Remote Sensing of Environment 259 (2021) 112414

Contents lists available at ScienceDirect

Remote Sensing of Environment

journal homepage: www.elsevier.com/locate/rse

Remote detection of marine debris using satellite observations in the visible

and near infrared spectral range: Challenges and potentials
Chuanmin Hu *
University of South Florida, 140 Seventh Avenue, South, St. Petersburg, FL 33701, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Despite the importance of remote detection of marine debris, nearly all published studies are focused on either
Remote sensing controlled experiments, or Sentinel-2 data with mixed band resolutions that are subject to large uncertainties. To
Marine debris date, key questions such as the following have not been addressed adequately: To what extent can the various
Microplastics
forms of marine debris be remotely detected and differentiated through satellite observations in the visible and
Resolution
Sensitivity
near infrared (NIR) spectral range, and how? Here, using published reflectance spectra of various types of
Spectral bands floating matters, I address these questions through sensitivity analyses, simulations, and spectral analyses of
Sentinel-2 satellite images. While the study is by no means comprehensive, several observations can still be made. First, it
PlanetScope/DOVE appears impossible to remotely detect marine microplastics from all existing and planned optical sensors. This is
simply because the contribution of these particles to the sensor signal, even when they are aggregated on the
water surface at the reported maximum particle density, is at least 60 times lower than the required signal
(~0.2% subpixel coverage) and 20 times lower than the sensor noise for a sensor with a signal-to-noise ratio
(SNR) of 200. In contrast, detecting macroplastics and other debris is possible when they form large patches
along ocean fronts or windrows. Second, assuming a SNR of 200, discriminating large patches of marine debris
from floating algae is only possible with a subpixel coverage of >0.3%. These threshold values are based on the
sensor SNRs only, and they represent the lower bounds of detection and discrimination, respectively. The real
threshold values above which a detection or discrimination is possible also depend on the observing conditions,
and therefore higher. Third, currently, Sentinel-2 MSI (Multi Spectral Instrument) sensors provide an optimal
trade between resolution and coverage, yet MSI sensors have SNRs <200, and interpretation of the MSI spectra
requires extra caution due to variable spatial resolutions in different bands, among other factors. From the
perspective of pure spectroscopy, it is possible to discriminate floating algae from non-algae floating matters but
difficult to differentiate the type of the latter (either plastic or non-plastic debris, foam, etc) because different
non-algae floating matters all show relatively flat reflectance spectral shapes in the vis-NIR spectral range.
Finally, based on these results, recommendations are made on algorithm designs and sensor designs, for example
spectral analysis should be performed over the difference spectra to minimize the impact of variable subpixel
coverage, and certain spectral bands are more important than others for the remote detection of marine debris.

1. Background on remote sensing of marine debris
By definition, marine debris refers to any persistent solid material
that is disposed (or abandoned) in the marine environment by natural
processes (including natural disasters such as Tsunami) and human ac­
tivities, for example microplastic particles, plastic bags or bottles,
cigarette butts, foam take-out containers, balloons, fishing gear, tree
branches/leaves, wood, among others. The impacts of marine debris on
marine animals and marine ecosystems have been increasingly
recognized in recent years, together with reports of increased land-based
sources of marine debris (Law et al., 2010). In addressing the pressing
needs of understanding the distributions and fates of various forms of
marine debris, inter-governmental agencies, environmental groups, and
the research community have started to work together toward an inte­
grated marine debris observing network (Maximenko et al., 2019), of
which remote sensing has been emphasized for its synoptic and frequent
coverage (see also Maximenko et al., 2017). The European Space Agency
is particularly keen on funding several pioneering efforts (Topouzelis

* Corresponding author.
E-mail address: huc@usf.edu.

