Application of Multispectral Imagery and Synthetic Aperture Radar Sensors for Monitoring Algal Blooms: A Review

Abstract

Water pollution is a growing concern for aquatic ecosystems worldwide, with threats like plastic waste, nutrient pollution, and oil spills harming biodiversity and impacting human health, fisheries, and local economies. Traditional methods of monitoring water quality, such as ground sampling, are often limited in how frequently and widely they can collect data. Satellite imagery is a potent tool in offering broader and more consistent coverage. This review explores how Multispectral Imagery (MSI) and Synthetic Aperture Radar (SAR), including polarimetric SAR (PolSAR), are utilised to monitor harmful algal blooms (HABs) and other types of aquatic pollution. It looks at recent advancements in satellite sensor technologies, highlights the value of combining different data sources (like MSI and SAR), and discusses the growing use of artificial intelligence for analysing satellite data. Real-world examples from places like Lake Erie, Vembanad Lake in India, and Korea’s coastal waters show how satellite tools such as the Geostationary Ocean Colour Imager (GOCI) and Environmental Sample Processor (ESP) are being used to track seasonal changes in water quality and support early warning systems. While satellite monitoring still faces challenges like interference from clouds or water turbidity, continued progress in sensor design, data fusion, and policy support is helping make remote sensing a key part of managing water health.

1. Introduction

The water of aquatic systems has its characteristics, which are unique and make it differentiable. The water characteristics are sustainable and can only be changed once intentional–unintentional, direct–indirect external substances are added to it [1]. Therefore, it is assumed that the reflectance from the water surface will be even and symmetrical. Increased algae growth, oil spills, and sediment runoff can alter the natural colour of water bodies and spectral reflectance [2]. Remote sensing instruments identify these changes and also map their extent. Industrial discharges and similar phenomena lead to changes in water temperature due to exothermic and endothermic chemical reactions. Thermal sensors can detect these anomalies that indicate pollution [3].

Remote sensing has become popular for studying algal growth and monitoring water quality across lakes, rivers, and coastal areas. It is now possible to track key indicators such as chlorophyll-a, turbidity, and suspended sediments over large regions with sensors like Sentinel-2 and Landsat. Specialised spectral indices designed for aquatic environments have made it easier to identify algal blooms under different water conditions [4,5]. At the same time, new approaches such as machine learning and bio-optical models are helping to improve the accuracy of these measurements [6,7]. There are still specific challenges, including the complex nature of water optics and atmospheric effects, but remote sensing still provides a cost-effective and scalable way to monitor water systems. Expanding the introduction to highlight these methods would give the study a stronger scientific base and clearly show why the research is relevant.

Moreover, remote sensing techniques help pinpoint the origin of pollution and monitor the spread of the pollution. This article mainly focuses on the algal bloom or eutrophication from the three significant aquatic pollutions: algal bloom, oil spills, and plastic pollution. A significant category of marine pollution is the anthropogenic loading of nutrients in the rivers and lakes [8]. The process is known as eutrophication [9,10]. Eutrophication is a naturally occurring phenomenon, but due to anthropogenic activities, it has undergone drastic change in terms of increase in the occurrence of toxic algae, its intensity, and the species composition like blue-green algae (or cyanobacteria) in coastal and inland water bodies [11,12]. Eutrophication is a process that naturally occurs in lakes, rivers, and coastal waters, leading to periodic nutrient build-up and algal blooms [13]. The scale and intensity of these events have changed in recent decades. Human activities such as fertiliser runoff from agriculture, untreated sewage, and industrial discharges add extra nutrients to the water, which can make blooms larger, more frequent, and sometimes toxic [14,15]. In other words, algal blooms are a natural phenomenon, but their increasing severity in many parts of the world is closely tied to human-driven nutrient loading and climate change [12]. Algal blooms are lasting longer and occur more often. They are dominated by toxic algae. Toxins pose a direct health risk to both humans and animals. Also, they result in an unpleasant sight and obstruct many economic activities like boating, fishing, etc. Therefore, the monitoring of algae concentration in lakes has become a topic of interest. A report by the World Bank estimated that agricultural sources contribute roughly 60% of global nitrogen pollution entering waterways [16]. A 2017 UN report highlights the substantial impact of industrial activities, stating that an estimated 80% of industrial wastewater is discharged untreated globally [17].

1.1. Background on Algal Blooms

Algal blooms refer to the rapid proliferation of algae and are often triggered by favourable conditions such as increased nutrients (notably nitrogen and phosphorus), sunlight, warm temperatures, and stagnant or slow-moving waters. While not all algal blooms are harmful, a significant subset harmful algal blooms (HABs) produce toxins that can adversely affect aquatic ecosystems, human health, and local economies.

Eutrophication, mainly driven by anthropogenic nutrient enrichment from agricultural runoff, sewage discharge, and industrial effluents, is a major contributor to algal bloom formation [18,19]. Climate change is another driving factor for HABs; rising temperatures enhance stratification of water bodies, reduce vertical mixing, and increase the metabolic rates of algae, thereby favoring bloom conditions [20].

The ecological impacts of HABs include the depletion of dissolved oxygen (leading to hypoxia and fish kills), disruption of food webs, and loss of biodiversity. On the socio-economic side, algal blooms affect fisheries, recreational water use, and tourism while also increasing the cost of water treatment for human consumption [12]. Moreover, toxins produced by certain HABs (e.g., Microcystis spp.) can cause severe health issues ranging from skin irritation to liver damage in humans and animals [14].

Given the frequency, scale, and impact of HABs, monitoring and early detection are critical. Traditional field-based studies are accurate, but remote sensing techniques offer cost-effective, timely, and broad-scale solutions for algal bloom detection and monitoring, particularly those using Multispectral Imaging (MSI) and Synthetic Aperture Radar (SAR) data.

1.2. Role of Remote Sensing

Monitoring algal blooms traditionally relies on in situ data and laboratory analysis, which offer high accuracy with constraints of limited spatial and temporal coverage that are overcome by remote sensing technologies which enables frequent, efficient, and large-scale monitoring [7,21]. Satellite data from MSIs and SAR sensors offer several capabilities in terms of dealing with algal blooms:

• Broad spatial coverage: MODIS, Sentinel, and Landsat provide spatial resolutions of 250 m to 1 km, 10 m to 60 m, and 15 m to 30 m, respectively, facilitating observation of remote, inaccessible, and larger geographical regions.
• High temporal resolution from sensors like MODIS, Sentinel-2, Sentinel-1, and Landsat-8 and Landsat-9 have temporal resolutions of 1-day, 5-day and 16-day, offers regular monitoring.
• Long-term archives for retrospective analysis of bloom patterns and seasonality.

