Remote Sensing-Based Water Quality Parameters Retrieval Methods: A Review
Abstract
Water quality is a sensitive global environmental issue, as it is important for long-term economic development and environmental sustainability. It is a general descriptor of water properties in terms of physical, chemical, thermal, and/or biological characteristics. Laboratory analysis is used to measure and analyze water quality parameter, however, it is a conventional, time-consuming, and expensive approach often providing discrete data at a single point in space and time and making it difficult to characterize a larger waterbody, while remote sensing methods is a cost effective and accurate methods of water quality monitoring with a high spatial and temporal resolution for large area of waterbodies. To this end, this review focused on novel findings in the field of water quality estimation using remote sensing techniques, and the result revealed that remote sensing method has used to retrieve water quality parameters which are optically active (Chlorophyll-a, Secchi Disk depth, Water temperature, Turbidity, Total Suspended Matter, Electrical conductivity, Sea Surface Salinity and Colored Dissolved Organic Matter), and optically non active (Dissolved Oxygen, Chemical Oxygen Demand, Biochemical Oxygen Demand, Total Nitrogen, Ammonia Nitrogen and Total Phosphorus). Various properties (spectral, spatial and temporal, etc.) of the more commonly employed multi spectral and hyper spectral sensors of both satellite and non-satellite-born data sources are tabulated to be used as a sensor selection guide. Furthermore, this paper summarizes the commonly used different retrieval algorisms (analytical, empirical, and artificial intelligence/machine learning (AI/ML)) employed in evaluating and quantifying the water quality parameters. As a whole, remote sensing technology is a permissible method for water quality monitoring across the world in its spatio-temporal coverage, accuracy, and its cost effectiveness
Downloads
References
Monitoring water quality (WQ) in aquatic environments is critical for the proper management of water resources to guarantee a sustainable use (Pizani1 and Maillard, 2022). It is also a means to get an insight on the dynamics of the surrounding human activities (Odermatt et al., 2008). The quality of these environments can be determined through their physical, chemical, and biological characteristics which will be addressed as WQ “parameters” (Gholizadeh et al.,2016). Water quality indicators including physical, chemical, and biological properties are traditionally determined by collecting samples from the field and then analysing the them in the laboratory. Although this in-situ measurement offers high accuracy, it is a labour intensive and time-consuming process that represent point estimations of the quality of water conditions in time and space, and obtaining spatial and temporal variations of quality indices in large water bodies is almost impossible (Pizani1 and Maillard, 2022). These limitations can in turn impact the overall data accuracy due to the inability of field measurements to capture heterogeneity of water quality measures throughout the extent of the study area as a whole (Liu et al., 2003). Moreover, conventional point sampling methods are not easily able to identify the spatial or temporal variations in water quality which is vital for comprehensive assessment and management of water bodies.
With advances in space science and the increasing use of computer applications and increased computing powers over recent decades, remote sensing techniques have become useful tools to achieve water quality monitoring goal. Remote sensing techniques make it possible to monitor and identify large scale regions and water bodies that suffer from qualitative problems more effectively and efficiently. Therefore, remotely sensed data can reinforce the abilities of water resources researchers and decision makers to monitor water bodies more effectively. These techniques involve the use of satellite imagery, aerial photography, and other technology to collect data about water bodies from a distance. The technique has been in used since the 1970’s and continue to be widely used in water quality assessment in the contemporary world (Giardino et al., 2014) to retrieve water quality parameters. The use of remotely sensed data has allowed for studies previously not feasible or practical due to inaccessibility of study areas. Various parameters such as chlorophyll-a concentration, water turbidity, and total suspended solids can be retrieved using satellite imagery. This is done by analysing the reflectance properties of the water surface, as different substances exhibit unique spectral signatures. Given the importance of remote sensing techniques for water quality estimation, reviewing remote sensing techniques is very critical for sustainable management of water resources. This review summarizes different data and information related to the application of remote sensing for water quality retrieval, and mainly discusses the research progress in terms of data sources and retrieval algorithms for specific water quality variables.