https://doi.org/10.1016/j.rse.2021.112414
Received 3 November 2020; Received in revised form 20 February 2021; Accepted 22 March 2021
Available online 5 April 2021
0034-4257/© 2021 Elsevier Inc. All rights reserved.
C. Hu
et al., 2019 & 2020), with more projects to start in 2020 and new pro­
jects funded recently by the U.S. National Aeronautics and Space
Administration (NASA).
Among the key questions are: can the various forms of marine debris
be remotely sensed and quantified? If so, how? What are the optimal
remote sensing approaches to address the various user needs?
Several pioneering studies attempted to address these questions from
different aspects, with reviews provided by Maximenko et al. (2017 &
2019) and by Martinez-Vicente et al. (2019). Among the various remote
sensing techniques such as synthetic aperture radar (Arii et al., 2014)
and thermal infrared sensors (Garaba et al., 2020; Goddijn-Murphy and
Williamson, 2019), passive optical remote sensing in the visible, near-
infrared (NIR), and shortwave-infrared (SWIR) spectral range is
perhaps the most used because most medium- (a few hundred meters) or
high-resolution (meters to tens of meters) optical sensors are equipped
with spectral bands in this range. While it is beyond the scope of this
paper to present a comprehensive review on the vis-NIR-SWIR optical
remote sensing on this subject, the following list presents key examples
of these studies:
Spectral reflectance measurements in the lab:
• Hu et al. (2015) showed the first measurements of spectral reflec­
tance of various large artificial (man-made) marine debris patches
(Fig. 1a). Reflectance is rather “flat” (i.e., without narrow-band
features) for up to 900 nm (Fig. 1b).
• Garaba and Dierssen (2018, 2020) showed the first laboratory
measurements of marine microplastics (Fig. 1b) and also laboratory
measurements of various macroplastics in the vis-NIR-SWIR wave­
lengths. Reflectance is also “flat” for up to 900 nm, with some C–H
absorption features in the SWIR wavelengths.
Airborne or satellite measurements of man-made targets under controlled
environments:
• Garaba et al. (2018) applied an airborne imager in the SWIR wave­
lengths to measure man-made plastic patches.
• Topouzelis et al. (2019 & 2020) used Sentinel-2 MSI images to detect
man-made plastic targets under a controlled experimental setting.
Spectral shapes of MSI pixels over these targets show a spike around
833 or 741 nm, and the NIR reflectance magnitude of ~0.04 in­
dicates mixed pixels.
Applications to satellite measurements over oceanic environments:
• Kikaki et al. (2020) used Sentinel-2 MSI and Landsat-8 OLI to search
for plastic debris in the Caribbean Sea, based on spectral
Fig. 1. (a) Artificial (man-made) garbage patches in a preliminary experiment. (b) Their corresponding reflectance measured in Tampa Bay using a hand-held
spectrometer (Hu et al., 2015). In (b), reflectance of wet microplastics (red curve, right y-axis) was obtained from Garaba and Dierssen (2020), who provided a
hyperspectral dataset (vis-NIR-SWIR) for different types of plastics (microplastics, macroplastics, plastics with different colors). (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
2
Remote Sensing of Environment 259 (2021) 112414
interpretation to separate debris from Sargassum. The NIR reflec­
tance of the debris endmember is about ~0.04, suggesting mixed
pixels.
• Biermann et al. (2020) also used Sentinel-2 MSI data and examined
spectral shapes of various endmembers to classify pixels containing
marine debris. Similar to those in Kikaki et al. (2020), the NIR
reflectance of the debris endmember is about ~0.04, suggesting
mixed pixels.
• Matthews et al. (2017) showed images from high-resolution com­
mercial satellites and airborne digital photos where marine debris
after a Tsunami near Japan could be detected as bright pixels in the
images/photos.
Field surveys:
• Law et al. (2010) showed distributions and areal density of marine
microplastics particles in the North Atlantic, based on >6600 surface
net tows from 1986 to 2010. Most of the data showed <200,000
plastic pieces km− 2. There was only one extreme data point with 26
million pieces km− 2 (Fig. 5 of Maximenko et al., 2017).
• Cózar et al. (2014) analyzed plastic debris data collected from 1127
surface net tows in 442 sites, and found that most plastic pieces have
their size below 5 mm (i.e., microplastics) with a log-normal distri­
bution and a mode of ~2 mm.
• Eriksen et al. (2014) compiled a dataset of marine plastic debris
measured at 1571 stations from 680 net tows and 891 visual survey
transects. They showed that the dominant majority of all compiled
and modeled particle density is <1 million pieces km− 2 for particles
<4.75 mm in size.
• van Sebille et al. (2015) compiled a first global inventory of small
floating plastic debris based on field measurements by many groups.
Results indicate that for the vast majority of data, particle density is
<1 million pieces km− 2.
Remote sensing perspectives:
• Goddijn-Murphy et al. (2018) showed a theoretical concept on
hyperspectral remote sensing of marine macroplastics based on
radiative transfer theory. The concept was evaluated later using a
carefully designed experiment (Goddijn-Murphy and Dufaur, 2018),
yet no satellite-based application has been demonstrated.
• Martinez-Vicente et al. (2019) provided a perspective on sensor re­
quirements for remote sensing of marine plastic debris, with a focus
on the NIR and SWIR wavelengths only.
C. Hu
• Maximenko et al. (2017) provided a review on remote sensing of
marine debris from a NASA workshop, where the strengths and
limitations of different technologies are discussed.
• Maximenko et al. (2019) provided a comprehensive review of all
possible techniques in observing marine debris, including in situ
sensing, remote sensing, and modeling, with future approaches
outlined toward addressing the identified challenges (for example,
sensor requirement).
Although field surveys of marine debris started in the 1970s, remote
sensing studies did not start until much more recently. From these pio­
neering studies above, it appears that the feasibility of using routine
satellite platforms (e.g., Sentinel-2, Landsat-8) to detect large patches of
marine debris has been proven (Biermann et al., 2020; Kikaki et al.,
2020), and the rest is simply to implement a monitoring system to
combine the published algorithms and satellite data for global moni­
toring. However, this does not appear to be the case, as shown below. In
particular, it appears impossible to remotely sense microplastics
regardless of the optical sensors or resolutions. Then, to what extent can
we sense macroplastics and other forms of marine debris using satellite
remote sensing in the vis-NIR-SWIR spectral range, and how?
With a focus on the vis-NIR wavelengths, the objective of this paper
is to provide a perspective on whether and under what conditions
various forms of marine debris can be detected and discriminated
against other floating matters. The paper is based on published end­
member spectra, sensor sensitivity, simulation experiments, and spectral
analyses of Sentinel-2 data for demonstration purpose, from which
recommendations are made on sensor design as well as on algorithms
and approaches toward vis-NIR remote sensing of marine debris. The use
of SWIR wavelengths is also discussed. Furthermore, because marine
debris is an emerging topic in the scientific literature across many dis­
ciplines, satellite remote sensing is only one of the many tools to help
answer multi-disciplinary questions (Maximenko et al., 2019); inter­
ested readers are therefore referred to the published literature on other
aspects of marine debris monitoring.
2. Field and satellite data
2.1. Endmember spectra
To date, laboratory or in situ measurements of optical properties of
marine debris are scarce, with the exception of some artificial (man-made)
garbage patches and field-collected micro plastic particles (Fig. 1). One
common characteristic from these various forms of marine debris is that
they are all spectrally flat without narrow-band features in the visible –
NIR wavelengths (400–900 nm). Although only one example of micro­
plastics spectra from Garaba and Dierssen (2020) is shown in Fig. 1b to
illustrate such spectral characteristic, all their miscroplastics spectra show
the same characteristic. This is because of lack of narrow-band vegetative
Fig. 2. Hyperspectral and multi-spectral reflectance of various floating matters from (a) in situ measurements (Hu et al., 2015 for Sargassum; Hu et al., 2017 for Ulva;
McKinna, 2010 for Trichodesmium) and (b) OLCI measurements (Qi et al., 2020), respectively. Other than red Noctiluca and pumice rafts, all floating matters show the
pigment absorption around 670 nm (black arrows). (c) Clear-water and turbid-water endmember spectra used in the simulated mixing experiments. (For inter­
pretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3
Remote Sensing of Environment 259 (2021) 112414
pigment absorption. Indeed, once normalized to 25% reflectance in the
NIR (e.g., around 860 nm), all these spectra appear rather similar
(Fig. 1b). Therefore, in this study, collectively they may represent one
marine debris endmember.
In the ocean, other forms of floating matters also exist. These include
Sargassum fluitans and Sargassum natans (Wang et al., 2018), Sargassum
horneri (Qi et al., 2017), Ulva (Hu et al., 2017), cyanobacteria Tricho­
desmium (Hu et al., 2010; McKinna, 2010), emulsified oil (Lu et al.,
2020), green Noctiluca, red Noctiluca, pumice rafts (Qi et al., 2020), and
foams (whitecaps) (Dierssen, 2019). Some of these have been measured
in the field, with in situ hyperspectral reflectance being available (e.g.,
Fig. 2 of Hu et al., 2015) but others have only been assessed using multi-
band satellite data (e.g., pumice rafts). Some of these spectra are
compiled in Figs. 2a & 2b, which show typical spectral shapes of non-
debris floating matters. Note that except for the spectra in Fig. 2a
whose magnitudes in the NIR are >0.2 and therefore may be used as
endmember spectra (i.e., 100% coverage within a pixel) for the spectral
mixing experiments below, the NIR reflectance in Fig. 2b are all <0.1,
suggesting that these spectra derived from medium-resolution (300-m)
image pixels are mixed between floating matters and water (about 16% -
40% subpixel floating matter coverage), therefore cannot be used as
endmembers. Also note that except for red Noctiluca and pumice rafts, all
spectra in Figs. 2a & 2b have the typical reflectance trough around 670
nm due to chlorophyll a absorption (black arrows), and they also show
the typical vegetation red-edge reflectance above 700 nm. These char­
acteristics are different from the marine debris endmember spectra, and
such characteristics may form the basis for spectral discrimination be­
tween floating algae (pigment rich) and floating debris (no pigment,
Fig. 1) using the vis-NIR bands. Such a concept is demonstrated in the
simulations below using “plastic bags” (Fig. 1b) and Sargassum (Fig. 2a)
to represent marine debris and floating algae, respectively. In such
simulations, for water endmember spectra, two scenarios were selected
to represent clear water and turbid water, respectively (Fig. 2c). These
spectra were collected in the Florida Keys (May 12, 2013, 24.5333oN,
81.4016oW, chlorophyll concentration ~ 0.1 mg m− 3) and near Tampa
Bay (August 24, 2012, 27.9313oN, 82.6512oW. chlorophyll concentra­
tion ~ 12.7 mg m− 3), respectively, using a hand-held spectrometer
following NASA Ocean Optics protocols.
2.2. Satellite data
While there are currently many satellite sensors in orbit, Sentinel-2
MSI was selected to represent high-resolution optical sensors because
it provides a trade between spatial resolution (10–20 m) and revisit
frequency (2–3 days) for most coastal waters. MSI covers wavelengths of
vis-NIR-SWIR, suitable for detecting and differentiating small floating
matters. The use of MSI is also to follow a number of recent studies by
Kikaki et al. (2020), Biermann et al. (2020), and Topouzelis et al. (2019
& 2020) for both real applications and controlled experiments. MSI has
C. Hu
the following spectral bands: 443 (60), 492 (10), 560 (10), 665 (10), 704
(20), 741 (20), 783 (20), 841 (10), 865 (20), 1614 (20), and 2202 nm
(20), where the numbers in the parentheses represent their ground
resolutions in meters.
Level-1 MSI data were downloaded from the U. S. Geological Survey,
and processed using the Acolite software to generate Rayleigh corrected
reflectance (Rrc(λ), dimensionless), from which Red-Green-Blue and
False-colour RGB composite imagery were generated. In the FRGB im­
agery, a NIR band was used to replace the green band in the RGB im­
agery, making it suitable to detect floating matters with enhanced NIR
reflectance (Qi et al., 2020).
3. Methods
Regardless of the floating matter type (either marine debris, floating
algae, or other types of floating matters), remote detection always re­
quires two steps. Step 1 is to detect a spatial anomaly, i.e., some pixels
“stand out” from their nearby background waters. Step 2 is to spectrally
differentiate the pixel type from the spatial anomaly. Step 2 is impos­
sible without Step 1. In simpler words, the two steps can be shortened as:
1) is there “something”? 2) what is that “something”? If the amount of
floating matter is to be quantified, then a third step is required to address
the question of how much is that “something.” Here, because Step 3 is
relatively easy, the focus is on Steps 1 and 2.
Using image examples and simulations, Hu et al. (2015) showed that
Step 1 relies only on the sensor’s sensitivity (i.e., signal-to-noise ratio or
SNR), while Step 2 requires specific spectral bands depending on the
targeted floating matter type and on the selected algorithms. Although
that study was targeted on Sargassum, the principle is the same for
marine debris and all other floating matters, therefore these steps will be
used for the sensitivity analysis below.
3.1. Sensitivity analysis
To be able to “stand out” in an image, a pixel needs to be significantly
different from the surrounding pixels. Mathematically, this requires (Qi
and Hu, 2021):
√̅̅̅
ΔR > 2 2σ (1)
√̅̅̅
where σ is the sensor noise in a pixel, 2 is to account for noise
propagation in pixel differencing between the target pixel and nearby
reference pixel (in this case, noise is the square root of sum squares from
√̅̅̅
two pixels, therefore the 2 term), 2 is to make the difference statisti­
cally significant (i.e., 2 times noise), and ΔR is the difference between
the target pixel and nearby reference pixel (i.e., water pixel):
ΔR = RT –RW = [χ RFM + (1 − χ) RW ]–RW = χ(RFM –RW ) (2)
where “T” stands for target, “FM” is for floating matter, “W” is for water,
and χ (0.0%–100%) is the subpixel proportion of floating matter. For
simplicity, the wavelength dependence of R is omitted.
From Eqs. 1 and 2, once σ is known, the subpixel detection limit, χdet,
can be estimated as:
√̅̅̅ /
χdet ≥ 2 2σ (RFM –RW ) (3)
From Eq. 2, assuming the endmember spectra of RFM and RW are
relatively stable, the spectral shape of ΔR is determined between RFM
and RW with equal weights, and the shape does not change with χ. In
other words, both RFM and RW contribute to ΔR with the same weights
regardless of χ. In contrast, their weights to RT are not equal but
determined by χ and (1-χ), respectively. This makes the spectral shape of
RT being dominated by RW when χ is very small (e.g., < 5% or < 10%) as
for the case of marine debris.
In practice, because the spectral contrast between floating matters
and water is mostly in the red-NIR-SWIR wavelengths, χdet can be
4
Remote Sensing of Environment 259 (2021) 112414
estimated using a single wavelength in the NIR (Eq. 3), or a combination
of these wavelengths (e.g., floating algae index, Hu, 2009). Because the
latter involves more bands and therefore more noise (Qi and Hu, 2021),
the lower detection limit is from a single band in the NIR, where the
spatial contrast is the highest between floating matters and water.
Once a pixel is determined to contain a certain type of floating
matter, there are several ways to discriminate the type, including a
similarity index between the pixel’s ΔR and (RFM – RW) where RFM and
RW are from the established spectral library (e.g., Figs. 1 & 2) and water
pixels, respectively. Because the spectral shapes of RW change the most
in the blue-green wavelengths in natural waters, the simplest way is to
restrict the similarity analysis to the red-NIR-SWIR wavelengths of RFM.
Comparing Figs. 1 and 2a, the most prominent difference between ma­
rine debris and floating algae occurs between 670 nm and ~ 750 nm: the
former is spectrally flat while the latter has a sharp increase in the NIR (i.
e., the red-edge reflectance). Therefore, the reflectance difference be­
tween the NIR and red bands (Richardson and Everitt, 1992) can be used
to discriminate marine debris from floating algae:
NRD = RNIR − Rred
Here NRD stands for NIR-red difference. For a pixel containing χ
floating matter and (1-χ) water, there is.
[( ) ( )]
ΔNRDFM = ΔRNIR − ΔRred = χ RFM NIR − RFM red + RW NIR –RW red (4)
Then, to be able to separate marine debris (MD) and floating algae
(FA), their difference in ΔNRD should be significantly higher than noise,
i.e.,
[( ) ( )]
ΔNRDFA − ΔNRDMD = χ RFA NIR –RFA red − RMD NIR –RMD red >> noise
= 2×2 σ
(5)
Here the first 2 in Eq. 5 represents statistical significance, and the
second 2 in Eq. 5 represents the cumulative noise using the square-root
√̅̅̅
rule (two bands, two types, therefore 4). The discrimination limit is
therefore:
/[( ) ( )]
χdis ≥ 4σ RFA NIR –RFA red − RMD NIR –RMD red (6)
Comparing with Eq. 3, Eq. 6 is very similar except for the factor of 4
√̅̅̅
instead of 2 2 because of more spectral bands involved.
3.2. Simulation experiment
In the experiment, reflectance of a pixel mixed with floating matter
and water (i.e., mixed pixel) was estimated with their endmember
spectra and χ. Then, the pixel’s spectra were compared with the end­
member spectra to determine their spectral similarity.
The similarity between two spectra was estimated using a spectral
angle measure (SAM, Kruse et al., 1993). The choice of SAM over other
similarity measures is because SAM is based on spectral shape only. This
is important because χ for marine debris or other floating matters is
often very small and also variable, thus the reflectance magnitude of the
mixed pixel should be deemphasized.
Mathematically, SAM is the angle between two spectral vectors,
defined as (Kruse et al., 1993):
[( ∑ )/(√̅̅̅̅̅̅̅̅̅̅̅̅ ∑ )]
∑ √̅̅̅̅̅̅̅̅̅̅̅̅
SAM (degrees) = cos− 1 xi yi x2i y2i (7)
where x and y represent two spectra and the summation is for band
number i from 1 to N. SAM = 0o means two parallel spectra in log space
(i.e., identical spectral shapes), while SAM = 90o means perpendicular
spectra (i.e., completely different spectral shapes). SAM < 5o indicates
that the two spectra are very similar (Garaba and Dierssen, 2018).
Four (4) endmember spectra were selected in the experiment:
Sargassum (Fig. 2a); plastic bags (Fig. 1); clear water; turbid water
C. Hu Remote Sensing of Environment 259 (2021) 112414