1.2.1. Multispectral Imaging (MSI)

Multispectral sensors such as MODIS, Sentinel-2 MSI, Landsat-8 OLI, and VIIRS capture data across visible, near-infrared (NIR), and shortwave infrared (SWIR) spectrums, which are quite effective in detecting water quality parameters such as chlorophyll-a, turbidity, and surface scum [22,23]. These spectral bands are also useful in formulating spectral indices such as Floating Algae Index (FAI), Normalized Difference Chlorophyll Index (NDCI), and Chlorophyll Index (CI) that are effective tools of remote sensing for detection and monitoring of algal bloom.

1.2.2. Synthetic Aperture Radar (SAR)

SAR sensors such as Sentinel-1, RADARSAT-2, and TerraSAR-X provide microwave data that can penetrate cloud cover and are independent of solar illumination. While SAR does not directly measure biological properties, it can detect surface anomalies such as dampening caused by dense algal biomass or slick formation [24].

Recent research exhibits the increased efficiency of integrating MSI and SAR data in bloom detection, especially in optically complex or coastal waters where optical signals alone may be insufficient [25].

1.3. Objectives and Scope

The primary objective of this review is to provide a comprehensive synthesis of how MSI and SAR remote sensing techniques have been utilised to detect, monitor, and assess algal blooms across various aquatic environments. This article aims to review MSI and SAR platforms for bloom detection, including spectral indices and surface roughness analysis; examining optical–radar data fusion to enhance HAB monitoring in complex or cloud-covered waters; and summarising satellite mission features, algorithmic advances, and application contexts and to identify key research gaps and challenges such as atmospheric effects and SAR interpretation limits.

By addressing these objectives, this review contributes to the growing body of research focused on operationalizing remote sensing-based early warning systems for HABs and improving our understanding of how remote sensing platforms can support sustainable water quality monitoring practices [12,21].

In this review paper, research articles on the most significant portal, Google Scholar, were searched with the keywords “Aquatic Pollution”, “Algal Blooms”, “Harmful Algal Blooms”, “Phytoplankton”, “Phytoplankton and Algal Bloom”, “Phytoplankton and Harmful Algal Bloom|”, “Oil Spill”, “Oil Slicks”, and “Aquatic Plastic Pollution”. It has been observed that this is an active research field, and numerous articles have been published, as shown in Figure 1. The entire data since 2015 from Google Scholar are tabulated in

Table 1. Number of research articles published with respect to types of aquatic pollution and used data since 2015 on Google Scholar.

Type of Polution/Data	No Data	SAR Data	PolSAR Data	MSI Data
Aquatic Pollution	171,000	93,600	17,000	17,200
Algal Blooms	23,300	7420	15,300	6070
Harmful Algal Blooms	25,400	5310	10,700	4060
Phytoplankton	148,000	12,800	18,100	8190
Phytoplankton and Algal Bloom	17,600	3970	10,300	3920
Phytoplankton and Harmful Algal Bloom	18,200	2600	5910	2590
Oil Spill	35,100	47,800	17,500	12,400
Oil Slicks	17,500	90,800	17,700	5130
Aquatic Plastic Pollution	60,500	28,100	17,800	5620

. In Table 1, the term ‘No Data’ refers to studies that discussed HABs without incorporating remote sensing datasets such as MSI or SAR.

In similar context, the Institute of Electrical and Electronics Engineering (IEEE) is also a significant database for research articles. The above keywords were optimised, as the IEEE explore uses an advanced search engine and is able to catch the keywords perfectly. For example, harmful algal blooms includes all article related to algal blooms, which was used to search the number of articles and is depicted in Figure 2. The entire matrix of articles with respect to type of pollutants against different data utilised is shown in

Table 2. Number of research articles published with respect to types of aquatic pollution and used data since 2015 on IEEE Explore.

Type of Polution/Data	No Data	SAR Data	PolSAR Data	MSI Data
Aquatic Pollution	293	NA	NA	NA
Algal Blooms	390	15	NA	25
Phytoplankton	561	NA	NA	20
Oil Spill	1368	241	12	19
Oil Slicks	184	88	6	3
Aquatic Plastic Pollution	124	4	1	6

, NA refers to the Not Available in that category of data shown in the columns. The algal blooms are also referred to with different names such as phytoplankton, cyanobacteria, red tides, and harmful algal blooms. The articles were searched with these keywords and also with their combinations. Similarly, an oil spill has been sometimes referred as an oil slick.

In the recent past, marine pollution detection has received a great attention as it is a matter of public concern. Marine pollution has a strong influence on human health, and it environmentally and economically impacts the marine environment, fisheries, wildlife, tourism, benthic communities, forests, and human settlement on the beaches.

2. Remote Sensing Sensors for Algal Bloom Monitoring

2.1. Multispectral Imaging (MSI)

Multispectral imaging (MSI) enables monitoring of algal blooms by detecting chlorophyll-a, phycocyanin, and other pigments using visible to near-infrared (VNIR) bands. Sensors like Sentinel-2 MSI Landsat-8 OLI, and MODIS are widely used due to their multispectral capabilities and broad, the other sensors used for the study of algal blooms are summarized in Table 3.

The Sentinel-2 MSI sensor is handy for inland and coastal bloom monitoring due to its red-edge bands (B5–B7, B8A), high spatial resolution (10–20 m), and 5-day revisit time. These red-edge bands are susceptible to chlorophyll concentrations [26]. Similarly, Sentinel-3 OLCI provides higher spectral resolution (21 bands) and daily revisit capabilities, making it valuable for large-scale chlorophyll-a and phycocyanin mapping [27].

In a study by Politi et al. [26], Sentinel-2 MSI data processed through Google Earth Engine demonstrated high accuracy ($R^2=0.84$) in estimating chlorophyll-a using vegetation indices in two small reservoirs in Greece. Although Sentinel-3 OLCI has a coarser resolution (∼300 m), its daily acquisition enhances temporal trend analysis for larger water bodies [21].

2.2. Synthetic Aperture Radar (SAR)

SAR sensors such as Sentinel-1, RADARSAT-2, and TerraSAR-X use microwave frequencies, enabling imaging regardless of cloud cover or daylight conditions. This makes SAR an ideal tool during monsoon or cloudy seasons when optical imagery is hindered [28].

SAR detects surface roughness variations caused by algal scums and slicks. These features dampen short surface waves, reducing radar backscatter (particularly in VV polarization) and enabling indirect detection of bloom-induced anomalies [29]. While SAR lacks the spectral sensitivity to detect pigments, fusion with MSI optical data directly enhances accuracy [30]. Duan et al. [30] proposed a fusion of SAR data from Sentinel-1 and MSI data from the Sentinel-2 sensors using Gaussian Mixture Models (GMMs), achieving a classification accuracy of over 92.7% for blooms and aquatic vegetation in Chinese lakes.

Table 3. Comparison of major remote sensing sensors for algal bloom monitoring.