METHODOLOGY
Publications on remote sensing-based water quality evaluation were searched in English language on different sources like Web of Science and Scopus using the terms “water quality” and “remote sensing” as topical subject, and then papers describing the monitoring and assessment of water quality for utilizing remote sensing methods have been collected, and then a detailed check was done on the selected literature by scanning the full text and excluded irrelevant literature, whose main topic is not water quality-related remote sensing (e.g., land-use and land-cover change). I also differentiated the water quality parameters and the methods of establishing the inversion algorithms, and remote-sensing data resource. The type of sensors, retrieval algorisms and different water quality parameters were then tabulated using the obtained information.
FINDINGS AND DISCUSSION
Fundamentals of Remote Sensing for Water Quality Monitoring
Water quality remote sensing relies on sunlight reflected by water and its constituents, and therefore we need to know how light interacts with matter. The remotely sensed signal is the portion of the incoming solar radiation that is scattered back and reflected into a light sensing instrument (Figure 1). There are a multitude of pathways by which light can reach the sensor. First, sunlight is both absorbed and scattered by the atmosphere. Some of this light reaches the line of sight of the sensor and therefore does not interact with the water. The light that reaches the water surface is a mixture of direct sunlight and diffuse light, i.e., light scattered by the atmosphere (Olmanson et al., 2015). At the water surface, some light reflects back and may reach the sensor as sun glint. The portion of light that finally reaches below the air-water interface is then absorbed and scattered by water molecules and substances within the water (Kirk, 1994). Again, only the portion of this light that is scattered back into the line of sight of the sensor and makes it back through the atmosphere can be measured by a satellite sensor. In order to estimate the water’s optical properties, the effect of atmospheric absorption and scattering has to be compensated for the atmosphere and sun glint, a process that is called atmospheric correction. Surface reflectance is the ratio of the light exiting the water surface upwards over the incoming sunlight (Schott, 2007). Once the spectrum of light reflected and scattered from the water column (i.e., spectral reflectance) is determined by remote sensing, we can apply retrieval algorithms to derive concentrations of constituent matter within the water column, which provides an estimate of water quality (Chen et al., 2015).
Figure 1: Schematic overview of the path of electromagnetic spectrum between the Sun, a waterbody, and a sensor
Source: Adopted from Batina and Andrija (2023)
The Remote Sensing Opportunity for Water Quality Monitoring
Development of remote sensing techniques for monitoring water quality began in the early 1970s, and it provides a clear opportunity to address the challenges of conventional water quality monitoring (Bazel et al., 2021). More recent interest in developing long-term environmental monitoring projects has furthered the development of new techniques for remote sensing primarily because of its ability to provide a perspective not available through any other avenue (Liu et al., 2003). Rapid development of remote sensing techniques, especially the launch of hyper-resolution satellites, enables the application of remotely sensed data for monitoring large-scale and long-term water quality (Haibo et al., 2022). The shift towards the use of remotely sensed data for aquatic ecosystem monitoring has taken many forms including analysis of hydrologic recharge, volumetric storage fluctuation rates, hydrologic connectivity, flow velocity of river systems and more (Pavelsky and Smith, 2009). Not only does remote sensing change the spatial and temporal scales which can be monitored, but it also elucidates the relationship between the hydrology, landscape, and organisms within the ecosystem (Liu et al., 2003). Several remote sensing programs provide historical data for studies of trends in water quality and the potential impacts of land use and land cover change on water quality. The real time availability of remote sensing data makes it possible to integrate it into early warning systems to protect the public from harmful algal blooms (Hicks et al., 2013). Continuous developments in satellite and sensor technologies, and research into parameter retrieval algorithms, will increase the use of remote sensing methods for water quality monitoring in the future (Dekker and Hestir, 2012).
Water Quality Parameters
Conventional water quality monitoring system includes three aspects of water indicators: physical indices (e.g., temperature, turbidity, smell, and electrical conductivity), chemical indices (e.g., pH, DO, COD, BOD, TOC, heavy metal ion and non-metallic poison), and microbiological indices (e.g., total bacteria and total coli forms). Based on the remote sensing techniques, water quality indicators can be classified into optically active and non-active parameters (Table 1).
Optically Active Constituents
Substances that alter the underwater light field are known as optically active constituents (OACs). They include the attributes that are related to water quality, such as chlorophyll a, suspended particulate matter, turbidity and coloured dissolved organic matter (CDOM) which affect the radioactive transfer process of waves through different absorption of the spectrum.