(Fig. 2c). While in nature other types of floating matters also exist and /
water reflectance can also change in both shape and magnitude, for χdis H ≥ 4σH 0.25 ≈ 0.3%
demonstration purpose, the first two are used to represent floating /
vegetation and floating debris, respectively, and the last two are used to χdis MSI ≥ 4σMSI 0.25 ≈ 1.0% (9)
represent typical clear open-ocean waters and turbid coastal waters, These estimates are based on the assumption that 1) for both floating
respectively. plastics (or other debris) and floating algae, their NIR reflectance is
The hyperspectral data of the 4 endmembers were first resampled to ~0.25 for χ = 100% (Fig. 1b & Fig. 2a), and 2) floating plastics (or other
MSI wavelengths using their relative spectral response (RSR) functions, debris) are spectrally flat in the red and NIR wavelengths (Fig. 1b).
and then mixed using different subpixel coverage (χ from 1% to 20%). Indeed, Garaba and Dierssen (2018 & 2020) showed that wet micro­
Then, both the mixed spectra, Rχ, and their contrasts from water, ΔRχ, plastics pieces have reflectance ~0.25 in the NIR and < 0.1 in the SWIR
were compared with the 2 floating matter endmembers to determine wavelengths for χ = 100%, and all spectra are rather flat between red
their similarities using Eq. 7. and NIR wavelengths. Therefore, such assumptions appear realistic.
In the above simulation experiment, because MSI bands have Then, from Eqs. 8 & 9, the detection and discrimination of microplastics
different spatial resolutions, the same experiment was conducted twice. and other marine debris are discussed separately below.
The first used imaginary MSI bands where their resolutions were all set
to 10 m. The second used realistic resolutions for individual bands 4.1.1. Microplastics
(either 60, 10, or 20 m). In the latter case, if a 10-m band had χ = 20%, From 11,854 surface trawls between 1971 and 2013, van Sebille
the 20-m band had χ = 20% / 4 = 5%. Therefore, χ varied between bands et al. (2015) compiled and analyzed microplastics distributions in global
in the same mixed-pixel spectra, causing distorted spectral shapes (see oceans (microplastics defined by particle size <5 mm, see also Garaba
below). and Dierssen, 2018). The dominant majority showed surface density of
<1 M pieces km− 2, and nearly the entire data archive showed <10 M
4. Results pieces km− 2. So far, the highest reported density is 26 M pieces km− 2
(Fig. 5 of Matthews et al., 2017). In an independent study, Eriksen et al.
4.1. Sensitivity (2014) compiled a dataset of marine plastic debris measured at 1571
stations from 680 net tows and 891 visual survey transects. They also
Table 1 shows σ from the proposed NASA mission (HyspIRI, showed that the dominant majority of all compiled particle density and
currently Surface Biology and Geology or SBG) assuming an SNR of 200, modeled particle density is <1 M pieces km− 2 for particles <4.75 mm in
and σ estimated from MSI measurements over clear-water scenes (Qi and size. Therefore, it is reasonable to assume that the maximum density of
Hu, 2021) following the approach of Hu et al. (2012). Here, σ represents microplastics in natural waters is ~10 M pieces km− 2. In another study
noise estimated from R instead of total at-sensor radiance. Only several by Cózar et al. (2014), the size of plastics from a compiled dataset
MSI bands in the green, red, and NIR wavelengths are listed because showed a log-normal distribution with most particles of <5 mm and the
these are the most relevant bands to detect floating matters. Rt,typical is histogram mode of ~2 mm.
the typical total reflectance over oceans under cloud-free and glint-free Assuming a mean size of 2.5 mm per piece, the maximum density of
conditions (Hu et al., 2012). MSI SNRs are lower than the proposed 10 M pieces km− 2 is equivalent to about 50 m2 microplastics km− 2 (i.e.,
HyspIRI SNRs, and the corresponding σ is 2–4 times higher than the χ = 0.005% of a pixel) if all pieces are laid on the very surface without
proposed HyspIRI σ. For simplicity, σMSI in the NIR is assumed to be blocking each other. Clearly, χ = 0.005% is << χH det (0.2%) and also <<
mean + 2 standard deviations (6 × 10− 4); Likewise, σH in the NIR is χMSI − 5
det (0.8%, Eq. 8). For χ = 0.005%, ΔR = 0.25 × χ ≈ 1 × 10 . Such a
assumed to be 2 × 10− 4. signal, corresponding to the maximum microplastics density reported in
Then, for a HyspIRI-like sensor, Eq. 1 suggests that in order for a pixel the literature, is 60 times lower than the required 6 × 10− 4 and 20 times
to “stand out” from the nearby background water pixels, ΔR needs to be lower than sensor noise for a sensor with SNRs of 200. In turn, in order
> ~ 6 × 10− 4. For MSI-like sensors with σ ~ 6 × 10− 4, ΔR needs to be > for microplastics particles to be detected, their density needs to be at
~ 2 × 10− 3. Assuming RNIR NIR
FM ≈ 0.25 (Figs. 1b & 2a) and RW ≈ 0 (Fig. 2c), least >600 M pieces km− 2 or 600 pieces m− 2. Even though, all these
from Eq. 3, the following lower detection limits can be derived for a particles need to be aggregated on the very surface without blocking
HyspIRI-like sensor and for MSI, respectively: each other in order to achieve a maximum reflectance signal.
√̅̅̅ / In the marine environment, because microplastics particles may
χdet H ≥ 2 2σH (RFM –RW ) ≈ 0.2%
actually be below water surface due to mixing or other processes
√̅̅̅ / (Kukulka et al., 2012; Kooi et al., 2016; Martinez-Vicente et al., 2019),
χdet MSI ≥ 2 2σMSI (RFM –RW ) ≈ 0.8% (8) their NIR reflectance can be much lower than when they are all aggre­
gated at the ocean surface. Therefore, the requirement is even higher
Similarly, assuming RNIR red NIR red
FA – RFA ≈ 0.25 (Fig. 2a) and RMD – RMD ≈ 0.0 than shown above, reinforcing the argument that remote detection of
(Fig. 1b), from Eq. 6, the following lower discrimination limits can be microplactics is nearly impossible.
derived for a HyspIRI-like sensor and for MSI, respectively: The requirement of >600 microplastics pieces m− 2 or χ > 0.2% for a
sensor with SNRs of 200 is way beyond the reported maximum density of
10 pieces m− 2. For the same SNRs, in order to detect a single micro­
Table 1 plastics piece with an average size of 2.5 mm, the pixel size must be 25
Reflectance noise (σ) used in the sensitivity analysis. Rt,typical is typical top-of- cm2 (i.e., 5-cm resolution) or smaller. Even if such a resolution is
atmosphere reflectance over the ocean (Hu et al., 2012). “H” is for HyspIRI possible, its SNRs are usually much lower than 200. Therefore, for all
specification. σMSI is from Qi and Hu (2021) estimated from clear-water scenes. existing and future non-military satellite sensors, such a requirement is
For simplicity, σMSI in the NIR is assumed to be mean + 2 standard deviations,
simply impossible to meet. Such a conclusion would not be altered even
about 6 × 10− 4. Likewise, σH in the NIR is assumed to be 2 × 10− 4.
if the microplastic particle size were assumed to be the maximum of its
Band (nm) Rt,typical SNRH σH σMSI category (5 mm) instead of the assumed average size of 2.5 mm. In this
(×10− 2) (×10− 4) (×10− 4)
case, the enhanced reflectance due to microplastics is 4 times higher, but
560 6.90 200 3.45 7.14 ± 2.70 still much lower than sensor noise.
665 3.71 200 1.86 6.58 ± 3.92 The take home message from the sensitivity analysis is simple: the
741 2.73 200 1.37 4.15 ± 1.00
865 1.85 200 0.93 3.70 ± 0.85
enhanced signal due to microplastics at the reported maximum density