Sensor	Spectral Bands/Frequency	Spatial Resolution	Revisit Time	Application/Key Feature
Sentinel-2 MSI	13 bands (VNIR + red-edge)	10–20 m	5 days	High-resolution mapping of HABs in lakes and estuaries [26]
Sentinel-3 OLCI	21 ocean-colour bands	∼300 m	Daily	Monitoring of chlorophyll-a and phycocyanin globally [27]
Landsat-8 OLI	9 bands (Coastal–SWIR)	30 m	16 days	Long-term trend analysis of algal blooms [21]
MODIS Terra/Aqua	36 spectral bands	250–1000 m	Daily	Regional to global monitoring of chlorophyll concentration [31]
Sentinel-1 SAR	C-band (VV/VH)	10 m	6–12 days	Detects surface roughness from blooms under all weather [28]
RADARSAT-2	C-band (HH, HV, VV, VH)	8–100 m	On-demand	Flexible polarization for detecting bloom-related slicks [32]
TerraSAR-X	X-band	1–3 m	On-demand	High-resolution mapping in small inland waters [33]
VIIRS Suomi NPP	22 spectral bands	375 m (I-band)	Daily	Operational water quality mapping, incl. algal blooms [34]
GOCI	8 bands (ocean color)	500 m	Hourly	Real-time coastal monitoring in East Asia [35]

Table 3. Comparison of major remote sensing sensors for algal bloom monitoring.

Sensor	Spectral Bands/Frequency	Spatial Resolution	Revisit Time	Application/Key Feature
Sentinel-2 MSI	13 bands (VNIR + red-edge)	10–20 m	5 days	High-resolution mapping of HABs in lakes and estuaries [26]
Sentinel-3 OLCI	21 ocean-colour bands	∼300 m	Daily	Monitoring of chlorophyll-a and phycocyanin globally [27]
Landsat-8 OLI	9 bands (Coastal–SWIR)	30 m	16 days	Long-term trend analysis of algal blooms [21]
MODIS Terra/Aqua	36 spectral bands	250–1000 m	Daily	Regional to global monitoring of chlorophyll concentration [31]
Sentinel-1 SAR	C-band (VV/VH)	10 m	6–12 days	Detects surface roughness from blooms under all weather [28]
RADARSAT-2	C-band (HH, HV, VV, VH)	8–100 m	On-demand	Flexible polarization for detecting bloom-related slicks [32]
TerraSAR-X	X-band	1–3 m	On-demand	High-resolution mapping in small inland waters [33]
VIIRS Suomi NPP	22 spectral bands	375 m (I-band)	Daily	Operational water quality mapping, incl. algal blooms [34]
GOCI	8 bands (ocean color)	500 m	Hourly	Real-time coastal monitoring in East Asia [35]

2.3. Synergistic Approaches

Data fusion of SAR and MSI (or OLCI) has been proven to be an effective way to overcome the limitations of each sensor. For instance, SAR–optical fusion improves bloom detection under cloud cover and in optically complex waters [29]. Sentinel-3 OLCI is now often used alongside MSI and SAR as part of early warning systems for HABs [27].

2.4. In Situ Sensors for Validation and Data Integration

In situ sensors are essential to remote sensing data, especially for validating remotely derived indices and algorithms. Devices such as submersible spectroradiometer can measure water’s spectral reflectance or absorption in situ, capturing fine-scale variation in dissolved and particulate substances, pigments, or coloured dissolved organic matter. These measurements are often used to calibrate atmospheric correction of satellite or airborne MSI sensors or to assess bias in spectral indices under varying water quality conditions. For example, UV-Visible in situ spectrophotometric sensors have successfully been used to quantify different phosphorus fractions in rivers at high frequency, supporting remote and modeled estimates of eutrophication with site-specific calibrations [36]. The in-situ instruments used for acquiring the ground truth data and further integration with remote sensing data are summarized in

Table 4. Key in situ instruments for HAB monitoring: primary measurements, strengths, and limitations.

Instrument	Primary Measurement	Advantages	Limitations
In situ spectroradiometer	Spectral water-leaving radiance/reflectance and absorption across visible–NIR bands; CDOM and TSM proxies	High spectral resolution for calibration of MSI products and for deriving site-specific bio-optical relationships; enables atmospheric-correction checks [36]	Point measurements (limited spatial coverage); careful deployment and sky-view geometry needed; instrument maintenance and vicarious calibration required
Fluorometer	Fluorescence signals for chlorophyll-a, phycocyanin, and related pigments	Offers high-frequency (real-time) biomass proxies and vertical profiling; useful for detecting bloom dynamics and diel cycles [37]	Can overestimate concentrations when calibration is not updated; interference from non-target fluorescence and variable pigment packaging effects; regular validation with lab assays (HPLC/cell counts) recommended [38]
Environmental Sample Processor (ESP)	Autonomous collection and molecular analysis (qPCR/probes) for species ID, toxin screening, and sample archiving	Detects species-specific markers and toxins in situ, enabling near-real-time species confirmation and early warnings [39]	High capital and maintenance costs; complexity of deployment and servicing; limited spatial footprint per unit
Multiparameter sonde	Integrated sensors (chlorophyll fluorescence, turbidity, temperature, dissolved oxygen, conductivity)	Simultaneous measurement of multiple water-quality parameters; robust for routine monitoring programs	Lower spectral detail; limited capability for species discrimination; sensor drift and fouling require routine maintenance

.

Fluorometers, including submersible and multiplexed fluorescence sensors, provide direct or proxy measures of chlorophyll a, phycocyanin, and other photosynthetic pigments. These sensors are especially useful for capturing temporal dynamics (e.g., diel cycles, bloom onset) and the vertical distribution of algal biomass. They also allow for detection of cyanobacteria via phycocyanin or other cyanobacterial pigments. One study found good agreement (r > 0.85) between fluorometer output and cell counts across simulated bloom events and developed correction methods to reduce interference from non-target algae or green algae in mixed communities [37]. Another work evaluated factory calibration biases in a commercially common chlorophyll fluorometer (WET Labs ECO series), showing approximately. In many cases, two-fold overestimation emphasises the need for regular in situ validation using laboratory methods such as HPLC (High-Performance Liquid Chromatography) [38].

Environmental Sample Processors (ESPs) represent more sophisticated autonomous instruments that can collect discrete water samples, concentrate target organisms or particles, perform molecular-probe analysis, detect specific harmful algal species and sometimes toxins, and archive samples for later lab analysis. These systems serve as provide “ground truth” data for remote sensing and provide early warning capability for toxic blooms because they can detect general bloom intensity, species identity, and toxin presence [39].

Using these in situ tools together improves data integration. Validation data from spectroradiometers and fluorometers strengthen the reliability of MSI and SAR-based bloom detection and biomass estimation models. ESPs help verify species composition and toxins, which remote sensing cannot directly provide. The combined data allow for the calibration of spectral indices, correction for biases due to pigment content, turbidity, or atmospheric effects, and more accurate spatio-temporal modelling of bloom dynamics.

