of the geometry of observation excluding surface reflected light. Based on these assumptions, radiative transfer computations were carried out through the 6SV-code framework [34,35] where a specific module was implemented to include the following bottom-of-atmosphere (BOA) reflectance:
Three in-orbit satellite missions were selected to exemplify realistic signals as potentially observed from the spectral bands: Maxar Technologies WorldView-3, Copernicus Sentinel-2 and Sentinel-3. Furthermore, these satellite missions represent observation at very high (∼1 m) high (∼10 m) and moderate (∼300 m) geo-spatial nominal pixel resolutions. TOA reflectance of the different bands i, $R_i^{TOA}$, was computed after convolution of the TOA signal with the relevant sensor relative spectral responses (RSR). The computation was carried out based on the hyperspectral radiative transfer simulations of the TOA upward radiance, $L_{sim}^{TOA}$, and downward irradiance, $E_d^{TOA}$, separately [38]:
3. Results
3.1. BOA reference observations
3.1.1. Spectral shapes and magnitude
Measured reflectance of the wet plastics just above water and subsurface at 2.5 cm depth are presented for the three water clarity cases (Fig. 3). Highest reflectance values were observed in all the just above water wet plastics over the UV-SWIR spectrum, consistent with prior studies [14,17,23]. Diagnostic absorption features found in weathered plastics [13,15,23,39] were also salient in the wet above-water samples located at ∼931, 1045, 1215, 1417, 1537, 1732, 2046 and 2313 nm (Fig. 3(a)). The plastics samples produced reflectance spectra with significant differences in the NIR-SWIR part of the spectrum [12]. For instance, the spectral shapes of the orange samples revealed that the PP rope target exhibits stronger absorption features around 1000 nm than the placemat sample where such features are practically undistinguishable. It was also noted the magnitude of the signal and depth of the absorption features was lower in the placemat compared to the PP rope target. In the visible spectrum, the apparent color of each target was consistent with the expected spectral shape of the measured signal. The blue object peaked around ∼465 nm, the white had a general flat shape and the orange had peaked ∼600 nm.
Fig. 3. Spectral reflectance of blue, orange, placemat and white plastic targets at varying water clarity inferred from suspended particulate matter, SPM. Wet samples were observed just above the water surface and the subsurface samples were at 2.5 cm depth below the surface. Vertical dotted lines indicate absorption features at 931, 1045, 1215, 1417, 1537, 1732, 2046 and 2313 nm reported in marine-harvested plastics [23].
Subsurface observations had negligible reflectance in the SWIR starting at ∼1200 nm a large decrease compared to the above water wet samples. Relevant diagnostic optical characteristics about submerged plastics are therefore presumed to be limited, from the UV to ∼1200 nm in clear to very turbid waters. The interaction of the inherent reflectance of the plastics with the water itself resulted in a change in the spectral shape with fewer absorption features that could be revealed in the spectra. All spectra had a consistent decrease in reflectance from 800–1000 nm followed by peak that started to appear between 1000–1200 nm. The white target was characterized by the highest reflectance values from the UV to the near infrared (NIR) whilst orange targets had lowest reflectance value until 600 nm. The blue PP rope target had two peaks centered ∼490 and 860 nm (Fig. 3).
3.1.2. Effects of water clarity and in-water depth of target
To ensure data quality control and to enable extrapolation at any depth, the aquatic optical model Eq. (3) was applied to all the measured spectra. Selected wavebands of the subsurface signal were modelled at various water depth and water clarity scenarios (Fig. 4). A good fit was determined which showed near linear to exponential decrease of reflectance with depth and water clarity. The decrease was pronounced at 1020 nm and was quasi-independent of the water clarity conditions suggesting that water absorption was the key variable that influenced the bulk signal observed from the submerged samples.
Fig. 4. Observed (dots) and modelled (dashed lines) reflectance values of submerged plastics as selected wavebands.
In the clear water scenario, measured above-water reflectance were observed to exhibit a near linear decline in intensity at the deepest point test of ∼0.32 m. As for the very turbid condition, the reflectance values reached a plateau that were independent of depth after the first ten centimeters. However, impact of submersion depth remained sensitive down to 0.3 m depth just before reaching the plateau where plastics became indistinguishable in the moderate water clarity case. A fitting model as applied here is proposed as a useful tool in studies investigating optical properties of submerged targets. Modelling can be applied from observed subsurface submersion (0- m) to the sub-centimeters despite the challenges of the actual practical approach in experimental setups. Measured and simulated spectra of in-water plastics at various water depths show the potential application of such a fitting model (Fig. 5).
Fig. 5. Measured (continuous lines) and simulated spectral reflectance (dashed lines) of submerged plastic targets in varying suspended particulate matter (SPM).
Based on the actual measurements plus simulated data for extra-depths, several spectral modifications with depth can be readily observed. For instance, a smoothing away of the two peaks that were observed in the blue PP rope at 490 and 860 nm (Fig. 3) are smoothed away at depths greater than the measurements at 0.32 m (Fig. 5). Reflectance at 590 nm in the clear water scenario was nearly constant at all water depths but was decreased as expected when clarity was reduced by increased SPM. It is assumed this complex behavior was caused by the change in the bulk scattering phase function of the water layer from the surface to the submerged plastic target. The phase function is close to isotropy for the clear water case and is significantly forward peaked following the addition of sediments for the more turbid cases. The smoothing out and decrease in magnitude were also found in the other plastic targets as the depth was increased to 1.5 m (Fig. 5). At greater depths, the simulated signal was similar to the corresponding Rblank measurement (Fig. 2 and Fig. 5), the spectral features of plastics were no longer detectable.