5
C. Hu
even with all particles on the very surface without blocking each other
(in reality, this is unlikely to happen due to mixing or other processes in
the ocean) is 60 times lower than the required signal and 20 times lower
than sensor noise (assuming SNRs = 200). To make them detectable, the
particle density needs to be at least 60 times higher than the reported
maximum, or the pixel size must be 25 cm2 (5-cm resolution) or smaller.
In the end, it would be extremely difficult, if not completely impossible,
to remotely sense microplastics from space using vis-NIR wavelengths.
Note that the above sensitivity analysis is only on whether a spatial
anomaly can be detected (i.e., is there “something”?) because if the
presence/absence detection is impossible, there is no need to discuss
discrimination.
These findings are certainly disappointing. However, a critical
assumption in the above calculations is that the microplastics pieces are
uniformly distributed on the surface. When the assumption is violated,
for example when the particles are heavily concentrated along narrow
ocean fronts, windrows, or in small-scale eddy convergence zones so
that particle density is >0.2% (i.e., > 600 m− 2), it might be possible to
detect the reflectance anomalies and therefore presence/absence of
microplastics particles.
The estimates above are based on the assumed SNR of ~200, as
proposed for the HyspIRI mission (currently SBG). High spatial-
resolution sensors typically have SNRs much lower than 200, resulting
in much higher sensor noise (Table 1). In such realistic cases, the
detection limit is also higher, for example χdet > 0.8% (or particle
density > 2400 m− 2) for MSI.
Finally, the above arguments are purely from the perspective of in­
strument sensitivity. In some special cases, microplastics may aggregate
among other larger floating matters, for example Sargassum. Because we
can currently estimate Sargassum density and distributions using both
coarse- and medium-resolution satellite sensors (Wang and Hu, 2016, &
2020; Wang et al., 2018), if a relationship between microplastics and
Sargassum density can be established from field surveys, the relationship
may be applied to the synoptic observations of Sargassum in the Atlantic
to estimate the total amount of microplastics around these large mac­
roalgae mats.
4.1.2. Macroplastics and other debris
Although made of different materials, both macroplastics (> 5 mm)
and other non-plastic debris have broad-band spectral response (e.g.,
Fig. 1b) similar to microplastics, and thus may be difficult to differen­
tiate from each other spectrally. For this reason, they are treated as the
same type, termed as macro debris.
Similar to the detection of microplastics, Step 1 in macro debris
detection is also to detect a spatial anomaly, where the requirement on
subpixel coverage is the same: χdet > 0.2% for a SNR of 200, assuming
macro debris is on the very surface as opposed to be submersed in water.
For a 10-m pixel, this means that the macro debris patch within the pixel
must be at least >0.2 m2. For MSI, the detection limit is χdet > 0.8% or
0.8 m2. Once a spatial anomaly is detected, spectral analysis is still
required in Step 2 to tell whether the anomaly is due to macro debris or
other floating matters (i.e., floating algae). From Eq. 9, χdis needs to be
>0.3% for a HyspIRI-like sensor, and > 1% for MSI. For a 10-m pixel,
this means 0.3 m2 and 1 m2, respectively.
Although still difficult, these detection and discrimination limits can
certainly be met under certain circumstances, for example around river
mouths, in frontal convergence zones, or along windrows of the ocean,
as shown in various digital photos of Biermann et al. (2020) and Kikaki
et al. (2020). However, in practice, as shown below, while detection and
discrimination of floating algae and non-algae features are possible,
discrimination of macro debris type is actually more demanding than
shown above due to spectral similarity among different types of macro
debris.
6
Remote Sensing of Environment 259 (2021) 112414
4.2. Simulation experiment: Spectral shape variations
While Section 4.1 is focused on the sensor’s sensitivity or SNRs to
detect and discriminate marine debris, this section illustrates how
spectral shape changes with spectral endmember, water type (clear or
turbid), and χ. Fig. 3 illustrates how R and ΔR from a mixed Sargassum-
water pixel change with χ for both clear and turbid waters. The same for
the “plastic bags” endmember is presented in Fig. 4. In these simulations,
Sargassum endmember spectrum is from Fig. 2a, plastic endmember
spectrum is from Fig. 1b (“plastic bags”), and clear/turbid water end­
members spectra are from Fig. 2c.
In both figures, the left column represents the results with imaginary
MSI bands all having 10-m resolution, while the right column is for the
results with realistic MSI bands where the 704, 741, 783, and 865-nm
bands have 20-m resolution but other bands (except 443 nm) have 10-
m resolution. In the latter case, the subpixel coverage of floating mat­
ters in the 20-m bands is only ¼ of the 10-m bands, creating spectral
distortion, for example an artificial local maximum at 841 nm and an
artificial local minimum at 704 nm. Worse than this, the degree of the
spectral distortion varies with water type and floating matter type as
well as χ. Thus, it would be difficult to use spectra of mixed pixels to
represent spectral endmembers, especially when χ is small (e.g., ~20%
in Topouzelis et al. (2019 & 2020), Kikaki et al. (2020), and Biermann
et al. (2020)). It is believed that in nearly all published papers on the use
of MSI to detect marine debris, the resolution mismatch among different
bands was not considered enough, and the spectral endmembers in the
published papers appear to be distorted in a similar fashion as shown
above, thus may result in large uncertainties in their interpretations.
The second observation from Figs. 3 & 4 is that, consistent with the
argument in Section 3.1, the use of ΔR is superior to R when evaluating
spectral shapes of mixed pixels. Indeed, in the log-scale plots, the
spectral shape in ΔR does not change with χ regardless of whether
imaginary or realistic MSI bands are used as long as ΔR is positive.
The third observation is that, when the blue bands of 443-nm and
492-nm are excluded, the spectral shape in ΔR of mixed pixels is also
stable between clear and turbid waters (i.e., the empty and solid symbols
almost overlap with each other). This has significant implications to
applications of satellite imagery in different water environments, as
water type (i.e., clear or turbid) may be excluded from consideration
when performing spectral analysis.
4.3. Simulation experiment: spectral similarity
While Figs. 3 & 4 show the spectral shapes after spectral mixing,
Figs. 5 & 6 show the spectral similarity measures between the mixed
pixel and floating matter endmember, expressed in SAM (Eq. 7). Lower
SAM values indicate higher similarity. To understand whether SAM is
different between the two endmembers (Sargassum and plastic), each
mixed pixel is compared to both endmembers.
From these results, the following can be summarized.
One, consistent with the findings from Figs. 3 & 4, ΔR (solid bars) is
way more efficient than R (empty bars) in differentiating floating mat­
ters, either Sargassum or plastic (i.e., compare the empty and solid bars
in the left column of both Figs. 5 & 6).
Two, such an ability is compromised for MSI spectra with mixed
band resolutions because their spectral shapes are distorted to the var­
iable χ in different bands (right column in both Figs. 5 & 6). Indeed, with
mixed band resolutions, it is difficult to determine whether the
Sargassum-containing pixel is spectrally more similar to Sargassum
endmember or to plastic endmember (Fig. 5 right column), or whether
the plastic-containing pixel is spectrally more similar to plastic end­
member or to Sargassum endmember (Fig. 6 right column).
Three, in contrast, if all MSI bands are forced to have the same 10-m
resolution, Sargassum-containing pixels and plastic-containing pixels
can be easily separated through comparing their SAM values to both
endmembers, and such a separation is possible down to at least 1%
C. Hu
subpixel coverage. This is shown by the solid bars between the two
colors in the left column of both Figs. 5 and 6.
Although the simulation experiment used only two endmembers of
Sargassum and plastic, because their spectral shapes can represent
floating algae and macro debris, respectively, the findings above can
help guide spectral analysis and algorithm development when applying
MSI imagery to detect marine debris and other floating matters, as
shown below. Indeed, most floating algae have similar red-edge reflec­
tance and similar 670-nm reflectance trough as in the Sargassum end­
member (Fig. 2), and nearly all macro plastics (except those monotonic
green or blue plastic bottles, Garaba and Dierssen, 2020) have similar
flat spectral shapes as those shown in Fig. 1b. Considering that hyper­
spectral and high-resolution sensors are currently unavailable, the most
achievable appears to be separating floating algae from non-algae while
further differentiating the non-algae type (e.g., plastics or non-plastic
debris or other objects) is rather challenging, as shown below.
4.4. Practical considerations
The above sensitivity analysis (Eqs. 8 and 9) is based on the ideal
situations where image noise is assumed to come from the sensor noise
only, and spectral shapes in either the marine debris endmembers or the
floating algae endmembers are stable. Therefore, both χdet and χdis
represent the lower bounds, i.e., below which detection and discrimi­
nation are impossible, but above which whether or not they are possible
still depend on other factors.
For example, the detection limit is estimated from a single band in
the NIR because this is where the maximum contrast occurs between
floating matters and water, but single-band images are difficult to
interpret as reflectance of the water background may change substan­
tially across the image (Wang and Hu, 2016). The use of floating algae
index (FAI, Hu, 2009) or other indexes (e.g., FRGB, Qi et al., 2020) may
7
Remote Sensing of Environment 259 (2021) 112414
Fig. 3. Simulated mixing experiment for Sargassum
using Sentinel-2 MSI band settings. The blue and
green dashed lines represent clear water and turbid
water, respectively. The legend shows χ from 1% to
20% for the 10-m bands. Left column represents re­
sults with imaginary MSI bands where all bands have
the same 10-m resolution. Right column represents
results with realistic MSI bands where the following
bands have 20-m resolution: 704, 741, 783, and 865
nm, and the 443-nm band has 60-m resolution; (a) R
for mixed pixels; (b) ΔR between mixed pixels and
water endmembers; (c) same as in (b) but ΔR is
plotted in log scale. Note the artificial peak at 841 nm
in the right column due to spectral distortion caused
by mixed band resolutions. (For interpretation of the
references to colour in this figure legend, the reader is
referred to the web version of this article.)
facilitate image visualization, but the detection limit may be compro­
mised to a higher value, for example to ~1% for floating algae with a
SNR of 200 (Hu et al., 2015). Likewise, in practice, a single pixel above
the detection limit is difficult to interpret (e.g., Gower et al., 2014, also
see Fig. 7). Rather, a spatially coherent feature (e.g., an image slick)
from at least several pixels is required, thus requiring higher χ than the
sensitivity-based estimates of 0.2% and 0.3% (for SNRs of 200) and 0.8%
and 1% (for MSI). For example, χdet and χdis for MSI may be 2–3 times
higher than the sensitivity-based estimates, being 2% and 3%,
respectively.
Furthermore, different sensors have different artifacts, which should
be considered carefully when analyzing the spatial and spectral anom­
alies. One example is the hardware parallax effect in push-broom sensors
such as MSI (ESA, 2015) and Landsat-8 OLI, where different bands are
not co-registered in time for a given pixel. Such an effect can create
colorful pepper noise in RGB composite imagery over moving targets (e.
g., waves, Fig. 7) when sun glint or sky glint is apparent (Liu et al.,
2020). Other disturbing image features include pepper noise due to
cosmic rays (Gower et al., 2014) and waves that can also confuse al­
gorithms to detect small features (Wang and Hu, 2020). Under these
circumstances, detecting marine debris becomes more difficult because
although a noise removal algorithm may be used to remove these arti­
facts (Wang and Hu, 2020), the same algorithm may also remove the
small features caused by marine debris as they are expected to be at
subpixel scales.
Similar to the sensitivity analysis, the simulation results in Sections
4.2 and 4.3 are also simplified to demonstrate the concept of 1) why ΔR
is preferred over R to differentiate floating matter type and 2) for MSI,
why the use of single pixels is not a practical way for spectral discrim­
ination even if ΔR is used. In real applications, other types of floating
matters as well as spectral modulations by image noise also need to be
considered. However, even such conceptual demonstrations may
C. Hu
Fig. 4. Same as Fig. 3, but the endmember is “plastic bags” (Fig. 1b) instead of Sargassum. Note the artificial peak at 841 nm in the right column due to spectral
distortion caused by the mixed band resolutions. Such a peak does not exist if all bands have the same resolution (left column).
provide some guidance on how to perform the spectral analysis.
For example, to avoid MSI spectral distortion in single pixels, mean
spectra from 5 × 5 pixels instead of a single pixel may be used to derive
the spectral shape in a more reliable way, as shown in Fig. 8. Although
the magnitude of the mean spectra is lower than from some of the 5 × 5
pixels, the spectral shape of the floating matter is largely retained, thus
facilitating analysis for spectral similarity. The two MSI images in Fig. 8
were collected from the southern Caribbean Sea where marine debris
has been reported (Kikaki et al., 2020). However, from the spectral
shapes alone, it is difficult to conclude whether the pixels contain ma­
rine debris, especially for the first case in Fig. 8a.
In Fig. 8a, the image slick appears greenish, indicating high NIR
reflectance because in the FRGB image, a NIR band is used to represent
the green channel. The mean ΔRrc spectrum in Fig. 8b from 5 × 5 pixels
(centered at the red “x” location) indeed shows red edge reflectance in
the NIR, strongly suggesting floating vegetation and very likely
Sargassum. Its SAM to the known Sargassum endmember is 3.8o, but 7.5o
to the plastic endmember. In contrast to the mean ΔRrc spectrum, ΔRrc
from 3 randomly selected pixels at the same location (Fig. 8b) show high
spectral noise, making them difficult to interpret.
Similar observations are obtained in Fig. 8c where the image slick
appears whitish, indicating lack of red-edge reflectance. The 5 × 5 mean
ΔRrc spectrum from the feature indicates non-algae materials (Fig. 8d).
Its SAM to plastic is half of that to Sargassum (8.2o versus 18.5o), but its
NIR reflectance is lower than the green-red reflectance. This charac­
teristic actually does not appear to resemble plastic, but more similar to
foam (whitecap) spectra (Dierssen, 2019), especially when considering
that the local minima at 741 nm are near the reported 756-nm minimum
in whitecap reflectance. However, it is also possible that the feature is a
8
Remote Sensing of Environment 259 (2021) 112414
mixture of marine debris and foam, or marine debris alone, especially if
the debris is partially submersed in water to cause a reflectance decrease
in the NIR wavelengths.
The results in Fig. 8 are for demonstration purpose only, as in reality
the plastic endmember may have slightly different spectra than used
here. If the plastic endmember spectra were replaced by the microplastic
spectra collected by Garaba and Dierssen (2020) (red curve in Fig. 1b),
the SAM to plastic would be 9.8o (Fig. 8a) and 5.3o (Fig. 8c), respec­
tively. This new result actually makes it easier to discriminate the two
image slicks between Sargassum-like and plastic-like, but it does not alter
the principles presented above.
Note that the reflectance magnitude in both cases is very small.
Assuming a NIR reflectance of 0.25 in both endmembers for χ = 100%,
Fig. 8b suggests mean χ ~ 6% while Fig. 8d suggests mean χ ~ 2%. Even
though, the slick features can be detected and discriminated.
From the analysis demonstrated here, once a spatially coherent
feature is detected in the MSI imagery, two techniques are recom­
mended to avoid the spectral “distortion” in spectral analysis or algo­
rithm development. The first is to use ΔR instead of RT in Eq. 2. This
way, both floating matter and water contribute equal weights to ΔR
regardless of χ (Figs. 3c & 4c). The second is to average several pixels to
minimize the impact of resolution mismatch. Without averaging, the
individual pixels in Figs. 8b & 8d show spectral shapes that are highly
noisy, making them impossible to compare with known floating matter
endmember spectra.
4.5. Case study over the West Florida shelf
With the findings above, a case study using MSI data is presented
C. Hu Remote Sensing of Environment 259 (2021) 112414