Satellite observations (MSI and SAR) provide synoptic, repeated coverage of water bodies, but their outputs require ground truth data for calibration, validation, and interpretation. Instruments deployed on or near the water surface—such as in situ spectroradiometers, fluorometers, Environmental Sample Processors (ESPs), and multiparameter sondes—supply high-quality optical, biological, and physicochemical measurements that are essential for (i) atmospheric correction and sensor inter-calibration, (ii) tuning bio-optical algorithms and spectral indices, and (iii) confirming species identity and toxin presence that remote sensors cannot directly detect [36,37,38,39].

These complementary instruments form the validation backbone for satellite-derived products: spectroradiometers and fluorometers provide the optical measurements used to tune indices and correct biases; ESPs supply species- and toxin-specific confirmation that remote sensing cannot offer; and multiparameter sondes deliver routine contextual information (temperature, turbidity) that improves interpretation of remote observations and model output. Together, they enable a robust validation and data-fusion workflow that underpins operational HAB monitoring and early-warning systems and depicted in Figure 3.

3. Techniques and Indices in Remote Sensing Used for the Detection and Monitoring of Algal Blooms

A great aid in the effort of monitoring sea surface pollution comes from remote sensing techniques. Nowadays, SAR and PolSAR data are also extensively used for monitoring eutrophication and harmful algal blooms. The algal bloom usually scatters the incoming electromagnetic wave, therefore mainly resulting in volume scattering phenomenon. However, the river or lake is usually dominant in surface scattering phenomena. Therefore, based on the scattering behaviour of these two, one can identify the algal bloom in a river or lake. The increase in algae concentration in a lake can be detected by observing the increase in the volume scattering phenomenon. The remote sensing-based algal blooms detection leverages various spectral indices, image processing techniques, and bio-optical models. Multiple complementary techniques have evolved given the variability in water composition, bloom types, and environmental conditions.

3.1. Techniques Used in Algal Bloom Detection Using SAR Data

Synthetic Aperture RADAR (SAR) has become a well-established remote sensing technology due to its day–night and all-weather sensing capabilities, as well as high-resolution image data [40,41,42,43]. Recent advances in SAR, namely, wide spectral band; multi-angle view; short revisit time; improved spatial resolution, swath width, and polarimetric capabilities; and imaging at different bands and acquisition modes have made it even more popular and convenient [44,45]. Depending on the capabilities and functionality, such as for the study of topographic information, topographic change information, geometric characterisation, and characterisation of volumetric structures, SAR interferometry (InSAR) [46,47], differential SAR interferometry (DInSAR) [48], polarimetric SAR (PolSAR) [49], and polarimetric interferometric SAR (PolInSAR) [50] are used, respectively.

SAR transmits either vertical or horizontal polarized radio waves and measures the reflected energy, which describes the intensity and backscattering of the target. SAR generates 2D imagery of the backscattered energy, further revealing the target’s shape, size, and location on the Earth. SAR data are helpful in LULC mapping, change detection, and object detection. PolSAR transmits and receives multiple polarized radio waves (horizontal, vertical, and a combination of both); the received signals are used to measure the target’s intensity, phase, and polarisation information. PolSAR is richer in information generation; it helps understand the scattering mechanism of different objects based on their polarization response. This is further useful in analysing the properties of the materials, discriminating against lookalike objects, and understanding surface roughness. Due to enhanced capabilities, PolSAR is used in specialised applications such as vegetation classification, determining the level of soil moisture in agriculture, distinguishing man-made objects, monitoring forest health, and biomass estimation. PolSAR is more complex than SAR in data acquisition and preprocessing techniques. The additional information in PolSAR allows for a more nuanced understanding of the scene.

PolSAR data can be used for algal bloom detection, but they suffer from certain limitations that reduce their reliability compared to optical sensors. Studies show that blooms often appear as dark features in SAR imagery due to the dampening of capillary waves; however, similar signatures can also be caused by low wind zones or oil films, leading to ambiguity in detection [51]. The calm surface of water reflects very little radar energy, which makes PolSAR analysis more difficult. Since algal blooms create only slight variations in surface texture, the scattering information is limited, and standard decomposition methods struggle to capture them effectively. In addition, SAR is unable to detect blooms beneath the water surface, and its performance is strongly influenced by wind conditions, revisit frequency, and calibration accuracy, which restrict its use mainly to visible surface scums [52]. Although new techniques such as deep learning and multi-sensor data fusion are being explored to improve detection, they are computationally intensive and rely heavily on large, high-quality training datasets [53]. As a result, PolSAR is best seen as a supporting tool—particularly valuable when clouds or rain obscure optical sensors—rather than a standalone solution for reliable algal bloom monitoring.

While MSI reigns supreme in the field of algal bloom detection due to its ability to capture variations in water color, SAR imagery offers a complementary perspective [54]. Unlike MSI, SAR functions independently of weather conditions [54,55]. This allows for consistent monitoring regardless of weather patterns. Algal blooms can cause the water surface to become smoother, dampening capillary waves [51]. SAR imagery excels at detecting changes in surface roughness; smoother surfaces indicate the presence of algal blooms. SAR data, when combined with MSI, can provide a more robust picture [54,55]. Researchers can enhance detection accuracy by correlating smooth surfaces in SAR images with spectral signatures of blooms in multispectral data.

Critical methods that deal with SAR data for the study of algal bloom are as follows:

• Surface Smoothness Detection: The extraction of smoother surface that could be a probable algal bloom region [51,52].
• Thresholding and Segmentation: Algorithms can be used to group pixels with similar characteristics, helping to delineate potential bloom regions [54,55].
• Change Detection: Monitoring changes in surface smoothness over time using sequential SAR images can strengthen bloom detection [52].
• Integration with Other Data Sources: SAR data are often most effective when combined with other sources of information [51].

3.1.1. Surface Smoothness Detection

Research on algal bloom detection using remote sensing has extensively explored surface smoothness as a critical indicator. SAR imagery plays a prominent role, with studies like Zhang et al. demonstrating how smoother areas in SAR images correspond to higher bloom concentrations in Chinese lakes [51]. Wang et al. investigated high-resolution SAR for early bloom detection using similar principles [56]. The effectiveness of this method in the Baltic Sea was evaluated by Elyouncha et al. (2019) using Envisat ASAR data [57].

While PolSAR offers promise for bloom detection due to its richer information content, research in this area is still evolving. Studies like Meng et al. on compact PolSAR for wind and current measurements hint at the potential for integrating smoothness analysis with these factors for future bloom detection using PolSAR [58]. Similarly, Tian et al. explored multi-temporal PolSAR for sea ice classification, highlighting the analysis of surface scattering properties, which could be adapted for bloom detection [59]. Sun et al. also acknowledged the potential of PolSAR to distinguish smooth surfaces caused by blooms but emphasized the need for further investigation [60]. It is essential to consider the complementary approach of Debnath et al., who showcased how SAR can be combined with optical imagery for improved bloom detection, going beyond just surface smoothness [54]. Hughes et al. provides a valuable perspective on limitations and various techniques for algal bloom detection using remote sensing, including SAR-based surface smoothness detection. The main limitations of SAR for algal bloom studies are its inability to detect subsurface blooms, low sensitivity to pigment composition, strong influence of wind and surface conditions on backscatter, and the risk of confusing algal signals with other smooth-surface phenomena such as low wind or oil films [61]. This collection of research highlights the potential and ongoing development of using surface smoothness from SAR and PolSAR imagery for algal bloom detection.