A limit of detection or maximum submersion depth, zmax, can be computed for each wavelength depending on the uncertainty attached of the water-leaving radiance measurements (see Eq. (4)). The zmax values were determined using relative uncertainties ε (1, 5%). As expected a lower ε produced a deeper zmax, a ε =1% in the clear water scenario (SPM = 0 mg/L) was determined to have a maximum depths ranging between 1–3.5 m (Fig. 6). Under these conditions the blue target cannot be distinguished at any depth around 590 nm. For the cases included SPM in the water column, it is expected that detection of the submerged plastics will be limited to the top 1 m layer as shown by the grey and red lines in Fig. 6. A significant change in zmax was also observed when changing ε was increased from 1% to 5%. The 5% uncertainty was adapted from the recent assessment of above-water radiometric measurements [40].
Fig. 6. Computed maximum submersion depth (zmax) for detection of plastic targets considering a radiometric uncertainty ε attached to the measurements in varied in-water suspended particulate matter (SPM).
3.2. TOA simulations
Assuming an aerosol load of maritime type and optical thickness of 0.1 at 550 nm, TOA radiances were derived for the measured water clarity conditions and submerged plastics as displayed in Fig. 7. In this figure, TOA signal was determined in radiance units to showcase and quantify the minimum absolute radiometric sensitivity required for in-orbit optical sensors to potentially detect submerged plastics. These simulations suggest that the presence of submerged plastic type significantly alter the TOA radiance magnitude whatever the concentration of SPM evaluated in this study. Nevertheless, the detection was confined to the top 0.1 m surface layer at the lowest water clarity condition.
Fig. 7. Simulated top-of-atmosphere (TOA) radiance spectra for the submerged plastics in different suspended particulate matter (SPM) levels obtained from laboratory measurements assuming an atmosphere with maritime aerosol type, aerosol optical thicknesses (aot) at 550 nm = 0.1 and solar zenith angle of 40 °.
In order to remove the solar spectral features, the simulated TOA radiances were normalized by the extraterrestrial irradiance, thus the intrinsic plastic signatures (in reflectance unit) can be further highlighted (Fig. 8). Different spectral shapes were obtained in the TOA reflectance spectra compared to BOA signal (see Fig. 5), suggesting observation with an intervening atmosphere were less sensitive to for sensing submerged plastics. It was more salient in the blue spectrum corresponding to the region the atmosphere is most turbid (i.e., high molecular scattering). The influence of the water vapor absorption as well as atmospheric gases was noticeable outside the atmospheric windows (∼900–1000 and 1100 - 1200 nm). Those results were confirmed by further sensitivity analyses performed in changing the aerosol types to continental, dust and urban scenarios for different viewing geometries.
Fig. 8. Simulated top-of-atmosphere (TOA) reflectance spectra for the submerged plastics in different suspended particulate matter (SPM) levels obtained from laboratory measurements assuming an atmosphere with maritime aerosol type, aerosol optical thicknesses (aot) at 550 nm = 0.1 and solar zenith angle of 40 °.
4. Satellite application and discussion
4.1. Aerosols and atmospheric correction
The TOA reflectances were simulated hyperspectrally for a series atmospheric conditions (considering a comprehensive set of viewing geometries and aerosol load and types) and for the different plastic and water conditions analyzed in this study. In remote sensing applications, the aerosol load and types must be identified to fully process and analyzed the signal originating from the water column itself, procedure known as atmospheric correction (AC) [41]. Different techniques of AC have been developed depending on the satellite sensor capabilities (number and width of spectral bands, viewing geometry…) showing different levels of robustness and accuracy [27]. The AC is therefore a critical step to retrieve the sought-after water information. This is particularly true for complex water types including moderate to highly turbid cases [42]. In this section, impacts of aerosols and imperfect atmospheric correction on the submerged plastic detectability are discussed.
In order to account for the potential impacts of AC uncertainty on the retrieved water signal, the procedure described section 2.3 was applied to simulate the three components of the TOA reflectance: (i) intrinsic atmospheric reflectance, $R_{atm}^{TOA}$, (ii) water reflectance at TOA, $R_w^{TOA}$, and (iii) the reflectance due to adjacency effect from the neighboring out-of-pixel water surface, $R_{env}^{TOA}$. Effective reflectance as seen from TOA level when assuming a horizontally homogeneous water column was derived as the summation of $R_{atm}^{TOA}$, $R_w^{TOA}$ and $R_{env}^{TOA}$. Here, the impact of aot uncertainties on these components was assessed following three scenarios with aot = 0.05, 0.2 and 0.5 at 550 nm. The analyses were performed by considering maritime, continental and urban aerosol types. Figure 9 shows the results for the case of the white PP rope target.
Fig. 9. Intrinsic atmosphere ($R_{atm}^{TOA}$), water ($R_w^{TOA}$), environment ($R_{env}^{TOA}$), total or apparent top-of-atmosphere (TOA) reflectance components for the subsurface white polypropylene rope target in clear water for maritime, continental and urban aerosol types with an aerosol optical thicknesses at 550 nm (aot = 0.05, 0.2, 0.5). The shaded backgrounds highlight the uncertainties in the retrieved aot by typical atmospheric correction algorithms. The solar and viewing zenith angles were set to 30° and 5°, respectively, for a relative azimuth of 75°.