Chlorophyll-A
Chl–a is photo synthetically active compounds that is used as a proxy measure of total algal biomass in aquatic systems (Kutser, 2009). The algal biomass of a water body controls its overall biological productivity, also known as trophic state, making it an ideal indicator of ecosystem integrity (Caballero and Navarro, 2021). While not all algal blooms are inherently harmful, blooms containing certain species, most commonly phycocyanin-producing cyanobacteria, are toxic to humans, livestock, and wildlife (Svircevet al., 2013). Optically, the spectral signature of chl-a varies depending on its concentration in relation to other water quality parameters and the composition of phytoplankton phenotypes producing the signal (Zhou et al., 2018). For low biomass, oligotrophic to meso trophic water bodies, the chl-a spectrum is characterized by a sun-induced fluorescence peak around 680 nm (Gower, 2004). The main absorption bands of chlorophyll are red and blue-violet light, which affect the optical properties of water bodies during radiation transmission. For high biomass, eutrophic water bodies, the florescence signal is masked by absorption features and backscatter peaks centred at 665 nm and 710 nm respectively (Matthews et al., 2012). The ratio between these two wavelengths has been used to accurately estimate chl-a concentrations in numerous studies. Various data sources, including hyperspectral data and multispectral data, have been well applied to the estimation of Chl–a concentration (Pyo et al., 2019). The concentration of Chl–a can be retrieved by using Band ratio model with the reflectance of the maximum reflectance value in the Nir band and the minimum reflectance value in R., first order differential model (Rundquist et al., 1996), three-band model using three-band data (Gitelson et al., 2008) and Artificial Intelligence model using an empirical neural network (Zhang et al., 2020).
Total Suspended Matter
Total suspended matter (TSM) concentration is one of the key parameters in main water quality components. TSM includes inorganic and organic particles suspended in water. The high concentration of TSM in aquatic ecosystem will disturb the optical properties of water, such as the depth of the true light layer, and then influence the growth of aquatic plants and the primary productivity of water bodies (Hou et al., 2017). Researchers have developed some methods for estimating the concentration of TSM using remote sensing data, and have achieved good results in practical applications (Chen et al., 2015).
Researchers have successfully calculated TSM concentrations using remote sensing data in practical applications (Chen et al., 2015). TSM concentrations estimated from remote sensing have been correlated with several optically inactive water quality indices to estimate small-scale phosphorous, mercury, and other metals.
Turbidity
In the context of water quality (WQ) monitoring, water turbidity represents a key factor controlling for water quality assessment. Water turbidity is an optical property of water, which scatters and absorbs the light rather than transmit it in straight lines. Turbidity is an optically active property of water that indicates the presence of particles in the water column that can provoke the scattering or absorption of light (Avdan et al., 2019). This is an important WQ indicator (Quang et al. 2017) adopted by most monitoring programs (Odermatt et al., 2008). Turbidity can be caused by the presence of matter of phytoplanktonic origin (Sabat-Tomala et al., 2018) and materials of mineral origin from soil erosion (Menken et al., 2006). High levels of turbidity imply lesser water transparency and can cause imbalances and damage to biological organisms (Quang et al. 2017). Turbidity detection assumes great significance for water management and environmental protection of aquatic ecosystems as it is a useful measure of light availability under water and is therefore related to many ecosystem processes (Dekker and Hestir, 2012). Due to the strong influence of suspended solids on water clarity, reflectance at 700 nm is most often used to derive turbidity from remotely sensed signals (Hicks et al., 2013). These turbidity measurements can provide important information about water quality, sediment transport, and ecological dynamics. Literature indicates that a good accuracy in turbidity prediction is possible using visible bands (Liu et al., 2019) and the combination of visible and infrared bands (Alparslan et al. 2010). Good results are described with both empirical and analytical models but the choice of spectral regions for the development of turbidity estimation algorithms may also be dependent on the season, especially in eutrophic environments (Dekker and Hestir, 2012).