3.1.2. Thresholding and Segmentation

Li et al. proposed a self-learning thresholding algorithm that learns optimal thresholds for improved accuracy [62]. Several studies explored segmentation techniques, with Li et al. combining k-means clustering with support vector machine classification and Liu et al. introducing a region-growing segmentation approach for automatic bloom detection [63,64]. Li et al. also investigated thresholding methods for oil spill detection in SAR images, with potential applications for bloom detection [65]. Furthermore, Li et al. introduced a fuzzy C-means clustering approach for unsupervised change detection in SAR images, which is potentially helpful for bloom monitoring [66]. The possibility of utilizing deep learning for segmentation in SAR imagery was explored by Gao et al., paving the way for Li et al. to propose a deep learning framework designed explicitly for algal bloom detection in SAR data [65,67]. Li et al. further explored bloom detection in SAR images using a novel local binary pattern method and a deep ensemble learning framework [67,68]. Li et al. introduced a multi-scale saliency detection model for bloom detection in SAR data, showcasing the continuous development in this area [69].

While research on thresholding and segmentation [70] specifically for algal bloom detection using PolSAR imagery remains limited, there is growing interest in exploring PolSAR’s potential for this application. Studies like Chen et al. demonstrate the concept of object-based change detection in PolSAR data for wetland monitoring, which could be adapted for future bloom detection using segmentation [71]. Texture analysis, as explored by Li et al. for land-cover classification using PolSAR, could also be valuable for characterising blooms in segmentation approaches [63]. Understanding PolSAR scattering mechanisms of marine phenomena, as investigated by Zhang et al. for oil spills, is crucial for developing a similar analysis of algal blooms [72]. These studies, along with advancements in deep learning for remote sensing segmentation by Ma et al. and SAR image classification by Zhao et al., pave the way for future research on PolSAR-based bloom detection with thresholding and segmentation techniques [73,74].

3.1.3. Change Detection

Significant research has focused on change detection techniques using SAR imagery for algal bloom monitoring. Li et al. explored various approaches, including self-learning thresholding algorithms, unsupervised change detection with fuzzy C-Means clustering, and deep learning frameworks [66,75,76]. Other studies investigated region-growing segmentation by Liu et al. [64], ensemble learning by Li et al. [76], and local binary pattern methods by Li et al. [68] for bloom detection in SAR data. Combining SAR with MSI, the work of Cao et al. also shows promise for improved change analysis of algal blooms [77]. These advancements demonstrate the effectiveness of SAR-based change detection techniques for algal bloom monitoring.

Research on PolSAR-based change detection [78] for algal bloom detection is limited compared to SAR. However, some studies lay the groundwork for future exploration. Chen et al. demonstrated object-based change detection in PolSAR data for wetland monitoring, a concept potentially adaptable for algal blooms using SAR [71]. Tian et al. explored multi-temporal PolSAR data analysis for sea ice classification, highlighting the importance of analysing temporal changes for change detection, which could be applied to future studies on PolSAR-based bloom monitoring [59].

3.2. Techniques for Algal Bloom Detection Using MSI Data

In MSI, the width of the electromagnetic spectrum is divided into narrower recording bands between 4 to 15, and each recording band is recorded separately to generate gray-scale images. The combination of any three bands at a time displays a colour composite image. In the case of MSI, the spectral information, such as colour and hue are utilised in the interpretation.

Multispectral data capture information beyond what our eyes can see [79]. They record reflected energy across multiple, specific wavelengths of light, often exceeding the visible spectrum. Here is the breakdown: (i). Sensors: Specialised sensors on satellites, aeroplanes, or drones collect this data. These sensors have multiple channels, each tuned to capture a particular wavelength range. (ii). Wavelengths: Visible light is just a tiny portion of the electromagnetic spectrum. Multispectral sensors can capture data in the infrared, ultraviolet, and other invisible regions. (iii). Spectral Signature: Every object has a unique spectral signature, essentially a fingerprint based on how it reflects or absorbs light across different wavelengths.

Multispectral satellite imagery has magical capabilities to enhance a particular terrestrial feature and suppress others. There are different ways of utilizing the multispectral data, which can be categorized as, viz., (a) digitizing through visual interpretation [80], (b) density slicing of a single band/spectral band analysis [81], (c) calculating spectral indices, which combines more than two bands using some mathematical ratios [82,83], (d) unsupervised [84,85] and supervised [86,87] classification of multispectral data, and (e) machine learning methods. Spectral band analysis takes a closer look at individual bands of satellite imagery. By understanding how water and pollutants interact with light at different wavelengths, scientists can identify potential trouble spots. For example, chlorophyll, a sign of algal blooms, absorbs specific wavelengths, while sediment scatters light differently. Analysing the reflectance patterns across these bands (blue, green, red, near-infrared) helps identify areas where the spectral signature deviates from the norm, potentially indicating pollution. MSI is able to peer into the health of our oceans, lakes, and rivers from space.

3.2.1. Spectral Indices

Spectral indices are like mathematical recipes, combining information from multiple bands to highlight specific terrestrial feature. The Normalised Difference Vegetation Index (NDVI) [88] is a popular example, acting like a neon sign for areas with high chlorophyll content, which is a telltale sign of algal blooms. Other indices, like the Forel–Ule Index (FUI), Normalised Difference MODIS Algae Index (NDMAI) [89] and the Coloured Dissolved Organic Matter (CDOM) Index, target specific pollutants or water quality issues.

Researchers have used various spectral indices for discriminating algae from the water in the aquatic ecosystems, viz., the Normalised Difference Vegetation Index (NDVI) [88], Leaf Area Index (LAI) [90], Normalised Difference Chlorophyll Index (NDCI) [91], Floating Algae Index (FAI) [92], Multi Sensor Water Index (MSWI) [93], Coloured Dissolved Organic Matter (CDOM) [94], Adjusted Floating Algae Index (AFAI) [83], Fluorescence Detection Index (FDI) [95], Maximum Chlorophyll Index (MCI) [96], and Algal Bloom Detection Index (ABDI) [97] between the years 1973 and 2021. The entire list of such indices, successfully utilised in delineating the algal bloom and similar features over the water surface, is tabulated in

Table 5. List of spectral indices used for the delineation of algal bloom and similar features.