Fig. 5. Spectral similarity between a mixed Sargassum pixel and Sargassum endmember (orange bars), and between a mixed Sargassum pixel and plastics endmember
(blue bars) using the following MSI bands: 560, 665, 704, 741, 841, 865 nm. (a) and (b): all MSI bands are 10 m; (c) and (d): the following bands are 20 m: 705, 741,
and 865. The solid orange bars (annotated with arrows) in (a) and (b) are close to 0 degrees, indicating high similarity. These results correspond to the spectra shown
in Fig. 3. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. Same as Fig. 5, but the mixed pixel is between water and plastic. The solid blue bars in (a) and (b) (annotated with arrows) are close to 0 degrees, indicating
high similarity. These results correspond to the spectra shown in Fig. 4. (For interpretation of the references to colour in this figure legend, the reader is referred to
the web version of this article.)

9
C. Hu Remote Sensing of Environment 259 (2021) 112414

as indicated by the 665-nm absorption feature (arrows in Figs. 9j & 9k).

Fig. 7. MSI FRGB image near Japan (centered around 31.7593oN 142.2510◦ E)
showing colorful pixels due to the sensor parallax effect. The spectral shapes of
these pixels in the inset figure appear abnormal as they do not resemble any
known spectral shapes, including those from marine debris.
here to demonstrate how to detect image features (spatial anomaly) and
how to discriminate the feature type.
For simplicity, the first step in remote detection of marine debris, i.e.,
detecting a spatial anomaly, is through visual inspection while sophis­
ticated image segmentation may be implemented in the future (e.g.,
Wang and Hu, 2020). Fig. 9a shows an MSI RGB sub-image west and
northwest of Tampa Bay (Florida, USA), where the full-resolution image
in several areas (denoted as “1” – “4”) reveals image slicks (Figs. 9d – 9
g). These slicks indicate the presence of floating matters.
Spectral analysis of representative pixels from the slicks, through the
use of ΔRrc as in Fig. 8, indicate different spectral shapes from these
slicks (Figs. 9h – 9 k). While slicks in Areas 1 & 2 show flat spectral
shapes, those in Areas 3 and 4 show typical shapes from floating algae,
Digital photos taken on the same day in Area 4 confirms that these slicks
are floating Trichodesmium mats (Fig. 9b). Indeed, even without such a
field validation, from knowledge of regional oceanography (Hu et al.,
2010) and from the spectral shapes in Figs. 9j & 9k, one can still
conclude that these slicks are very likely due to surface aggregation of
Trichodesmium.
In contrast, although it is clear that the slicks in Areas 1 & 2 are not
floating algae, it is very difficult to discriminate their type based on the
spectral shapes. The 443-band may be excluded because the spectral
distortion by this 60-m band cannot be removed even after 5 × 5 pixel
averaging. Then, except for the residual band-resolution effect in several
NIR bands (704-nm, 741-nm, 783-nm, 865-nm, all having 20-m reso­
lution), all spectra are featureless (Figs. 9h & 9i). The question becomes
what could cause these image slicks, marine debris or other floating
matters. Unfortunately, there is no easy answer.
First, foams or white caps may be ruled out because these slicks show
red-rich spectra (in contrast, whitecaps are blue-rich, Dierssen, 2019)
and because the ocean was very calm (wind is mostly <3 m s− 1 for two
consecutive days before the imaging time, Fig. 9c). Weather data
showed almost no rainfall prior to the imaging time, and there is no
major river nearby, suggesting these materials are unlikely to be riverine
origin. Furthermore, there was no news report of large debris patches in
this region, and routine cruise surveys to this region in the past (almost
monthly, as part of the red tide monitoring effort) never encountered
large debris patches in this region. Therefore, although from the
perspective of image spectroscopy these slicks appear like marine debris,
it is difficult to conclude without a direct field validation, let alone
further differentiating the debris type (e.g., plastic or non-plastic). At the
present, whether the slicks in Areas 1 and 2 are due to marine debris or
other type of floating matters remains to be a puzzle. On the other hand,
detecting and discriminating floating matters (algae versus non-algae)

Fig. 8. (a) MSI FRGB image on 29 November 2015 in the SW Caribbean Sea. The red “x” (16.0649oN, 86.4025oW) and black “x” symbols mark the locations where
spectra were extracted for analysis in (b). The Rrc spectra of 5 × 5 pixels from the two locations are shown to the right y-axis, while their ΔRrc spectra are shown to the
left y-axis. In addition to the 5 × 5 mean spectrum (solid green circle), spectra from 3 individual pixels are also shown. (c) MSI FRGB image on 30 August 2018 in the
SW Caribbean Sea. The red “x” (15.7479oN, 88.1656oW) and black “x” symbols mark the locations where spectra were extracted for analysis in (d), where the spectral
legend is the same as in (b). Note that the spectral shapes from individual pixels are all distorted (compared to the mean spectral shapes) due to mixed band res­
olutions. Annotated on the images are the SAM values to Sargassum and plastic endmembers, respectively, and mean χ of the marked pixels. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