From those computations, it can be readily observed that the main source of uncertainty comes from the $R_w^{TOA}$ component whatever the aerosol type considered. For each aerosol load condition, this component is mainly impacted by the atmospheric transmittance through the contribution of the aerosols to light extinction. The uncertainties due to $R_{atm}^{TOA}$ and $R_{env}^{TOA}$ were considered second order. Interestingly, the two distinct behaviors of $R_w^{TOA}$ and $R_{env}^{TOA}$ lead to minimization of the uncertainties due to aerosols in the total reflectance for the maritime aerosol type. Even if it is not generalizable to the urban aerosols for example, this could be further exploited for plastic identification since the observed TOA signal can be seen as virtually independent on the atmospheric correction procedure. As shown by several studies, the adjacency effect remains significant over spatial extent of more than a kilometer [43,44]. Therefore, at the satellite level, the environment reflectance will be enhanced due to adjacency effect over large plastic patches (> 1 km, horizontal distribution). In other words, the fourth column of Fig. 9 shows that the spectral features could be used to detect large patches of subsurface plastics without applying AC due to the strong impact of the environment reflectance.
Aggregated plastic patches tend to be heterogeneous, scattered or in windrows and surrounded by plastic-free water. In this case, the corresponding $R_{env}^{TOA}$ originates from the darker water masses with reflectance one order of magnitude lower than that of the plastic patch. It can be presumed that the adjacency effects over the plastic patch are negligible. The total TOA reflectance values assuming null adjacency for maritime, continental and urban aerosol types were calculated for the subsurface submerged plastic targets (Fig. 10). In this case, those findings suggest that the aerosol type has less impact on the uncertainties compared to the effect of aot.
Fig. 10. Simulated top-of-atmosphere (TOA) reflectance spectra of submerged subsurface plastics at the assuming maritime, continental and urban aerosol types with an aerosol optical thicknesses at 550 nm (aot = 0.05, 0.2, 0.5). Plastics are assumed to be surrounded by plastic-free water thus null adjacency effects. Viewing geometry is similar to that of Fig. 9.
A change in aerosol load affected the magnitude and shape of the total reflectance when sensing was focused on small aggregated plastic patches. The combination of these two effects could lead to ambiguities when evaluating uncorrected TOA reflectance of subsurface plastics, suggesting the relevance of applying accurate AC to account for aerosols. Nevertheless, as discussed above, the main impact of the aerosol is significant in the computation of the atmospheric transmittance. As such, computation of the diffuse sky light (${R_{atm}^{TOA}}$) does not need to be as accurate as the retrieval of the spectral values of the aerosol optical thickness that strongly impact the spectral atmospheric transmittance. Therefore, it can be appropriate to use some gridded spectral aot provided by global atmospheric monitoring systems instead of performing AC procedure directly on the satellite measurements as previously recommended [45]. First, the accuracy of the estimation of spectral aot will be most likely strongly limited in the complex water types and heterogeneous environments retrieved in polluted areas [46,47]. Second, infrastructures such as Copernicus Atmosphere Monitoring Service (CAMS) [48,49] or Modern-Era Retrospective Analysis for Research and Applications-2 (MERRA-2) [50,51] provide 3-hourly estimates of spectral aot data that can be used with a reasonable accuracy for such an usage [52].
4.2. Impact of satellite sensors characteristics
Computation accounting for the uncertainties in AC were also performed to generate hyperspectral TOA reflectance signals of submerged plastics as detailed in section 3. Such robust simulations have relevance for current and future hyperspectral missions [53] like the German EnMap, Italian PRISMA, NASA PACE or ESA CHIME. Multispectral missions are also important as they exhibit very narrow spectral responses (e.g., Sentinel-3) to pan-chromatic ones (e.g., WorldView-3) whilst pixel resolution ranges from kilometric on Sentinel-3, decametric on Sentinel-2 and submetric on WorldView-3.
Spectral responses for best representative current missions (Sentinel-3, Sentinel-2, WorldView-3) were simulated for the clear waters consisting of submerged plastics at subsurface, 0.12 and 0.32 m depth under an atmosphere with maritime aerosols and aot = 0.2 (Fig. 11). The horizontal bars highlight the spectral bandwidth whilst the uncertainties in AC are represented by the vertical error bars. Note that the simulations were performed under the assumption of “small plastic patch” with negligible adjacency effects. Obviously, no signal is detectable for the satellite bands corresponding to the part of the spectrum showing no impact of plastic in the hyperspectral computations (see Fig. 10). More interestingly, for the blue PP rope target, the error bars of the blue wavebands are overlaid on each other at the three selected depths. This suggests identification of blue coloured submerged plastics might be challenging as a result of AC related uncertainties. It was also found out that narrowing the spectral responses from WorldView-3 to Sentinel-3 mission did not significantly improve the sensing of submerged plastics. In the red-NIR spectrum, most of the spectral wavebands could distinguish the submerged plastic signal depths without error bar superposition.