Secchi Disk Depth (SDD)
The water transparency assessment represents an important factor in the monitoring and management of water resources for many reasons. The Secchi depth is inversely correlated with the amount of TSM present in the waterbodies. Therefore, remote sensing can be an ideal tool for monitoring water transparency and estimating the SDD. Lee et al. (2015) introduced a model to estimate the SDD, which unlike the classical model that relies strongly on the beam attenuation coefficient, the new model relies only on the diffuse attenuation coefficient at a wavelength corresponding to the maximum transparency for such interpretations. The main remote sensing algorithms applied to estimate SDD were based on empirical or semi analytical relationships (Liu et al., 2019). Red and blue regions of the electromagnetic spectrum are widely used to assess water clarity. The reflectance in the red band is directly proportional to the increase in turbidity, while the blue band responds inversely in proportion to the optical properties of substances that promote greater water turbidity ((Liu et al., 2019). Combinations of visible bands (red, green, and blue) have often been used in the construction of algorithms for SDD estimates in optically complex continental water (Vundo et al., 2019).
Remote sensing models with good levels of significance for SDD estimation have been applied to MODIS (Liu et al. 2019), IKONOS (Keith et., 2014) and MERIS products (Zhang et al., 2016) in large water bodies > 160 ha (Liu et al. 2019). Studies that measure SDD from optical sensors like MSI and OLCI sensors have also a relatively high level of success in part because SDD is a direct consequence of all optical characteristics of water and the elements it contains ( Underberg et al. 2020). The existing literature showed that SDD can be quantified using visual spectral bands and various band ratios. For example, Thiemann and Kaufmann (2002) used HyMap and CASI data for Secchi disk transparency Mecklenburg Lake District, Germany, and they used the area between a base line and the spectrum from 400 to 750 nm and found a good correlation with the in situ measured Secchi disk transparency (SDT). Powell et al. (2008) suggested a regression equation related to in-situ Secchi disk transparency measurements by using the blue, green, and red bands of TM.
Water Temperature
Water temperature (WT) is an important parameter for the physical and biochemical processes occurring within water as well as in air-water interactions because temperature regulates physical, chemical, and biological processes in water. Hence, WT is regarded as one of the crucial indicators of water quality and ecosystem health (Gholizadeh et al., 2016). Water temperature also influences the solubility, and thus availability of various chemical constituents in water. Most importantly, this parameter affects dissolved oxygen concentrations in water; as oxygen solubility decreases with increasing water temperature. It is also very important to analyse the temporal variations due to seasonal changes. Remote sensing can provide accurate surface WT measurements, and water temperature parameter retrieval using remote sensing techniques has been an active area of research in recent years. Satellite-based remote sensing platforms equipped with thermal sensors can provide valuable information on water surface temperature over large areas and at regular intervals. These sensors measure the emitted thermal radiation from the water surface, which is correlated with water temperature. To retrieve water temperature from remote sensing data, various algorithms are employed that utilize the relationship between surface temperature and the captured radiance values. These algorithms incorporate atmospheric correction, which accounts for the interference of atmospheric conditions on the thermal signals (Batina and Krtalic, 2023). These data are crucial for understanding the thermal dynamics of aquatic ecosystems, assessing the impacts of climate change, and managing water resources effectively.
Coloured Dissolved Organic Matter (CDOM)
CDOM is an optically active component of the coloured fraction of organic matter dissolved in the water (Coelho et al., 2017). It consists of a mixture of organic molecules originating from sources both allochthonous (humic and fulvic substances from the land portion of the drainage basin) and autochthonous (phytoplankton and other organisms present in the water body) (Sabat-Tomala et al., 2018). CDOM is an important WQ indicator that impacts on the potability of the water (Chen et al., 2017) and its property to absorb solar radiation results indirectly in the protection from pathogenic organisms resulting from the photochemical reactions that occur in the interaction of light and water (Kutser et al., 2005). The presence of CDOM in an aquatic environment can provoke its brownification, a phenomenon causing the water to acquire a yellow/brown tint as a response to the high concentration of organic matter. As concentrations increase, absorption of low-wavelength light by CDOM regulates the light availability of primary producers, controlling net productivity and trophic structure (Carvalho et al., 2013). Continued monitoring of CDOM directly, and as a proxy for total dissolved organic carbon, provides a better understanding of carbon inputs and processing in freshwater systems.
The phenomenon can negatively affect the quality of the water by changing the amount of nutrients, the pH, the thermal stratification, and the whole food chain. Unlike TSM or chl-a, there are no recognized specific spectral band associated with CDOM. However, the visible absorption bands (blue and green) associated with other bands (red edge, NIR) is important to increase chances of producing good estimates (Hestir et al., 2015).