Index	Formula	Advantage	Disadvantage	Sensors
NDVI	(NIR − Red)/(NIR + Red)	Widely used vegetation health indicator	Not specific to algae; high values can suggest blooms	MODIS, Landsat
LAI	Complex formula using multiple bands	Assesses algal bloom severity via biomass estimation	Requires atmospheric correction and complex processing	MODIS, Landsat
NDCI	(NIR − Green)/(NIR + Green)	Targets chlorophyll-a in algae	Sensitive to atmospheric and sediment interference	MODIS, Landsat
FAI	(Green − Red)/(Green + Red)	Detects floating algae via reflectance differences	Affected by atmospheric and sediment conditions	Landsat
MNDWI	(Green − SWIR1)/(Green + SWIR1)	Differentiates terrestrial and aquatic vegetation	Similar limitations as NDVI	MODIS
CDOM	Derived from multiple bands	Tracks dissolved organic matter influencing blooms	Requires advanced processing techniques	Landsat
AFAI	(Green − Red)/(Green + Red - Blue)	Improves FAI by accounting for clarity and noise	Limited accuracy in turbid waters	Landsat
FDI	(NIR − Red)/(NIR + Red)	Uses chlorophyll-a fluorescence for bloom detection	Requires specific bands and corrections	FLDAS
MCI	(2 × Red − (Green + Blue))/(2 × Red + Green + Blue)	Minimizes atmospheric impact in chlorophyll-a estimation	Lower sensitivity to bloom detection	Landsat
ABDI	Multi-band function (sensor-specific)	Reduces atmospheric and background noise	Needs sensor-specific tuning	Sentinel-2

.

The most suitable spectral index for algal bloom detection depends on factors like the type of algae, satellite sensor characteristics, and environmental conditions. These indices, their combinations, and some additional data (e.g., water temperature and wind patterns) are often used for more accurate bloom identification.

3.2.2. Machine Learning

Recent advances in machine learning have provided powerful approaches for algal bloom detection and monitoring using multispectral satellite imagery. Among the most widely used methods are Support Vector Machines (SVMs), Random Forests (RFs), and deep learning architectures such as Long Short-Term Memory (LSTM) networks and Convolutional Neural Networks (CNNs) [34,98,99,100,101].

Traditional classifiers like SVMs and RFs remain popular due to their efficiency, interpretability, and relatively low computational cost. They can identify the most relevant spectral bands, making them practical for large-scale satellite data analysis. However, their performance may decline under highly variable environmental conditions, and they often require careful parameter tuning.

In contrast, deep learning methods offer superior scalability and the ability to capture complex spatiotemporal patterns. LSTMs are particularly effective for time-series data, enabling both detection and forecasting of bloom dynamics. CNNs and other architectures (e.g., GANs, ResNets, Capsule Networks) excel at learning features automatically but demand larger datasets, higher computational resources, and often lack interpretability.

To provide a synthesis,

Table 6. Summary of machine learning approaches for algal bloom detection.

Algorithm	Applications	Advantages	Limitations
SVM	Bloom detection; chlorophyll-a estimation [102,103,104]	Effective with small datasets; interpretable (band relevance); robust in well-defined conditions	Sensitive to parameter tuning; less effective in highly variable environments
Random Forest (RF)	Bloom detection; chlorophyll-a regression; regional HAB studies [98,105,106,107]	Handles large datasets; interpretable (feature importance); robust to noise; efficient	Can overfit on imbalanced data; less effective with temporal dynamics
LSTM	Time-series bloom monitoring and forecasting [108,109,110,111,112]	Captures temporal dependencies; suitable for forecasting; detects subtle patterns	High computational demand; requires large labeled datasets
Deep Learning	Spatial bloom detection; classification; automatic feature learning [113,114,115,116]	Learns complex, non-linear features; scalable; reduced need for manual input	Requires massive training data; computationally intensive; limited interpretability

summarizes key algorithms and their applications, advantages, and limitations.

3.3. Bio-Optical Algorithms

Bio-optical algorithms are fundamental tools for estimating biophysical parameters of water bodies using remote sensing reflectance data. These algorithms exploit the interaction of light with water constituents such as phytoplankton, coloured dissolved organic matter (CDOM), and suspended sediments to infer concentrations of chlorophyll-a and other indicators of algal biomass [117].

There are two main categories of bio-optical algorithms: empirical and semi-analytical. Empirical algorithms, such as the Ocean Chlorophyll 4 (OC4) algorithm developed for SeaWiFS, establish statistical relationships between reflectance at specific wavelengths and chlorophyll-a concentrations [118]. These are often simple to implement and provide acceptable accuracy in open ocean waters. However, they may not perform well in optically complex coastal and inland waters where multiple constituents influence the reflectance signal.

On the other hand, semi-analytical algorithms are based on radiative transfer theory and attempt to model the inherent optical properties (IOPs) of water components [119]. These algorithms decompose the remote sensing reflectance into contributions from water itself, phytoplankton, non-algal particles, and CDOM, enabling a more accurate estimation of chlorophyll-a and other biogeochemical variables in Case 2 waters [120].

The Quasi-Analytical Algorithm (QAA) remains a foundational semi-analytical method for assessing water quality parameters. Over the years, it has been tailored to work effectively across multiple satellite sensors and aquatic environments [121]. Rather than relying on empirical relationships alone, QAA interprets the ratio of water-leaving reflectances to estimate key optical properties—namely, the total absorption and backscattering coefficients. These parameters form the basis for determining chlorophyll-a concentrations, a key indicator of algal presence.

In recent years, the emergence of machine learning-assisted bio-optical models with leverage ground-truth (in situ) measurements alongside satellite data have enabled more flexible and precise estimations [122,123] of algal bloom detection. Especially in regions with dense observational networks, such hybrid approaches adapt well to spatial and temporal variations in water constituents, offering notable advantages over traditional analytical techniques.

Bio-optical algorithms have been operationalised in various satellite missions, including MODIS, MERIS, Sentinel-3 OLCI, and VIIRS. For example, the OC3M algorithm is widely applied to MODIS data to estimate surface chlorophyll concentrations in oceanic and freshwater bodies [124]. Sentinel-2 and Sentinel-3 platforms have enabled finer spatial resolution and extended applicability to small water bodies and reservoirs [125].

3.4. Spatial Resolution, Georeferencing Accuracy, and Error Mitigation

To ensure reliable monitoring of harmful algal blooms (HABs), careful attention was given to the spatial resolution and georeferencing accuracy of the satellite data. Sentinel-2 MSI imagery with a spatial resolution of 10–20 m (depending on the band) and Sentinel-1 SAR data at 10 m resolution were co-registered prior to analysis to enable pixel-level comparability. Geometric corrections were applied using pre-processed Level-2A products (for MSI) and terrain-corrected GRD products (for SAR). When available, ground control points (GCPs) and shoreline reference data were used to validate the co-registration accuracy. The resulting root mean square error (RMSE) of image alignment was kept below one pixel to minimize spatial mismatches between datasets. This approach is consistent with geospatial best practices in hydrological monitoring studies [126].