Fig. 11. Spectral band responses of Sentinel-3, Sentinel-2 and WorldView-3 for submerged plastics in clear water with a maritime aerosol type and aot = 0.2. Horizontal bars correspond to the spectral bandwidth and vertical error bars correspond to the atmospheric correction uncertainties.
The detection of the submerged plastics in clear to turbid waters from Sentinel-3, Sentinel-2 and WorldView-3 was evaluated for the white PP rope sample. The target was submerged at subsurface, 0.12 and 0.32 m depth under an atmosphere with maritime aerosols and aot = 0.2 (Fig. 12). The results indicate detection was feasible in the top surface layers at all water clarity levels. However, in the very turbid waters detection was limited to the near surface submersion depths, < 0.1 m, as shown by the similar reflectance signals for plastics submerged at 0.12 and 0.32 m (Fig. 12).
Fig. 12. Spectral band responses of Sentinel-3, Sentinel-2 and WorldView-3 for submerged white polypropylene rope plastics in varied in-water suspended particulate matter (SPM) concentrations with a maritime aerosol type and aot = 0.2. Horizontal bars correspond to the spectral bandwidth and vertical bars correspond to the atmospheric correction uncertainties.
The red-NIR spectrum is considered to be the appropriate for monitoring floating macro-plastic [11,12,54]. This study confirms the potential interest for exploitation of such spectral bands for submerged plastics as well consistent with recent findings [17]. The potential use of such wavebands is even suitable in highly turbid water conditions for detecting plastics submerged within the top layers in the subsurface waters. The extensive simulations of the satellite responses were performed under the assumption of no adjacency radiance coming from neighbouring out-of-pixel water masses. This hypothesis stands if the plastic patch is sufficiently small in comparison to the pixel size and/or the typical scale of adjacency effects which are roughly greater than a few kilometres. Such conditions appear to be reasonable since plastic debris are likely concentrated into convergence zones, eddies or fronts of submesoscale extent [1,7]. From our numerical simulations, it can be presumed that the characteristics of Sentinel-3, Sentinel-2 and WorldView-3 provide nearly similar capabilities to remotely sense submerged plastic types in aggregated patches. This is also true for the pan-chromatic broadband of WorldView-3 which possesses a spatial resolution of ∼0.3 m. As a plastic patch might be small in comparison to the pixel size and due to the difficulty to interpret subpixel signatures, it could be of particular interest to exploit such highly-spatially resolved band to ensure robust plastic detection and help to build more accurate algorithms for optical monitoring of polluted waters.
5. Conclusion and outlook
Laboratory measurements of virgin plastics were investigated for a set of submersion depths and several water clarity conditions from clear to highly turbid waters. The measured reflectance spectra gathered from an above-water spectroradiometric system, were used to parametrize an optical model permitting both minimization of the surface reflection impacts and quality control the actual hyperspectral reflectance values. We believe our proposed optical model parameterization could be advantageously adapted to develop novel retrieval algorithms dedicated to submerged plastics. Diagnostics based on the optical model demonstrated that maximum submersion depth, where tested plastic target became undetectable was limited to the first meter under the water surface. Spectral characteristics about submerged plastics were shown to be limited to the range from the UV to ∼1200 nm in clear to very turbid waters. Reflectance signals were also shown to be influenced by the apparent color of the plastic and polymer types. Clearly, there is a research gap to advance knowledge about these optical properties that could assess both virgin and weathered plastics with a wider range of polymer types as well as apparent colors.
Full radiative transfer computations based on the laboratory data were used to assess the radiometric and spectral capabilities required for monitoring submerged plastics from space. Hyperspectral modelling was done by considering various geometries, aerosol loads and types in space sensing of submerged plastics in clear to turbid waters. Spectral responses for representative current missions (Sentinel-3, Sentinel-2, WorldView-3) were simulated based on the hyperspectral simulations. These extensive analyses showed that the sensitivity of TOA usable information is limited in the blue spectrum due to uncertainty of the state-of-the-art atmospheric corrections. Prospective applications of the red-NIR spectrum is envisioned for potential detection of submerged plastics. The findings also suggested that the aerosol type had less impact on the uncertainties compared to the aerosol optical thickness through its effects on the atmospheric transmittance. It was demonstrated that narrowing the spectral responses from WorldView-3 to Sentinel-3 mission did not significantly improve the potential sensing of submerged plastics. For practical image processing, the 3-hourly estimates of spectral aot data provided by CAMS or MERRA infrastructures could be used to correct dedicated optical satellite mission for the atmospheric transmittance.
In this work, numerical solutions were obtained from a small set of plastic targets. Further efforts are needed to assess spectral responses in presence of more complex settings with mixture of submerged and floating macroplastics with or without adjunction of other microplastics. Nevertheless, solutions will have to be found in real-world conditions to test our model predictions based on both in-situ and satellite measurements. Relevant measurements as well as knowledge is expected from the various research projects e.g. funded by National Aeronautics and Space Administration and Portugal Space Agency on remote sensing of marine litter. Through the support of the IOCCG Task Force on the Remote Sensing of Marine Litter and Debris, Sentinel-2 collected some data over the Great Pacific Garbage Patch that could be utilized to analyze the retrieved spectral band information provided that ground-truths are available. Another aspect to improve our monitoring capabilities could stem from other optical properties such as polarization which are for the moment poorly exploited. Such efforts would for sure benefit from approaches linking theoretical modelling and laboratory measurements as achieved for light intensity by this study.
