IEEE Geoscience and Remote Sensing Magazine - June 2023 - 19

a pixel's area, the possibility of detecting smaller debris
decreases [116].
Data preprocessing in the detection of plastic in the
ocean is commonly minimal (without noise reduction and
normalization [117]) and very often omits atmospheric
correction to eliminate the risk of removing information in
wavelengths dominated by water absorption [115]. Atmospheric
correction may have a minimal impact on narrow
bands of HSIs, which may adversely affect detection and
classification based on pollutants' spectral profiles [113],
[118]. Therefore, level 1 MSIs and HSIs are often exploited
[116], [119], [120], and HSIs may be utilized to distinguish
different categories of garbage (e.g., fishing nets and plastics)
[113] and seawater [115]. Since supervised methods
require ground truth, which is most often obtained with
in situ methods in extremely time-consuming processes,
unsupervised and semisupervised techniques have been
increasing in the field [116]. Finally, it is worth mentioning
that tracking microplastics at the ocean surface layer
requires very detailed radiative transfer analysis and the
development of high signal-to-noise sensors, and it constitutes
another exciting avenue of emerging research [121].
DETECTION OF WATER POLLUTION
Water pollution poses a serious threat to ecosystems; therefore,
monitoring it is an important element of various countermeasures.
Due to the nature of pollution, several types
of events can be distinguished here. Oil spills are one of
the main sources of marine pollution [122], resulting in
not only oil stains on the water surface but also the death
of animals and vegetation and damaged beaches, which
translates into losses in the economy and tourism [122]. Locating
oil spots and monitoring their displacement allows
us to track their environmental impact, take preventive actions
[123], and ensure justice (and compensation) when a
spill source can be identified [124].
Remote sensing can help monitor the displacement of
oil spills while ensuring high scalability over large areas.
SAR imagery can be exploited here, as oil spots are manifested
as contrasting elements to surrounding clean water
[125], [126]. As a result, SAR images show black spots representing
oil spills and brighter areas representing clear water
[127]. Such imaging can effectively work regardless of
weather conditions, such as the degree of cloudiness and
changes in lighting [123]. The research area of detecting oil
spills from SAR is very active and spans classic [127], [128]
and deep machine learning [122], [123], [125], [126], [129].
The less common processing of MSI data from Landsat-8
[130] can also be precise in locating oil spills.
Garbage that flows down rivers to the ocean is a remnant
of human activity [113]. Another threat to water quality and
ecosystems is the human impact on water management
through eutrophication [92] and industrial [131] and mining
[65] activities. Litter contaminants are characterized by
significant variation in their composition (plastic, wood,
metal, and biomass) [112], which is a challenge to detecting
JUNE 2023 IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE
and tracking them [113]. HSI analysis enables capturing the
characteristics of such materials; for example, initial simulations
confirmed that the absorption characteristics of 1,215
and 1,732 nm have applications in detecting plastic [113].
Also, the use of the SWIR range allows us to eliminate the
bathymetric influence [115], which should be considered
due to water vapor absorption. Spatial resolution is of great
importance in the case of contamination detection because
with the increase of the area per pixel, the possibility of
detecting smaller plastic debris decreases [116]. Industrial
impacts reduce water quality,
especially inland, through
heavy metals [88]. Keeping
track of these changes and determining
their nature is thus
essential for human health.
The assessment of the state
of rivers should form the basis
for planning agricultural
and industrial activities [83].
The assessment of the mineral
composition of water may be
based on HSI and MSI analysis
through observing spectral
signatures [65], [83].
DETECTION OF HABs AND WATER
QUALITY MONITORING
HABs are a serious threat both to humans and other living
organisms present in the aquatic environment [132]. HABs
cause the extinction of some organisms, due to limited light
and the inhibition of photosynthesis [133], and they reduce
fishing [132], deteriorate water quality [133], and may be a
threat in the case of power plants located near reservoirs, as
they may cause blockages of cooling systems, as was the case
with the Fangchenggang Nuclear Power Plant, in China [134].
Monitoring such harmful blooms can, therefore, result in socioeconomic
benefits [133]. The development of algal blooms
is influenced by environmental conditions, such as temperature
[133] and water fertility [135], and man-made infrastructure,
e.g., river dams limiting the movement of water [136].
Algae occurrence estimation can be performed using in
situ methods [134]. This approach allows us to study conditions
by using buoys equipped with a sensor collecting data
from aquatic ecosystems (temperature, salinity, dissolved
oxygen, chlorophyll a, and the concentration of algae [134]).
In situ measurements can also validate estimation based on
HSIs [137] and MSIs. Due to the influence of seasonality on
the value of the water surface temperature [138], which affects
the presence of HABs, the use of UAVs may limit the
monitoring of changes over time. Satellite imagery, apart
from providing massive scalability (in situ techniques are
extremely costly and labor-intensive for periodic measurements),
enables generating algal maps taking into account
information on spatial distribution and variability over time
[132], [129]. Automated machine learning detection methods
19
A QUANTITATIVE ANALYSIS
OF THE EXISTING STATE OF
THE ART IS PIVOTAL TO
MAKE AN INFORMED
DECISION ABOUT THE
RESEARCH AREA THAT WILL
BE TACKLED IN AN
UPCOMING MISSION.
IEEE Geoscience and Remote Sensing Magazine - June 2023

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2023