10
C. Hu Remote Sensing of Environment 259 (2021) 112414

Fig. 9. An example showing the possibility of detecting and discriminating floating algae and non-algae matters as well as the challenge in identifying the type of
non-algae floating matters using MSI data. (a) MSI RGB image on 10 February 2021 over part of the west Florida shelf (NW off Tampa Bay); (b) Digital photo taken on
the same day in area “4” of the image showing surface Trichodesmium mats (photo credit: Charles Tilney and Matt Garrett of Florida Fish and Wildlife Conservation
Commission; photo location annotated as “X” in (g)); (c) Windspeed measured from a nearby marine buoy, where the marked dates represent the beginning of the day
(0:00 GMT); (d) – (g) Enlarged sub-images corresponding to areas 1–4 in (a), where surface slicks are visible; (h) – (k) ΔRrc spectra (5 × 5 pixels, vertical bars indicate
standard deviations) extracted from MSI pixels of the image slicks in (d) – (g), respectively, where the pixel locations are marked in red. The pigment absorption
features in (j) and (k) are marked with arrows. In (h) – (k), mean χ of each 5 × 5 pixels is annotated. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)

are shown to be possible for χ ≥ 3% from a group of pixels, a result
consistent with the sensitivity analysis and practical considerations
above.
5. Discussions: Looking forward
5.1. Which wavelengths (bands) to use
While some current and future satellite sensors are equipped with
spectral bands in the SWIR wavelengths and SWIR bands have proven
useful in detecting and quantifying emulsified oil (Clark et al., 2010; Lu
et al., 2020; Hu et al., 2021), this study is focused on the vis-NIR bands for
several reasons. First, the spectral contrast between floating matters and
background waters is mostly in the NIR, regardless of the type of floating
matters (either Sargassum, Ulva, Trichodesmium, or plastics, see references
in Figs. 1 & 2). Reflectance of floating matters in the SWIR bands is several
times lower than in the NIR bands, making it more difficult to detect if
SWIR bands are to be used. This is different from emulsified oil that may
have higher SWIR reflectance than in the NIR. Second, due to mixing and
other processes, floating matters may be slightly below surface, making the
SWIR reflectance quickly disappear due to strong water absorption. For
example, water absorption coefficient is ~130 m− 1 at 1.2 μm and ~ 700
m− 1 at 1.6 μm. If the floating matter is 1 cm below surface, its SWIR
reflectance would be reduced by 14 times at 1.2 μm and 1 million times at
1.6 μm. In contrast, the 765-nm reflectance would be reduced by only 6%.
Furthermore, when implementing a detection scheme, although the
use of a single NIR may maximum the pixel-to-pixel contrast at local
scale, interpretating single-band images is usually difficult because of
relatively large gradience (compared to noise) across the image.
Therefore, band-combination indexes, such as FAI (Hu, 2009), may be
used to mitigate such effects at the price of reduced sensitivity to detect
spatial anomalies. This is exactly the reason why the same 200 SNR led
to the ~1% detectability estimate in Hu et al. (2015) but ~0.2% in this
study. The FAI design can be changed to use different band combinations
depending on band availability and application needs, for example
through the alternative FAI (AFAI, Wang and Hu, 2016; Qi et al., 2017)
or the floating debris index (FDI, Biermann et al., 2020). In the end, a
combination of a red band (665 nm), a NIR band (754 nm), and another
NIR band (865 nm) or a SWIR band (1.2 μm or 1.6 μm) should be suf­
ficient in detecting the presence of floating matters.
Then, the inclusion of a green band (560 nm) with the 665-nm band
and NIR bands makes it easy to calculate SAM in order to discriminate
between floating algae and non-algae floating matters (Figs. 5 & 6). The
chlorophyll absorption feature around 665 nm (arrows in Figs. 2a & 2b,
Figs. 9j & 9k) represents a key difference between most floating algae
and non-algae floating matters, therefore can be used in either the SAM
or other indexes to differentiate the two types. Once an image feature is
being identified as floating algae, spectral shapes in the visible wave­
lengths can be used to further differentiate whether the floating algae is
Sargassum, Ulva, Trichodesmium, or others (Qi et al., 2020) as long as χ is
above the discrimination threshold (Qi et al., 2019).
5.2. Challenges in discriminating the type of non-algae floating matters
Once non-algae floating matters are identified using the above SAM-
based or other similar approaches, further discriminating the type of
non-algae floating matters represents a technical challenge as most non-
algae debris appear to be similar in spectral shapes (Fig. 1b, Figs. 9h &