Funding
Discovery Element of the European Space Agency’s Basic Activities (4000132037/20/NL/GLC); Deutsche Forschungsgemeinschaft (417276871).
Acknowledgements
We are grateful for the support during data collection by Johan Mijnendonckx, Sindy Sterckx and Robin de Vries. Measurements were obtained through an experiment-of-opportunity and collaboration with Els Knaeps, Johan Mijnendonckx and Sindy Sterckx. The authors thank the three anonymous reviewers for their insightful comments and meticulous reviews.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are available in Dataset 1, Ref. [55].
References
1. E. Van Sebille, S. Aliani, K. L. Law, N. Maximenko, J. M. Alsina, A. Bagaev, M. Bergmann, B. Chapron, I. Chubarenko, A. Cózar, P. Delandmeter, M. Egger, B. Fox-Kemper, S. P. Garaba, L. Goddijn-Murphy, B. D. Hardesty, M. J. Hoffman, A. Isobe, C. E. Jongedijk, M. L. A. Kaandorp, L. Khatmullina, A. A. Koelmans, T. Kukulka, C. Laufkötter, L. Lebreton, D. Lobelle, C. Maes, V. Martinez-Vicente, M. A. Morales Maqueda, M. Poulain-Zarcos, E. Rodríguez, P. G. Ryan, A. L. Shanks, W. J. Shim, G. Suaria, M. Thiel, T. S. Van Den Bremer, and D. Wichmann, “The physical oceanography of the transport of floating marine debris,” Environ. Res. Lett. 15(2), 023003 (2020). [CrossRef]
2. V. Martínez-Vicente, J. R. Clark, P. Corradi, S. Aliani, M. Arias, M. Bochow, G. Bonnery, M. Cole, A. Cózar, R. Donnelly, F. Echevarría, F. Galgani, S. P. Garaba, L. Goddijn-Murphy, L. Lebreton, H. A. Leslie, P. K. Lindeque, N. Maximenko, F. R. Martin-Lauzer, D. Moller, P. Murphy, L. Palombi, V. Raimondi, J. Reisser, L. Romero, S. G. H. Simis, S. Sterckx, R. C. Thompson, K. N. Topouzelis, E. van Sebille, J. M. Veiga, and A. D. Vethaak, “Measuring marine plastic debris from space: Initial assessment of observation requirements,” Remote Sens. 11(20), 2443 (2019). [CrossRef]
3. N. Maximenko, P. Corradi, K. L. Law, E. Van Sebille, S. P. Garaba, R. S. Lampitt, F. Galgani, V. Martinez-Vicente, L. Goddijn-Murphy, J. M. Veiga, R. C. Thompson, C. Maes, D. Moller, C. R. Löscher, A. M. Addamo, M. Lamson, L. R. Centurioni, N. Posth, R. Lumpkin, M. Vinci, A. M. Martins, C. D. Pieper, A. Isobe, G. Hanke, M. Edwards, I. P. Chubarenko, E. Rodriguez, S. Aliani, M. Arias, G. P. Asner, A. Brosich, J. T. Carlton, Y. Chao, A. M. Cook, A. Cundy, T. S. Galloway, A. Giorgetti, G. J. Goni, Y. Guichoux, B. D. Hardesty, N. Holdsworth, L. Lebreton, H. A. Leslie, I. Macadam-Somer, T. Mace, M. Manuel, R. Marsh, E. Martinez, D. Mayor, M. Le Moigne, M. E. M. Jack, M. C. Mowlem, R. W. Obbard, K. Pabortsava, B. Robberson, A. E. Rotaru, M. T. Spedicato, M. Thiel, A. Turra, and C. Wilcox, “Towards the integrated marine debris observing system,” Front. Mar. Sci. 6, 1 (2019). [CrossRef]
4. T. Acuña-Ruz, D. Uribe, R. Taylor, L. Amézquita, M. C. Guzmán, J. Merrill, P. Martínez, L. Voisin, and C. Mattar B, “Anthropogenic marine debris over beaches: Spectral characterization for remote sensing applications,” Remote Sens. Environ. 217, 309–322 (2018). [CrossRef]
5. M. C. Evans and C. S. Ruf, “Toward the Detection and Imaging of Ocean Microplastics With a Spaceborne Radar,” IEEE Trans. Geosci. Remote Sens. 60, 1–9 (2022). [CrossRef]
6. K. Topouzelis, A. Papakonstantinou, and S. P. Garaba, “Detection of floating plastics from satellite and unmanned aerial systems (Plastic Litter Project 2018),” ITC J. 79, 175–183 (2019). [CrossRef]
7. N. Maximenko, J. Hafner, and P. Niiler, “Pathways of marine debris derived from trajectories of Lagrangian drifters,” Mar. Pollut. Bull. 65(1-3), 51–62 (2012). [CrossRef]
8. W. G. Pichel, T. S. Veenstra, J. H. Churnside, E. Arabini, K. S. Friedman, D. G. Foley, R. E. Brainard, D. Kiefer, S. Ogle, P. Clemente-Colón, and X. Li, “GhostNet marine debris survey in the Gulf of Alaska – Satellite guidance and aircraft observations,” Mar. Pollut. Bull. 65(1-3), 28–41 (2012). [CrossRef]
9. Y.-J. Park, Y.-J. Park, Y.-J. Park, S. P. Garaba, S. P. Garaba, S. P. Garaba, and B. Sainte-Rose, “Detecting the Great Pacific Garbage Patch floating plastic litter using WorldView-3 satellite imagery,” Opt. Express 29(22), 35288–35298 (2021). [CrossRef]
10. S. P. Garaba, M. Arias, L. Biermann, P. Corradi, L. Lorenzoni, F.-R. Martin-Lauzer, V. Martinez-Vicente, D. Mitra, and H. Murakami, “IOCCG Task Force on Remote Sensing of Marine Litter and Debris - Terms of Reference,” (2021).