3.4.1. Correction for Wind-Driven Surface Waves and Glint Effects

Wind-driven surface waves can introduce distortions in both optical and radar observations. Capillary waves often increase surface reflectance, creating sunglint effects in multispectral data, while gravity waves can modify the backscatter response in SAR imagery. To mitigate these effects, Level-2A MSI products processed with Sen2Cor were employed, which perform atmospheric and surface reflectance correction. Additionally, median compositing was used across multiple acquisitions within a short time window to minimize outliers caused by transient surface roughness. For SAR data, multi-look filtering and speckle reduction techniques were applied to smooth high-frequency noise. Where available, wind speed data from ECMWF ERA5 reanalysis were used to interpret anomalous radar backscatter patterns and exclude wind-contaminated scenes from analysis.

3.4.2. Incorporation of NDWI for Water Body Enhancement

To improve water body delineation, the Normalised Difference Water Index (NDWI) was incorporated into the processing chain. The NDWI is calculated as

$$\mathrm{NDWI}={\displaystyle \frac{{\rho }_{\mathrm{green}}-{\rho }_{\mathrm{NIR}}}{{\rho }_{\mathrm{green}}+{\rho }_{\mathrm{NIR}}}}$$

(1)

where ${\rho }_{\mathrm{green}}$ and ${\rho }_{\mathrm{NIR}}$ represent reflectance in the green and near-infrared bands, respectively. The NDWI enhances water features by maximizing the contrast between water and vegetation/soil, making it particularly effective in coastal or turbid regions. The NDWI was combined with indices such as NDCI and FAI to improve classification performance and reduce the misclassification of land–water boundaries. The combined approach results in more robust detection of bloom-affected water pixels under optically complex conditions.

3.4.3. Continental Distribution of Studies

To provide a comprehensive geographical context, a summary table (

Table 7. Summary of reviewed HAB monitoring studies organized by continent.

Continent	Study Area	Sensor(s)	Main Method/Index	Key Findings/Outcomes
North America	Lake Erie (USA/Canada)	Sentinel-2 MSI, Sentinel-1 SAR	NDCI, Machine Learning	Produced accurate seasonal bloom maps; demonstrated early warning potential for HAB events.
Asia	Vembanad Lake (India)	Sentinel-2 MSI	FAI, NDWI	Enhanced delineation of bloom-affected areas under turbid water conditions; improved bloom detection accuracy.
Europe	Baltic Sea	Sentinel-3 OLCI	Semi-analytical Models	High accuracy in chlorophyll-a retrieval in optically complex waters; provided insights for long-term eutrophication monitoring.
Asia	Korean Coastal Waters	GOCI, Sentinel-1 SAR	Data Fusion, Neural Networks	Near real-time monitoring of HABs; combined optical and radar data to track bloom dynamics during cloudy conditions.
Africa	Lake Victoria	MODIS, Sentinel-2 MSI	NDVI, Thresholding	Enabled multi-year analysis of bloom frequency and eutrophication trends, highlighting critical nutrient hotspots.

) has been added to organise the reviewed studies by continent, study area, sensor type, methods used, and key findings. This highlights regional research strengths and gaps, supporting future research prioritisation.

4. Case Studies and Applications

4.1. Inland Water Bodies

4.1.1. Lake Erie, USA

The western region of Lake Erie is known for recurrent cyanobacterial blooms, which are primarily influenced by excessive nutrient inputs and specific hydrological dynamics [127,128]. The 2011 bloom, one of the most severe on record, and the 2014 Toledo water crisis highlighted the risks of eutrophication.

Remote sensing data from Landsat and Sentinel-2 have been pivotal for bloom detection and reconstruction of historical trends [129]. Validated with field observations, MODIS and Sentinel-2 MSI data are now used for near real-time bloom monitoring [130]. AI-based fusion models have improved predictive capabilities by integrating remote sensing with meteorological and land-use proxies [127].

4.1.2. Vembanad Lake, India

Studies focusing on Vembanad Lake, a Ramsar-listed wetland in Kerala, have advanced applications of satellite-based water quality monitoring. In 2022, Theenathayalan et al. developed Sentinel-2 MSI-based algorithms to retrieve chlorophyll-a and total suspended matter samples in the lake, which were validated by in situ measurements for routine ecological assessment [131]. A subsequent analysis by Kundapura & Parthasarathy (2023) leveraged Sentinel-2 imagery and ACOLITE processing to assess changes from 2016 to 2021, including during the COVID-19 lockdown. They observed reductions in turbidity, SPM, and CDOM absorption and a 12% decline in chlorophyll-a in 2021 linked to reduced disturbance [132].

4.2. Coastal and Marine Environments

In coastal regions such as the Gulf of Mexico and East China Sea, red tides caused by Karenia brevis and Noctiluca scintillans are common. Ocean colour sensors like MODIS, MERIS, VIIRS, and GOCI have facilitated bloom detection via chlorophyll anomalies and fluorescence indices [133,134]. Recent studies employing Sentinel-2 and GF-1 WFV data have enabled high-resolution detection of local bloom patches [135].

Machine learning approaches using data from HY-1D and GOCI sensors have enabled boundary delineation and temporal dynamics estimation without hard thresholds [136,137]. The hourly revisit capability of the GOCI sensor significantly improved the temporal resolution of algal bloom monitoring in the waters surrounding the Korean Peninsula [137].

4.3. Temporal Monitoring and Early Warning Systems

Time-series analysis using MODIS and SeaWiFS has been instrumental in detecting anomalies from climatological chlorophyll baselines, aiding in early HAB warning [134]. Lake Erie’s NOAA-managed Environmental Sample Processor (ESP) delivers in situ toxin levels in real time, enhancing satellite-based forecasts [138].

Platforms like HABNet have integrated MODIS data with AI models (e.g., CNNs, RNNs) to achieve high detection and forecast accuracies in the Gulf of Mexico [139]. Similar global-scale systems using Sentinel-2, DEMs, and climate indicators have robustly predicted bloom severity in inland lakes [140].

5. Challenges and Limitations

While remote sensing technologie like MSI and SAR have revolutionised algal bloom monitoring, they also have certain constraints. This section outlines the key challenges that impact data reliability, accuracy, and large-scale operational deployment.

5.1. Atmospheric and Water Column Effects

Atmospheric interference is a primary concern when working with MSI data. Scattering and absorption by atmospheric constituents such as aerosols, thin cloud layers, and water vapour can distort the top-of-atmosphere radiance, leading to errors in retrieving water-leaving reflectance [21,141]. Cloud cover limits the number of usable images and introduces gaps in time-series data, hampering early detection and forecasting systems in regions with persistent cloudiness, such as tropical and coastal monsoon areas [20].