11
C. Hu
9i). This is especially true considering the usually small χ values .
Indeed, the pixels from the same image slick in Fig. 9h are likely to be of
the same type, yet their pixel-to-pixel variations in their spectral shapes
appear to encompass those from different plastic and non-plastic ma­
terials (Fig. 1b, also see more examples in Garaba and Dierssen, 2020).
Such pixel-to-pixel variations also appear to overwhelm the spectral
variations in the visible due to changes in the background waters.
Therefore, at the present, it appears that discrimination of non-algae
floating matters from floating algae is doable but further pinpointing
the non-algae floating matter type based on spectroscopy alone is not.
Adding more spectral bands, for example from hyperspectral measure­
ments, might help, but this is a subject of future research.
One exception might be the separation of plastic versus non-plastic
marine debris, as the former shows specific, narrow, hydrocarbon ab­
sorption features in the SWIR wavelengths (Garaba and Dierssen, 2020).
Once hyperspectral data at those wavelengths are available, it might be
possible to fingerprint floating plastics. This is similar to the use of these
features in detecting and quantifying emulsified oil on the ocean surface
(Clark et al., 2010; Lu et al., 2020; Hu et al., 2021). However, one
drawback is the small magnitude of these features from marine debris,
which may be extremely difficult to detect for the reasons outlined
above.
Despite these difficulties, discriminating marine debris from other
non-algae floating matters may still be possible through non-
spectroscopy methods. For example, most of the non-algae floating
matters are rare in the ocean, with often known locations, thus can be
easily ruled out with some a priori knowledge. Inspection of wind data
can also help rule out the possibility of whitecaps.
On the other hand, although discrimination between floating algae
and floating debris (and further discrimination of the type of floating
debris) is desirable, it is not always necessary in an ecological perspec­
tive. This is because biofouling of marine debris is common, which has
implications on both the ecosystem and the fate of marine debris
(Maximenko et al., 2019). Nevertheless, the feasibility of discriminating
the type of non-algae floating matters needs to be explored more in a
practical sense with the existing satellite data.
5.3. Automation considerations
Based on these observations, it appears possible to implement a step-
wise approach to automate the detection and quantification of both
floating algae and non-algae floating matters (including macro debris),
for example:
Step 1 – detecting spatial anomaly and delineating image features.
This can be based on a single band or a combination of bands using
image segmentation techniques (e.g., Wang and Hu, 2020);
Step 2 – discriminating between floating algae and non-algae floating
matters. This is based on their difference around 665 nm, which can be
quantified through the use of SAM or other indexes. In this step, the
spectral shape around 665 nm is derived from reflectance difference
(ΔRrc) in order to maximize the spectral contrast, where the background
water pixels for the individual slick pixels can be found using a nearest-
neighbor approach (Hu et al., 2000).
Step 3 – quantifying χ in each mixed pixel. This is through the use of
locally tuned lower-bound threshold to represent χ = 0% and pre-
defined upper-bound threshold to represent χ = 100% (e.g., upper
bound for a single NIR band may be 0.25, but for AFAI may be 0.1 ac­
cording to the endmember spectra presented in Figs. 1b & 2a).
In all steps above, the fundamental property is the spectral reflec­
tance derived from satellite measurements. Ideally, atmospherically
corrected surface reflectance should be used to remove the variable ef­
fects due to Rayleigh scattering, gaseous absorption, aerosol scattering,
sun glint, and solar/viewing geometry. However, this is not always
possible from a pixel-wise atmospheric correction approach (e.g.,
currently being used in the NASA processing software SeaDAS and ESA
processing software SNAP) because the presence of floating matter will
12
Remote Sensing of Environment 259 (2021) 112414
violate atmospheric correction assumptions on negligible or predictable
(predicted from the red band) NIR-SWIR surface reflectance (Stumpf
et al., 2003; Wang and Shi, 2007). Therefore, a nearest-neighbor at­
mospheric correction (Hu et al., 2000) or a dark-target based image-wise
atmospheric correction approach, such as that implemented in the
Acolite software (Vanhellemont and Ruddick, 2018), may be used. On
the other hand, because ΔR rather than R is used in Steps 2 and 3, both
Rrc and Rt can be used. This is because atmospheric properties over
adjacent pixels are assumed to be the same, leading to ΔRrc = ΔRt = ΔR.
Automation also requires cloud masking and other steps to mask
pixels that are impossible to determine whether they contain floating
matters. These pixels are treated as no-observation pixels to avoid
biasing statistics. This is perhaps more difficult than implementing the
above steps, as there is no universal way to mask these pixels. For
example, the standard cloudmasking algorithm in SeaDAS uses a
threshold of Rrc(865 nm) > 0.027 to mask clouds, and the modified al­
gorithm uses a threshold of Rrc(2130 nm) > 0.0175 to mask clouds
(Wang and Shi, 2006). Because of the enhanced NIR and SWIR reflec­
tance due to floating matters, these cloud masking schemes may falsely
mask some floating matter pixels as clouds. To overcome this difficulty,
customized cloudmasking algorithms have been used for different sen­
sors (Wang and Hu, 2016 & 2020), yet their global applicability needs to
be evaluated. Likewise, cloud shadows in high-resolution imagery,
among other “artifacts”, also need to be identified and masked. These
detailed image processing requirements are beyond the scope of this
paper, and interested readers may refer to the relevant literature for
more information. In the end, a regional near real-time system may be
developed to monitor both floating algae and non-algae floating matters,
similar to the Sargassum Watch System (SaWS, Hu et al., 2016) estab­
lished for the Atlantic Ocean and Gulf of Mexico.
5.4. “Optimal” optical sensors
In a recent paper, Martinez-Vicente et al. (2019) provided an initial
review of the current satellite sensor capability, including SAR and LIDAR,
with the focus on sensors equipped with NIR and SWIR bands to emphasize
the contrast between marine debris and water in these bands. For passive
optical sensors, none of the existing satellite sensors was designed to
monitor floating matters, especially for the case of marine debris. While
designing a “perfect” optical sensor (including satellite orbital character­
istics) is beyond the scope of this paper, the observations above may pro­
vide some general guide if an “optimal” sensor/mission is to be designed.
The characteristics of passive optical remote sensing can be gener­
alized in the following 4 resolutions: spatial, spectral, radiometric, and
temporal resolutions, with the last two defined by SNRs and site revisit
frequency. The importance of the first three resolutions has been dis­
cussed in this paper, and the last resolution depends on both sensor and
satellite orbital designs. Because there is always a trade-off between all
four resolutions, for a coastal region, an optimal sensor/mission should
be able to detect and discriminate non-algae floating matter patches of
several m2 in size every few days. Such a capacity may enable the sensor
to search for missing fishing gears or large solid objects in the ocean due
to Tsunami or other disasters. Thus, the following may serve as an
optimal trade: 3–4 m spatial resolution, 4–6 spectral bands (443 nm,
560 nm, 620 nm, 665 nm, 754 nm, 865 nm, with 665-nm band having
10–20 nm bandwidth), SNRs of 50–100 at typical ocean radiance inputs,
and revisit frequency of 3–4 days. The 865-nm band is used to form a
baseline with the 670-nm band to calculate AFAI. If only 5 bands are
allowed, the 865-nm band may be removed because the 754-nm and
670-nm bands can be used to calculate the normalized difference
vegetation index (NDVI). The 620-nm band is to differentiate floating
algae colors (green-rich or orange-rich) but can be sacrificed if only 4
bands are allowed. The other 4 bands (443, 560, 670, and 754 nm)
represent the core requirements for effective detection and discrimina­
tion of floating algae and non-algae floating matters.
Currently, the Sentinel-2 MSI sensors almost meet these
C. Hu
requirements, yet the mixed band resolutions degrade their capacity in
discriminating floating matter types. On the other hand, the Planet
Scope/DOVE constellation of hundreds of miniature satellite sensors
(CubeSats) provides 3–4 m resolution data in 3–4 spectral bands with
2–3 day revisit frequency in many coastal areas (http://www.planet.
com), thus almost meeting the “optimal” requirements above. Unfortu­
nately, DOVE spectral bands are too wide (60–90 nm) to discriminate
floating algae against non-algae floating matters. Nevertheless, these
existing satellite sensors, combined with other high-resolution sensors
such as those discussed in Martinez-Vicente et al. (2019), may provide a
practical solution to address the challenges in sensor requirements,
while the challenges in algorithm development, as demonstrated above,
still need to be addressed.
6. Summary
By no means is this paper meant to provide a comprehensive guide on
how to remotely detect marine debris from space especially when
considering its focus on passive optical remote sensing. Instead, the paper
addresses two challenges identified by Martinez-Vicente et al. (2019) on
SNR requirements and methodology requirements, respectively, through
answering the following questions posed earlier: to what extent can marine
debris be remotely sensed, and how?
From the above sensitivity analysis, simulation experiments, and
case studies using MSI imagery, the following observations and sug­
gestions can be generalized:
1) Regardless of the floating matter type, either marine debris, oil
slicks, or floating vegetation, from the perspective of spectroscopy,
remote detection is always through spatial anomaly and remote
discrimination is always through spectral shape similarity as
opposed to spectral magnitude. Both depend on sensor’s SNRs and
band settings.
2) From a theoretical basis, with only 4 bands around 560 nm, 665 nm,
and two NIR wavelengths, both χdet and χdis depend only on sensor
sensitivity (SNRs). They are 0.2% and 0.3% for a sensor with SNR of
200, and 0.8% and 1.0% for MSI. Below these limits, floating matters
are simply not detectable. Considering other practical requirements
(e.g., a spatially coherent image feature instead of a single pixel
needs to “stand out” from the background, these limits may be
increased by 2–3 times in order to detect and discriminate floating
matters. For example, χdet and χdis for MSI may actually be 2% and
3%, respectively (Figs. 8 & 9).
3) The detection limits suggest that detecting microplastics is likely
impossible even for the reported maximum density and even when
all particles are at the very surface. Therefore, future efforts may be
dedicated to frontal zones or windrows only, or to other surrogates
(e.g., Sargassum) that may aggregate microplastics. Whether these
approaches are feasible, however, still requires more research.
4) While the detection of presence/absence of floating matters can be
through single-band images or band-combination indexes with each
having its own strengths and weaknesses, spectral similarity (or
anomaly) needs to be analyzed through the difference spectra (ΔR in
Eq. 2) because this is the only way to retain the spectral shape of the
floating matter endmember when χ is typically small (e.g., <10%).
Such a practice actually started from Gower et al. (2006) when
differentiating Sargassum from Trichodesmium in the Gulf of Mexico.
For the same reason, using spectral shapes derived from mixed pixels
as endmembers are subject to large uncertainties because these
shapes depend not only on the floating matter endmember, but also
on the unknown χ as well as on changes in the water endmember in
the real environment.
5) While detecting macro debris appears possible from MSI, more work
is required to make the previously reported cases more convincing,
as spectral shapes of a mixed pixel can be modulated by the variable
χ, variable water reflectance, mismatch in band spatial resolution,
13
Remote Sensing of Environment 259 (2021) 112414
waves, and sensor artifacts (e.g., parallax effect). This is true even for
the same type of floating matter, not to mention the possibility that
more than one type of floating matter may exist in the same region.
For the reasons mentioned above, in the case of lack of pure spectral
endmember (i.e., χ = 100%), it is better to use ΔR rather than R to
represent the endmember.
6) Because the spectral similarity between macro plastics and non-
plastics debris (i.e., relatively flat, wide-band reflectance in the vis-
NIR wavelengths), it appears difficult to separate them spectrally.
This is also partially due to the lack of a relatively complete spectral
library of various marine debris. More measurements are therefore
required to complement those reported in previous studies focused
on plastics. On the other hand, because marine debris may be a
mixture of different types of materials, discriminating a specific type
may be unnecessary.
7) The spectral analysis here using Sargassum and plastic is for
demonstration purpose only. In the marine environment, other