11. S. P. Garaba, T. Acuña-Ruz, and C. B. Mattar, “Hyperspectral longwave infrared reflectance spectra of naturally dried algae, anthropogenic plastics, sands and shells,” Earth Syst. Sci. Data 12(4), 2665–2678 (2020). [CrossRef]
12. S. P. Garaba, M. Arias, P. Corradi, T. Harmel, R. de Vries, and L. Lebreton, “Concentration, anisotropic and apparent colour effects on optical reflectance properties of virgin and ocean-harvested plastics,” J. Hazard. Mater. 406, 124290 (2021). [CrossRef]
13. L. Goddijn-Murphy and J. Dufaur, “Proof of concept for a model of light reflectance of plastics floating on natural waters,” Mar. Pollut. Bull. 135, 1145–1157 (2018). [CrossRef]
14. E. Knaeps, S. Sterckx, G. Strackx, J. Mijnendonckx, M. Moshtaghi, S. P. Garaba, and D. Meire, “Hyperspectral-reflectance dataset of dry, wet and submerged marine litter,” Earth Syst. Sci. Data (2021).
15. L. Corbari, A. Maltese, F. Capodici, M. C. Mangano, G. Sarà, and G. Ciraolo, “Indoor spectroradiometric characterization of plastic litters commonly polluting the Mediterranean Sea: toward the application of multispectral imagery,” Sci. Rep. 10(1), 1–12 (2020).
16. J. A. Guffogg, M. Soto-Berelov, S. D. Jones, C. J. Bellman, J. L. Lavers, and A. K. Skidmore, “Towards the Spectral Mapping of Plastic Debris on Beaches,” Remote Sens. 13(9), 1850 (2021). [CrossRef]
17. M. Moshtaghi, E. Knaeps, S. Sterckx, S. P. Garaba, and D. Meire, “Spectral reflectance of marine macroplastics in the VNIR and SWIR measured in a controlled environment,” Sci. Rep. 11(1), 5436 (2021). [CrossRef]
18. C. Hu, “Remote detection of marine debris using satellite observations in the visible and near infrared spectral range: Challenges and potentials,” Remote Sens. Environ. 259, 112414 (2021). [CrossRef]
19. G. Neukermans, T. Harmel, M. Galí, N. Rudorff, J. Chowdhary, O. Dubovik, C. Hostetler, Y. Hu, C. Jamet, K. Knobelspiesse, Y. Lehahn, P. Litvinov, A. M. Sayer, B. Ward, E. Boss, I. Koren, and L. A. Miller, “Harnessing remote sensing to address critical science questions on ocean-atmosphere interactions,” Elem Sci Anth 6(1), 71 (2018). [CrossRef]
20. J. Reisser, B. Slat, K. Noble, K. Du Plessis, M. Epp, M. Proietti, J. De Sonneville, T. Becker, and C. Pattiaratchi, “The vertical distribution of buoyant plastics at sea: An observational study in the North Atlantic Gyre,” Biogeosciences 12(4), 1249–1256 (2015). [CrossRef]
21. M. Kooi, J. Reisser, B. Slat, F. F. Ferrari, M. S. Schmid, S. Cunsolo, R. Brambini, K. Noble, L. A. Sirks, T. E. W. Linders, R. I. Schoeneich-Argent, and A. A. Koelmans, “The effect of particle properties on the depth profile of buoyant plastics in the ocean,” Sci. Rep. 6(1), 1–10 (2016). [CrossRef]
22. Y. K. Song, S. H. Hong, S. Eo, M. Jang, G. M. Han, A. Isobe, and W. J. Shim, “Horizontal and Vertical Distribution of Microplastics in Korean Coastal Waters,” Environ. Sci. Technol. 52(21), 12188–12197 (2018). [CrossRef]
23. S. P. Garaba and H. M. Dierssen, “An airborne remote sensing case study of synthetic hydrocarbon detection using short wave infrared absorption features identified from marine-harvested macro- and microplastics,” Remote Sens. Environ. 205, 224–235 (2018). [CrossRef]
24. C. Jamet, A. Ibrahim, Z. Ahmad, F. Angelini, M. Babin, M. J. Behrenfeld, E. Boss, B. Cairns, J. Churnside, J. Chowdhary, A. B. Davis, D. Dionisi, L. Duforêt-Gaurier, B. Franz, R. Frouin, M. Gao, D. Gray, O. Hasekamp, X. He, C. Hostetler, O. V. Kalashnikova, K. Knobelspiesse, L. Lacour, H. Loisel, V. Martins, E. Rehm, L. Remer, I. Sanhaj, K. Stamnes, S. Stamnes, S. Victori, J. Werdell, and P.-W. Zhai, “Going Beyond Standard Ocean Color Observations: Lidar and Polarimetry,” Front. Mar. Sci. 6, 251 (2019). [CrossRef]
25. R. A. Martins, J. M. Felicio, S. A. Matos, J. R. Costa, and C. A. Fernandes, “Preliminary characterization of microwave backscattering of floating plastic,” 2021 Telecoms Conf. ConfTELE 2021 (2021).