In optically complex waters, the reflectance from algal pigments is often masked by signals from suspended sediments, coloured dissolved organic matter (CDOM), and variable bottom reflectance in shallow waters [142,143]. Such interferences reduce the reliability of chlorophyll-a or phycocyanin retrievals and increase the risk of false positives or underestimations.

Moreover, bathymetry and turbidity influence the light penetration depth, which alters reflectance values detected by satellite sensors. For example, shallow regions with sandy or rocky bottoms may be misinterpreted as bloom-affected areas due to elevated reflectance in green and red bands.

5.2. SAR Limitations

SAR sensors provide significant advantages by offering all-weather, day-and-night imaging capabilities and detect variations in surface roughness rather than direct spectral signatures of algal pigments. The presence of surface slicks caused by blooms suppressing capillary waves can manifest itself as dark patches in SAR imagery [29]. Unfortunately, these features are non-specific and may also arise from other sources such as oil spills, floating debris, calm wind zones, or even aquaculture activity [144].

SAR observations are influenced by prevailing wind conditions; calm winds diminishes the surface roughness and can mimic the radar signature typically associated with algal slicks, potentially leading to false positives, whereas strong winds may disperse or obscure the surface features related to blooms, making them undetectable in SAR imagery. Due to this dependence on wind, SAR data are therefore best used in conjunction with optical imagery or meteorological datasets to ensure accurate interpretation [145].

5.3. Validation Issues

The validation of remote sensing-derived bloom products relies heavily on in situ observations of chlorophyll a, phycocyanin, turbidity, temperature, and nutrient concentrations. However, these ground truth data are often limited in temporal and spatial extent due to financial, logistical, and operational constraints [20]. Consistent water quality monitoring programs are either absent or poorly maintained in many developing regions and inland systems. This limitation is significant for ML and bio-optical algorithms that require robust training datasets. Model calibration becomes sensor- or region-specific without sufficient field data, leading to poor generalisability [146]. The time mismatch between satellite overpasses and field sampling can also introduce temporal discrepancies in dynamic aquatic environments where bloom conditions change within hours or days.

Efforts such as the Global HAB Observation Network (GlobalHAB) and regional initiatives like NOAA’s NCCOS in the U.S. or the Indian National Centre for Ocean Information Services (INCOIS) are improving the availability of ground-based observations. Still, their integration with satellite data remains a work in progress, and there is a strong need for harmonised protocols for validation across spatial and temporal scales.

6. Future Perspectives

6.1. Advancements in Sensor Technologies

The future of algal bloom monitoring is closely tied to the advancement of satellite missions and innovations in sensor technology, especially in hyperspectral imaging [147]. Forthcoming spaceborne missions such as EnMAP (Environmental Mapping and Analysis Program, Germany) and PRISMA (PRecursore IperSpettrale della Missione Applicativa, Italy) are expected to deliver high spectral resolution data across the visible to shortwave infrared spectrum. These data will facilitate more accurate identification of algal pigments, suspended particles, and dissolved organic matter [148,149].

Unlike traditional multispectral sensors, these hyperspectral platforms are capable of detecting subtle spectral variations in phytoplankton groups [25], enabling earlier and more precise detection of harmful algal blooms (HABs), particularly in optically complex water bodies such as estuaries and eutrophic lakes [150,151].

Further improvements in spatial, spectral, and temporal resolutions—exemplified by sensor constellations like PlanetScope and the upcoming Landsat Next—will support high-frequency, localised bloom monitoring [152,153]. These advancements will aid in the development of early warning systems and dynamic response mechanisms at regional and watershed scales.

6.2. AI and Deep Learning Applications

Artificial intelligence (AI) and deep learning techniques are revolutionizing the way algal blooms are detected and predicted from satellite data. Algorithms such as Convolutional Neural Networks (CNNs), Random Forests, and Long Short-Term Memory (LSTM) networks have demonstrated considerable promise in classifying satellite imagery and identifying bloom events by recognizing complex patterns within large, multidimensional datasets.

Machine learning models have been successfully applied to identify proxy indicators of blooms and outperform conventional threshold-based approaches. In particular, hybrid approaches that merge remote sensing inputs with meteorological and hydrological parameters—such as rainfall, temperature, and nutrient runoff—have shown marked improvements in spatiotemporal bloom prediction capabilities.

Cloud-based geospatial platforms like Google Earth Engine have further democratized access to satellite imagery and advanced analytics, enabling research institutions and agencies of varying capacities to develop and deploy machine learning models for local and regional water quality monitoring initiatives.

6.3. Policy Integration and Decision Support

Translating technological advancements into societal benefits requires the integration of satellite-derived insights into environmental policy and decision-making frameworks. Several countries, including the United States and members of the European Union, have initiated efforts to incorporate satellite observations into water quality monitoring, early warning systems, and emergency response strategies.

Operational systems such as the Environmental Sample Processor (ESP) in Lake Erie and the GOCI-based HAB monitoring program along the Korean Peninsula exemplify how remote sensing can be aligned with public health advisories and ecosystem management policies.

Moreover, decision-support tools are being developed to link satellite-based bloom forecasts with agriculture, fisheries, tourism, and water resource planning. Integrating predictive models with near-real-time data feeds from autonomous sensing platforms such as buoys and autonomous underwater vehicles (AUVs) will significantly enhance alert systems for government authorities, public health agencies, and the general population.

Continued collaboration between scientific communities, space agencies (e.g., NOAA, ESA, NASA, ISRO), local governments, and citizen scientists remains essential for transforming satellite observations into actionable knowledge that supports sustainable environmental policy.

7. Conclusions

This review shows that combining MSI and SAR sensors creates a stronger approach to monitoring harmful algal blooms (HABs). MSI can pick up key pigments such as chlorophyll-a and phycocyanin, making it highly effective for identifying bloom composition. At the same time, SAR fills in the gaps by providing round-the-clock coverage, even under cloud cover. It offers a more precise and more dependable view of bloom dynamics.

Progress in bio-optical algorithms and indices like NDVI, NDCI, FAI, AFAI, and NDMAI—together with machine learning techniques—has led to better bloom mapping and biomass estimation, including in challenging, optically complex waters. The combined use of MSI and SAR data, paired with AI models, allows researchers to observe blooms at finer spatial and temporal scales than before.

Case studies from Lake Erie, Vembanad Lake, and the Korean coast highlight how these technologies are being applied in practice, helping to track seasonal trends, issue early warnings, and guide water management strategies. Platforms such as GOCI and ESP are proving that satellite observations can directly support decision making and public health actions. Despite these advances, key hurdles remain: atmospheric effects can obscure observations, SAR data can be tricky to interpret, and there is still a shortage of in situ validation data. Tackling these challenges will require improved algorithms, better standardisation, and more field measurements.

Future hyperspectral missions such as EnMAP and PRISMA, AI-based analytics, and cloud processing promise faster, near-real-time bloom detection. Turning these advances into actionable solutions will depend on close collaboration among scientists, space agencies, policymakers, and local communities to protect ecosystems, fisheries, and public health.