Optically Inactive Constituents
Optically inactive water quality parameters refer to those parameters that do not directly affect the reflectance or absorption of light in water. The second category includes the indicators without explicit optical properties (e.g., DO, TN, TP, BOD, COD). Such water quality parameters with poor optical properties and low signal-to-noise ratio (Gholizadeh et al., 2016) are often assessed based on statistical relationships with other optically active indicators. There are several indirect methods for estimating optically inactive parameters. Publications leveraging relationships between optically inactive constituents, which have no detectable spectral signal, and the optically active constituents listed above have provided remote sensing models for between optically inactive constituents like nitrogen and phosphorous (Torbick et al., 2013) and dissolved oxygen (Toming et al., 2016). The optically active water quality variables such as chlorophyll-a, Secchi disk depth (SDD) and total suspended solids (TSS) have a high correlation coefficient with TN and TP concentrations.
Artificial neural network (ANN) model and linear regression (LR) model were mostly utilized to determine the relationship of OLI data and TP and TN concentrations. Wu et al. (2010), Song et al. (2011) and Yang et al. (2012) showed that there is a high correlation between TP-TN and total suspended solids (TSS), chlorophyll-a and Secchi disk depth (SDD). Therefore, the single band or band ratio, which is used to estimate TSS, SDD and chlorophyll-a, can be utilized for predicting the TN-TP (Isenstein and Park, 2014; Chen and Quan, 2012; Song et al., 2011). For example, the band ratio of Landsat TM data which had the highest correlation with SSD and chlorophyll-a was utilized by Wu et al. (2010) to estimate the concentration of TP in the Qiantang River in China. Shang et al. (2023) have also used Sentinel-2 images using four machine learning algorithms (eXtreme Gradient Boosting (XGBoost), Support Vector Regression (SVR), Random Forest (RF), and Artificial Neural Network (ANN)) to retrieve chlorophyll-a (Chl-a), dissolved oxygen (DO), and ammonia-nitrogen (NH3-N) for inland reservoirs in China, and the water quality parameters obtained by the machine learning model were very close to the in-situ measurements and could be well implemented for the retrieval of non-optically active parameters for small and medium-sized Water.
Table 1: The most commonly measured qualitative parameters of water by means of remote sensing
Water quality parameter Abbreviation Units Optical activity
Chlorophyll-a Chl-a Mg/l active
Secchi Disk depth SDD M active
Water temperature WT 0c active
Turbidity TUR NTU active
Total Suspended Matter TSM Mg/l active
Electrical conductivity EC active
Sea Surface Salinity SSS PSU active
Coloured Dissolved Organic Matter CDOM Mg/l active
Total Organic Carbon TOC mg/L active
Dissolved Oxygen DO mg/L inactive
Chemical Oxygen Demand COD mg/L inactive
Biochemical Oxygen Demand BOD mg/L inactive
Total Nitrogen TN mg/L inactive
Ammonia Nitrogen NH3-N mg/L inactive
Total Phosphorus TP mg/L inactive
Orthophosphate PO4 mg/L inactive
Available Data Sources for Remote Sensing Water Quality Retrieval
Observing sensors are divided into two main categories based on the platforms on which they are situated. Airborne sensors are those that are mounted on a platform within the Earth’s atmosphere (i.e., a boat, a balloon, a helicopter, or an aircraft), and spaceborne sensors are carried by a spacecraft or satellite to locations outside of the Earth’s atmosphere. Understanding the properties of these sensors is necessary to choose an appropriate sensor for the objectives of the study. Therefore, various remote sensing satellites (Table 2) and airborne systems (Table 2) commonly used in water quality assessments, along with their spectral properties including spatial resolution, spectral bands, and revisit interval are presented. This information is helpful when designing water quality assessment studies, and can be used for the selection of appropriate sensors among many other available sensors.