26. J. Chowdhary, P. W. Zhai, E. Boss, H. Dierssen, R. Frouin, A. Ibrahim, Z. Lee, L. A. Remer, M. Twardowski, F. Xu, X. Zhang, M. Ottaviani, W. R. Espinosa, and D. Ramon, “Modeling atmosphere-ocean radiative transfer: A PACE mission perspective,” Front. Earth Sci. 7, 100 (2019). [CrossRef]
27. R. J. Frouin, B. A. Franz, A. Ibrahim, K. Knobelspiesse, Z. Ahmad, B. Cairns, J. Chowdhary, H. M. Dierssen, J. Tan, O. Dubovik, X. Huang, A. B. Davis, O. Kalashnikova, D. R. Thompson, L. A. Remer, E. Boss, O. Coddington, P. Y. Deschamps, B. C. Gao, L. Gross, O. Hasekamp, A. Omar, B. Pelletier, D. Ramon, F. Steinmetz, and P. W. Zhai, “Atmospheric Correction of Satellite Ocean-Color Imagery During the PACE Era,” Front. Earth Sci. 7, 145 (2019). [CrossRef]
28. H. R. Gordon, J. W. Brown, R. H. Evans, O. B. Brown, Baker Smith, and D. K. Clark, “A semianalytic radiance model of ocean color,” J. Geophys. Res. 93(D9), 10909–10924 (1988). [CrossRef]
29. Z. Lee, K. L. Carder, C. D. Mobley, R. G. Steward, and J. S. Patch, “Hyperspectral remote sensing for shallow waters I A semianalytical model,” Appl. Opt. 37(27), 6329 (1998). [CrossRef]
30. Z. Lee, S. Shang, C. Hu, K. Du, A. Weidemann, W. Hou, J. Lin, and G. Lin, “Secchi disk depth: A new theory and mechanistic model for underwater visibility,” Remote Sens. Environ. 169, 139–149 (2015). [CrossRef]
31. J. R. V. Zaneveld, “A theoretical derivation of the dependence of the remotely sensed reflectance of the ocean on the inherent optical properties,” J. Geophys. Res. 100(C7), 13135–13142 (1995). [CrossRef]
32. R. W. Preisendorfer, “Secchi Disk Science: Visual Optics of Natural Waters,” Limnol. Oceanogr. 31(5), 909–926 (1986). [CrossRef]
33. M. A. Branch, T. F. Coleman, and Y. Li, “Subspace, interior, and conjugate gradient method for large-scale bound-constrained minimization problems,” SIAM J. Sci. Comput. 21(1), 1–23 (1999). [CrossRef]
34. S. Y. Kotchenova, E. F. Vermote, R. Matarrese, and F. J. Klemm, “Validation of a vector version of the 6S radiative transfer code for atmospheric correction of satellite data. Part I: Path radiance,” Appl. Opt. 45(26), 6762–6774 (2006). [CrossRef]
35. E. F. Vermote, D. Tanre, J. L. Deuze, M. Herman, and J. J. Morcrette, “Second Simulation of the Satellite Signal in the Solar Spectrum, 6S: An overview,” IEEE Trans. Geosci. Remote Sens. 35(3), 675–686 (1997). [CrossRef]
36. C. Cox and W. H. Munk, “Statistics of the sea surface derived from sun glitter,” J. Mar. Res. 13(2), 198–227 (1954).
37. R. C. Levy, S. Mattoo, L. A. Munchak, L. A. Remer, A. M. Sayer, F. Patadia, and N. C. Hsu, “The Collection 6 MODIS aerosol products over land and ocean,” Atmos. Meas. Tech. 6(11), 2989–3034 (2013). [CrossRef]
38. O. Burggraaff, “Biases in reflectance convolution,” Opt. Express 28(9), 13801–13816 (2020). [CrossRef]
39. S. P. Garaba and H. M. Dierssen, “Hyperspectral ultraviolet to shortwave infrared characteristics of marine-harvested, washed-ashore and virgin plastics,” Earth Syst. Sci. Data 12(1), 77–86 (2020). [CrossRef]
40. G. Zibordi, B. N. Holben, M. Talone, D. D’Alimonte, I. Slutsker, D. M. Giles, and M. G. Sorokin, “Advances in the Ocean Color Component of the Aerosol Robotic Network (AERONET-OC),” J. Atmos. Ocean. Technol. 38(4), 725–746 (2021). [CrossRef]
41. H. R. Gordon, “Atmospheric correction of ocean color imagery in the Earth Observing System era,” J. Geophys. Res. 102(D14), 17081–17106 (1997). [CrossRef]
42. N. Pahlevan, A. Mangin, S. V. Balasubramanian, B. Smith, K. Alikas, K. Arai, C. Barbosa, S. Bélanger, C. Binding, M. Bresciani, C. Giardino, D. Gurlin, Y. Fan, T. Harmel, P. Hunter, J. Ishikaza, S. Kratzer, M. K. Lehmann, M. Ligi, R. Ma, F. R. Martin-Lauzer, L. Olmanson, N. Oppelt, Y. Pan, S. Peters, N. Reynaud, L. A. Sander de Carvalho, S. Simis, E. Spyrakos, F. Steinmetz, K. Stelzer, S. Sterckx, T. Tormos, A. Tyler, Q. Vanhellemont, and M. Warren, “ACIX-Aqua: A global assessment of atmospheric correction methods for Landsat-8 and Sentinel-2 over lakes, rivers, and coastal waters,” Remote Sens. Environ. 258, 112366 (2021). [CrossRef]