known (e.g., Fig. 2) and unknown floating matters may exist, and
some of them may have similar spectral shapes as marine debris (e.g.,
Figs. 9d & 9e; the thought-to-be sea jelly schools in Qi et al., 2020). In
the end, it may be impossible to discriminate marine debris from
other unknown floating matters from a perspective of pure spec­
troscopy. In such cases, other ancillary information (e.g., knowledge
of local oceanography, news reports of pollution events, etc.) may be
used to help interpret the detected image features.
8) Although the paper demonstrates under what conditions remote
detection of marine debris might be possible using the reflected vis-
NIR sunlight, it also shows conditions that make such a detection
impossible through the sensitivity analysis. Clearly, there is still a lot
to do after the recent pioneering studies on this topic, including
establishment of a more complete spectral library of various floating
matters and development of more robust algorithms to detect debris-
like features in noisy remote sensing images.
Finally, the paper uses MSI data to demonstrate the concept and chal­
lenges in remote sensing of marine debris in the marine environment. The
arguments may change substantially for remote sensing of marine debris
when they may be heavily concentrated on beaches. In both environments,
sensors with finer spatial resolution such as WorldView (2 m) or Planet­
Scope/Dove (3–4 m) offer a better capacity in detecting smaller debris
patches, although detecting such features in the marine environment re­
quires more effort because these sensors were not designed for marine
applications. In any case, given the increased reports of marine debris and
their increasingly recognized importance in the marine environment, more
effort is required to develop reliable algorithms and feasible approaches to
monitor and track marine debris, thus contributing to a final roadmap in
remote sensing of marine debris (Maximenko et al., 2019).
Credit author statement
Chuanmin Hu: Conceptualization; Methodology; Data curation;
Writing- Original draft preparation, revision, editing; Funding
acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
The authors declare the following financial interests/personal re­
lationships which may be considered as potential competing interests:
Acknowledgements
This work was supported by the U.S. NASA through its Ocean Biology
and Biogeochemistry program (NNX16AR74G, 80NSSC20M0264) and
C. Hu
Ecological Forecast program (NNX17AF57G, 80NSSC21K0422). The
author thanks Drs. Eric Hochberg, Lianbo Hu, Lachlan McKinna, and
Shungudzemwoyo Garaba for providing endmember spectra for Sargassum,
Ulva, Trichodesmium, and microplastics, respectively. The author is grateful
to Dr. Aikaterini Kikaki (National Technical University of Athens, Greece)
for the discussions on marine debris detection in the Caribbean Sea. Three
anonymous reviewers provided insightful comments to help improve the
presentation of this work, whose efforts are appreciated.
References
Arii, M., Koiwa, M., Aoki, Y., 2014. Applicability of Sar to marine debris surveillance
after the great East Japan earthquake. IEEE J. Sel. Top. Appl. Earth ObsRemote Sens.
7, 1729–1744.
Biermann, L., Clewley, D., Martinez-Vicente, V., Topouzelis, K., 2020. Finding plastic
patches in coastal waters using optical satellite data. Sci. Rep. 10, 1–10.
Clark, R.N., Swayze, G.A., Leifer, I., Livo, K.E., Kokaly, R., Hoefen, T., Lundeen, S.,
Eastwood, M., Green, R.O., Pearson, N., 2010. A method for quantitative mapping of
thick oil spills using imaging spectroscopy. US Geological Survey Open-File Report
1167, 1–51.
Cózar, Andrés, Fidel, Echevarría, Ignacio González-Gordillo, J., Irigoien, Xabier,
Úbeda, Bárbara, Hernández-León, Santiago, Álvaro T. Palma, et al., 2014. Plastic
debris in the Open Ocean. Proc. Natl. Acad. Sci. 111 (28), 10239–10244.
Dierssen, H., 2019. Hyperspectral measurements, parameterizations, and atmospheric
correction of whitecaps and foam from visible to shortwave infrared for ocean color
remote sensing. Front. Earth Sci. 7, 14. https://doi.org/10.3389/feart.2019.00014.
Eriksen, M., Lebreton, L.C.M., Carson, H.S., Thiel, M., Moore, C.J., Borerro, J.C.,
Galgani, F., Ryan, P.G., Reisser, J., 2014. Plastic pollution in the world’s oceans:
more than 5 trillion plastic pieces weighing over 250,000 tons afloat at sea. PLoS
One 2014 (9), e111913.
ESA (2015). Sentinel-2 User Handbook. https://sentinel.esa.int/documents/247904/
685211/Sentinel-2_User_Handbook Accessed date: 24 October 2020.
Garaba, S.P., Dierssen, H.M., 2018. 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.
Garaba, S.P., Dierssen, H.M., 2020. Hyperspectral ultraviolet to shortwave infrared
characteristics of marine-harvested, washed-ashore and virgin plastics. Earth Syst.
Sci. Data 12 (77–86), 2020. https://doi.org/10.5194/essd-12-77-2020.
Garaba, S.P., Aitken, J., Slat, B., Dierssen, H.M., Lebreton, L., Zielinski, O., Reisser, J.,
2018. Sensing ocean plastics with an airborne hyperspectral shortwave infrared
imager. Environmental Sci. & Techol. 52, 11699–11707.
Garaba, S.P., Acuña-Ruz, T., Mattar, C.B., 2020. Hyperspectral longwave infrared
reflectance spectra of naturally dried algae, anthropogenic plastics, sands and shells.
Earth Syst. Sci. Data 12, 2665–2678.
Goddijn-Murphy, L., Dufaur, J., 2018. Proof of concept for a model of light reflectance of
plastics floating on natural waters. Mar. Pollut. Bull. 135, 1145–1157. https://doi.
org/10.1016/j.marpolbul.2018.08.044.
Goddijn-Murphy, L., Williamson, B., 2019. On thermal infrared remote sensing of plastic
pollution in natural waters. Remote Sens. 11, 2159.
Goddijn-Murphy, L., Peters, S., van Sebille, E., James, N.A., Gibb, S., 2018. Concept for a
hyperspectral remote sensing algorithm for floating marine macroplastics. Mar.
Pollut. Bull. 126, 255–262.
Gower, J., Hu, C., Borstad, G., King, S., 2006. Ocean color satellites show extensive lines
of floating Sargassum in the Gulf of Mexico. IEEE Trans. Geosci. Remote Sens. 44,
3619–3625.
Gower, J., King, S., Young, E., 2014. Global remote sensing of Trichodesmium. Int. J.
Remote Sens. 35 (14), 5459–5466. https://doi.org/10.1080/
01431161.2014.926422.
Hu, C., 2009. A novel ocean color index to detect floating algae in the global oceans.
Remote Sens. Environ. 113, 2118–2129.
Hu, C., Carder, K.L., Muller-Karger, F.E., 2000. Atmospheric correction of SeaWiFS
imagery over turbid coastal waters: a practical method. Remote Sens. Environ. 74,
195–206.
Hu, C., Cannizzaro, J., Carder, K.L., Muller-Karger, F.E., Hardy, R., 2010. Remote
detection of Trichodesmium blooms in optically complex coastal waters: examples
with MODIS full-spectral data. Remote Sens. Environ. 114, 2048–2058.
Hu, C., Feng, L., Lee, Z., Davis, C.O., Mannino, A., McClain, C.R., Franz, B.A., 2012.
Dynamic range and sensitivity requirements of satellite ocean color sensors: learning
from the past. Appl. Opt. 51, 6045–6062.
Hu, C., Feng, L., Hardy, R.F., Hochberg, E.J., 2015. Spectral and spatial requirements of
remote measurements of pelagic Sargassum macro algae. Remote Sens. Environ. 167,
229–246. https://doi.org/10.1016/j.rse.2015.05.022.
Hu, C., Murch, B., Barnes, B.B., Wang, M., Marechal, J.P., Franks, J., Johnson, D.,
Lapointe, B., Goodwin, D.S., Schell, J.M., Siuda, A.N.S., 2016. Sargassum watch
warns of incoming seaweed. Eos 97 (22), 10–15. https://doi.org/10.1029/
2016EO058355.
Hu, L., Hu, C., He, M.X., 2017. Remote estimation of biomass of Ulva prolifera
macroalgae in the Yellow Sea. Remote Sens. Environ. 192, 217–227. https://doi.org/
10.1016/j.rse.2017.01.037.
Hu, C., Lu, Y., Sun, S., Liu, Y., 2021. Optical remote sensing of oil spills in the ocean:
What’s really possible? J. Remote Sens. 9141902. https://doi.org/10.34133/2021/
9141902.
14
Remote Sensing of Environment 259 (2021) 112414
Kikaki, A., Karantzalos, K., Power, C.A., Raitsos, D.E., 2020. Remotely sensing the source
and transport of marine plastic debris in bay islands of honduras (Caribbean Sea).
Remote Sens. 12, 1727. https://doi.org/10.3390/rs12111727.
Kooi, M., Reisser, J., Slat, B., et al., 2016. The effect of particle properties on the depth
profile of buoyant plastics in the ocean. Sci. Rep. 6, 33882.
Kruse, F.A., Lefkoff, A.B., Boardman, J.W., Heidebrecht, K.B., Shapiro, A.T., Barloon, P.
J., Goetz, A.F.H., 1993. The spectral image processing system (SIPS)—interactive
visualization and analysis of imaging spectrometer data. Remote Sens. Environ. 44,
145–163.
Kukulka, T., Proskurowski, G., Morét-Ferguson, S., Meyer, D.W., Law, K.L., 2012. The
effect of wind mixing on the vertical distribution of buoyant plastic debris. Geophys.
Res. Lett. 39 (L07601).
Law, K.L., Moret-Ferguson, S., Maximenko, N.A., Proskurowsk, G., Peacock, E.E.,
Hafner, J., Reddy, C.M., 2010. Plastic accumulation in the North Atlantic subtropical
gyre. Science 329, 1185–1188.
Liu, Y., Xu, B., Zhi, W., Hu, C., Dong, Y., Jin, S., Lu, Y., Chen, T., Xue, W., Liu, Y.,
Zhao, B., Lu, W., 2020. Space eye on flying aircraft: from Sentinel-2 MSI parallax to
hybrid computing. Remote Sens. Environ. 246, 111867. https://doi.org/10.1016/j.
rse.2020.111867.
Lu, Y., Shi, J., Hu, C., Zhang, M., Sun, S., Liu, Y., 2020. Optical interpretation of oil
emulsions in the ocean – Part II: applications to multi-band coarse-resolution
imagery. Remote Sens. Environ. 242, 111778. https://doi.org/10.1016/j.
rse.2020.111778.
Martinez-Vicente, V., Clark, J.R., Corradi, P., et al., 2019. Measuring marine plastic
debris from space: initial assessment of observation requirements. Remote Sens. 11,
2443. https://doi.org/10.3390/rs11202443.
Matthews, J.P., Ostrovsky, L., Yoshikawa, Y., Komori, S., Tamura, H., 2017. Dynamics
and early post-tsunami evolution of floating marine debris near Fukushima Daiichi.
Nat. Geosci. 10, 598–605.
Maximenko, N., Arvesen, J., Asner, G., Carlton, J., Castrence, M., Centurioni, L.,
Chao, Y., Chapman, J., Chirayath, V., Corradi, P., et al., 2017. Remote sensing of
marine debris to study dynamics, balances and trends. Available online. https://ec
ocast.arc.nasa.gov/las/Reports%20and%20Papers/Marine-Debris-Workshop-2017.
pdf.
Maximenko, N., Corradi, P., Law, K.L., Van Sebille, E., Garaba, S.P., Lampitt, R.S.,
Galgani, F., Martinez-Vicente, V., Goddijn-Murphy, L., Veiga, J.M., et al., 2019.
Toward the integrated marine debris observing system. Front. Mar. Sci. 6, 447.
https://doi.org/10.3389/fmars.2019.00447.
McKinna, L.I.W., 2010. Optical Detection and Quantification of Trichodesmium spp.
within the Great Barrier Reef. PhD dissertation. James Cook University, Australia,
312pp.
Qi, L., Hu, C., 2021. To What Extent Can Ulva and Sargassum Be Detected and Separated
in Satellite Imagery? Harmful Algae doi:10.1016/j.hal.2021.102001.
Qi, L., Hu, C., Wang, M., Shang, S., Wilson, C., 2017. Floating algae blooms in the East
China Sea. Geophys. Res. Lett. 44 https://doi.org/10.1002/2017GL075525.
Qi, L., Tsai, S.F., Chen, Y., Le, C., Hu, C., 2019. In search of red Noctiluca scintillans
blooms in the East China Sea (2019). Geophys. Res. Lett. 46, 5997–6004. https://
doi.org/10.1029/2019GL082667.
Qi, L., Hu, C., MIkelsons, K., Wang, M., Lance, V., Sun, S., Barnes, B.B., Zhao, J., der
Zande, D.V., 2020. In search of floating algae and other organisms in global oceans
and lakes. Remote Sens. Environ. 239, 111659. https://doi.org/10.1016/j.
rse.2020.111659.
Richardson, A.J., Everitt, J.H., 1992. Using spectral vegetation indices to estimate
rangeland productivity. Geocarto International 7 (1), 63–69.
Stumpf, R.P., Arnone, R.A., Gould, R.W., Martinolich, P.M., Ransibrahmanakul, V., 2003.
A partially coupled ocean-atmosphere model for retrieval of water-leaving radiance
from SeaWiFS in coastal waters. In: Hooker, S.B., Firestone, E.R. (Eds.), SeaWiFS Post
Launch Technical Report Series, NASA Tech. Memo. 2003-206892, Vol. 22. NASA
God- dard Space Flight Center, Greenbelt, Maryland, pp. 51–59.
Topouzelis, K., Papakonstantinou, A., Garaba, S.P., 2019. Detection of floating plastics
from satellite and unmanned aerial systems (plastic litter project 2018). Int. J. Appl.
Earth Obs. Geoinf. 79, 175–183.
Topouzelis, K., Papageorgiou, D., Karagaitanakis, A., Papakonstantinou, A.,
Ballesteros, M.A., 2020. Remote sensing of sea surface artificial floating plastic
targets with Sentinel-2 and unmanned aerial systems (plastic litter project 2019).
Remote Sens. 12, 2013. https://doi.org/10.3390/rs12122013.
van Sebille, E., Wilcox, C., Lebreton, L., Maximenko, N.A., Hardesty, B.D., van
Franeker, J.A., Eriksen, M., Siegel, D., Galgani, F., Law, K.L., 2015. A global
inventory of small floating plastic debris. Environ. Res. Lett. 10 (12), 124006.
https://doi.org/10.1088/1748-9326/10/12/124006.
Vanhellemont, Q., Ruddick, K., 2018. Atmospheric correction of metre-scale optical
satellite data for inland and coastal water applications. Remote Sens. Environ. 216,
586–597.
Wang, M., Hu, C., 2016. Mapping and quantifying Sargassum distribution and coverage
in the central West Atlantic using MODIS observations. Remote Sens. Environ. 183,
356–367. https://doi.org/10.1016/j.rse.2016.04.019.
Wang, M., Hu, C., 2020. Automatic extraction of Sargassum features on Sentinel-2 MSI
images. IEEE Trans. Geosci. Remote Sens. https://doi.org/10.1109/
TGRS.2020.3002929.
Wang, M., Hu, C., Cannizzaro, J., English, D., Han, X., Naar, D., et al., 2018. Remote
sensing of Sargassum biomass, nutrients, and pigments. Geophys. Res. Lett. 45
https://doi.org/10.1029/2018GL078858.
Wang, M., Shi, W., 2006. Cloud masking for ocean color data processing in the coastal
regions. IEEE Trans. Geosci. Remote Sens. 44 (11), 3196–3205.
Wang, M., Shi, W., 2007. The NIR-SWIR combined atmospheric correction approach for
MODIS Ocean color data processing. Opt. Express 15, 15722–15733.