Satellite-Borne Remote Sensing Data
Multispectral Data
Multispectral data typically consists of a few discreet spectral bands (usually ranging from 3 to 30 bands) across the visible and near-infrared spectrum. These bands can be used to estimate water quality parameters such as chlorophyll-a concentration, turbidity, and suspended solids. By analysing the reflectance values at specific bands, mathematical models can be developed to correlate these values with the desired water quality parameters. Multispectral data, available for remote sensing water quality retrieval, includes MSS, TM, ETM+, OLI, ESA’s Sentinel-2, ENVISAT MERIS, France’s SPOT satellite data, and NOAA’s AVHRR and China’s GF series (Batur and Maktav, 2018). Considering the spatial, temporal, and spectral resolution and the accessibility, Landsat series data are the most commonly used for water quality monitoring, such as TSM, COD and TP (Vakili and Amanollahi, 2019; Wang et al., 2019). Generally, due to resolution limitations, multi-spectral remote sensing water quality retrieval models are mainly constructed with empirical methods, which are more suitable for a certain specified period or water area.
Hyperspectral Data
Hyperspectral data provides a much higher spectral resolution than multispectral data, typically consisting of hundreds of narrow contiguous bands across the electromagnetic spectrum. This detailed spectral information allows for more accurate and precise retrieval of water quality parameters. Hyperspectral data can be used to estimate specific water quality parameters such as chlorophyll-a, dissolved organic matter, and mineral content (Yang et al., 2022). These parameters can be retrieved by analysing the absorption and reflectance patterns at different wavelengths in the data. Hyperspectral satellites have multiple bands with about 0.01 m spectral resolution. Hyperspectral data, including the USA’s Hyperion data and HIS data of China’s HJ-1 satellite have been used for water quality retrieval (Cao et al., 2018). Higher spectral resolution data have plentiful bands, which can be selected precisely and optimally for establishing water quality retrieval methods to distinguish the spectral mixing differences in multispectral data, thus greatly improving the accuracy of water quality parameters retrieval algorithms and showing good application potential (Hestir et al., 2015).
Table 2: Available satellites for remote sensing water quality retrieval (bands more than 5).
Satellite Sensor Launch Date Spatial resolution (m) Spectral Resolution Band Temporal Resolution (Day)
Multi-spectral NIMBUS-7 CZCS 1978.10 825 6 6
Landsat-5/7/8/9 1984-2020 30 5 16
SeaWiFS 1997.8 1130 8 16
NOAA-16AVHRR 2000.10 1100-4000 6 9
EO-1 AL1 2000.11 10 9 16
WorldView-2/3 2009/2014 1.85/1.24 8 1.1
MERIS 2002.3 300-1200 15 1
MODIS 1900.12 250-500-1000 9 0.5
Landsat-8 OLI 2013.2 30 7 16
Sentinel-2 A 2015 10 13 5
Sentinel-3A/ OLCI 2016 300-1600 21 27
Sentinel-2 B 2017 10 13 10
Sentinel-3B 2018 300-1200 21 4
Hyper-spectral HY-1A COCTS 2002.5 1100 10 3
PROBA CHRIS 2001.10 18-36 19 7
Hyperion 2000.11 30 42 16
HJ-1A HSI 2008.9 100 128 4
MICO 2009.9 100 128 10
VIRS 2011.10 375-750 22 0.5
OHS 2018.4 10 32 2
GFS-AHSI 2018.5 30 330 3
ZYI-02D 2018.9 30 166 3
ZK-VNR-FPG4S0 / 0.09 270 /
Gala Sky-mini / 0.04 176 /
Non-Satellite Remote Sensing Data
In addition to satellite remote sensing data, non-satellite remote sensing techniques can also be used for water quality parameter retrieval. By using sensors mounted on aircraft or drones, airborne remote sensing can provide detailed spatial information and better temporal resolution compared to satellites. With the rapid progress of UAV technology, the light and small UAV system, equipped with multi-spectral camera, high spectrometer, infrared sensor, and Lidar, is convenient and effective in water environment management (Ouma et al., 2018). Similar techniques and algorithms as satellite remote sensing, such as spectral analysis and mathematical models, can be applied to airborne imagery to retrieve water quality parameters. For example, the USA’s AVIRIS with 220 channels and Canada’s Compact Airborne Spectrographic Imager (CASI) with 48 channels were extensively used for water environment monitoring. Similarly, airborne Chinese Imaging spectrometer (CIS) data have also applied to monitor water environment. For example, the SVC HR-1024 portable field spec radiometer has 1024 bands with a range of 350~2500 nm (Li et al., 2015). Non-satellite spectrometers have the strength of having higher spectral and spatial resolution and they can provide continuous ground feature spectral curves. Although shortwave infrared (SWIR) and near infrared (NIR) are mainly used for turbid waters and clear waters individually, the combination of SWIR and NIR can improve the application (Liu et al., 2019). In addition, ground-based remote sensing techniques involve collecting data from the water surface or near-shore areas using handheld sensors, spectroradiometers, or underwater cameras, and provide close-range measurements and can be used for monitoring water quality parameters in specific locations. They are particularly useful for studying near-shore environments, littoral zones, and small water bodies. Ground-based remote sensing data can be processed using similar techniques as satellite or airborne data for water quality parameter retrieval.