43. B. Bulgarelli, G. Zibordi, and V. Kiselev, “Simulation and analysis of adjacency effects in coastal waters: a case study,” Appl. Opt. 53(8), 1523–1545 (2014). [CrossRef]
44. B. Bulgarelli and G. Zibordi, “Adjacency radiance around a small island: implications for system vicarious calibrations,” Appl. Opt. 59(10), C63–C69 (2020). [CrossRef]
45. T. Harmel, M. Chami, T. Tormos, N. Reynaud, and P.-A. Danis, “Sunglint correction of the Multi-Spectral Instrument (MSI)-SENTINEL-2 imagery over inland and sea waters from SWIR bands,” Remote Sens. Environ. 204, 308–321 (2018). [CrossRef]
46. S. Hlaing, T. Harmel, A. Gilerson, R. Foster, A. Weidemann, R. Arnone, M. Wang, and S. Ahmed, “Evaluation of the VIIRS ocean color monitoring performance in coastal regions,” Remote Sens. Environ. 139, 398–414 (2013). [CrossRef]
47. L. Biermann, D. Clewley, V. Martinez-Vicente, and K. Topouzelis, “Finding Plastic Patches in Coastal Waters using Optical Satellite Data,” Sci. Rep. 10(1), 1–10 (2020). [CrossRef]
48. J. Flemming, A. Benedetti, A. Inness, R. J. Engelen, L. Jones, V. Huijnen, S. Remy, M. Parrington, M. Suttie, A. Bozzo, V.-H. Peuch, D. Akritidis, and E. Katragkou, “The CAMS interim Reanalysis of Carbon Monoxide, Ozone and Aerosol for 2003–2015,” Atmos. Chem. Phys. 17(3), 1945–1983 (2017). [CrossRef]
49. J.-J. Morcrette, O. Boucher, L. Jones, D. Salmond, P. Bechtold, A. Beljaars, A. Benedetti, A. Bonet, J. W. Kaiser, M. Razinger, M. Schulz, S. Serrar, A. J. Simmons, M. Sofiev, M. Suttie, A. M. Tompkins, and A. Untch, “Aerosol analysis and forecast in the European Centre for Medium-Range Weather Forecasts Integrated Forecast System: Forward modeling,” J. Geophys. Res. 114(D6), D06206 (2009). [CrossRef]
50. V. Buchard, C. A. Randles, A. M. da Silva, A. Darmenov, P. R. Colarco, R. Govindaraju, R. Ferrare, J. Hair, A. J. Beyersdorf, L. D. Ziemba, and H. Yu, “The MERRA-2 Aerosol Reanalysis, 1980 Onward. Part II: Evaluation and Case Studies,” J. Clim. 30(17), 6851–6872 (2017). [CrossRef]
51. R. Gelaro, W. McCarty, M. J. Suárez, R. Todling, A. Molod, L. Takacs, C. A. Randles, A. Darmenov, M. G. Bosilovich, R. Reichle, K. Wargan, L. Coy, R. Cullather, C. Draper, S. Akella, V. Buchard, A. Conaty, A. M. da Silva, W. Gu, G.-K. Kim, R. Koster, R. Lucchesi, D. Merkova, J. E. Nielsen, G. Partyka, S. Pawson, W. Putman, M. Rienecker, S. D. Schubert, M. Sienkiewicz, and B. Zhao, “The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2),” J. Clim. 30(14), 5419–5454 (2017). [CrossRef]
52. C. A. Gueymard and D. Yang, “Worldwide validation of CAMS and MERRA-2 reanalysis aerosol optical depth products using 15 years of AERONET observations,” Atmos. Environ. 225, 117216 (2020). [CrossRef]
53. A. Ibrahim, B. Franz, Z. Ahmad, R. Healy, K. Knobelspiesse, B. C. Gao, C. Proctor, and P. W. Zhai, “Atmospheric correction for hyperspectral ocean color retrieval with application to the Hyperspectral Imager for the Coastal Ocean (HICO),” Remote Sens. Environ. 204, 60–75 (2018). [CrossRef]
54. H. M. Dierssen and S. P. Garaba, “Bright oceans: Spectral differentiation of whitecaps, sea ice, plastics, and other flotsam,” in Recent Advances in the Study of Oceanic Whitecaps: Twixt Wind and Waves, P. Vlahos and E. C. Monahan, eds., Springer I (2020), pp. 197–208. [CrossRef]
55. S. Garaba, R. de Vries, E. Knaeps, J. Mijnendonckx, and S. Sterckx, “Spectral reflectance measurements of dry and wet virgin plastics at varying water depth and water clarity from UV to SWIR (SEV-1),” .4TU.ResearchData: Version 1, 7 July 2020, https://doi.org/10.4121/uuid:9ee3be54-9132-415a-aaf2-c7fbf32d2199.