Table 3: Specification of the more commonly used airborne sensors in water quality assessments.
Types of Sensors Number of Bands Spectral Range (µm) Resolution (m)
Multispectral MIVIS 102 VIS/NIR (28), MIR (64) TIR (10) VIS (0.43–0.83), NIR (1.15–1.55), MIR (2.0–2.5) TIR (8.2–12.7) 3 to 8 depending
on altitude
MSS 0.42–14.00 25
Hyperspectral AVIRIS 224 0.40–2.50 17
HYDICE 210 0.40–2.50 0.8 to 4
HyMap 128 0.40–2.50 3 to 10
APEX Up to 300 VIS/NIR (114), SWIR (199) VIS/NIR (0.38–0.97), SWIR1 (0.97–2.50) 2 to 5
CASI-1500 Up to 228 0.40–1.00 0.5 to 3
EPS-H VIS/NIR (76), SWIR1 (32), SWIR2 (32) TIR (12) VIS/NIR (0.43–1.05), SWIR1 (1.50–1.80), SWIR2 (2.00–2.50), TIR (8–12.50) Dependent upon
flight (min 1 m)
DAIS 7915 VIS/NIR (32), SWIR1 (8), SWIR2 (32), MIR (1),
TIR (12) VIS/NIR (0.43–1.05), SWIR1 (1.50–1.80), SWIR2 (2.00–2.50), MIR (3.00–5.00), TIR (8.70–12.30) 3 to 20 depending
on altitude
AISA Up to 288 0.43–0.90 1
Water quality retrieval algorithms and modelling approaches
Retrieval algorithms can be developed in a variety of ways, ranging from simple empirical relationships between in situ samples and radiance reflectance at certain wavelengths, through to spectral additive models based on radiative transfer theory (Politi et al., 2015). The basic principle of inversion of water quality using remote sensing methods consists in situ water quality monitoring data and equivalent remote sensing imagery for model establishment. Retrieval algorithms are used to estimate the concentration of a water quality parameter from the spectral water-leaving radiance measured by a sensor. There are several algorithms and modelling approaches used for retrieving water quality parameters from remote sensing data. Here are some commonly employed methods (Table 4).
Empirical Models
Empirical algorithms are established through statistical relationships between water quality parameters and remotely sensed data. Empirical algorithms require in situ data on each water quality variable to determine a statistical relationship between the reflectance of spectral bands and the concentration of constituents at the time of image capture (Olmanson et al., 2015). These algorithms are usually based on a training dataset that includes field measurements of water quality parameters and corresponding remote sensing data. The algorithm then uses this training dataset to develop a mathematical relationship that links the observed spectral signatures with the target water quality parameter. Then the inversion algorithm is obtained via the statistical analyses between the water quality parameters and selected characteristic bands or band combinations (Cheng et al., 2015, Zhou and Wang, 2015). Typical empirical methods include linear regression, single-band method, channel-combination, principal component analysis and band-combination method (e.g., band-ratio, band difference). Most empirical models use multivariate regression because inland waterways are optically complex (Ouma et al., 2018). The empirical method is most commonly used to assess TUR, chl-α, and trophic status (Giardino et al., 2010).
In addition to purely empirical approaches, there is another family of empirical model so called Semi-empirical methods that combine empirical and analytical methods (Keller, 2001). Semi-empirical models require statistical and measured spectral analysis (Li, 2009). The semi-empirical method integrates waterbody reflectivity with measured parameter concentration, giving it physical significance and ease of implementation and improve the parameter’s spectral properties and reduce optical parameter noise (Keller, 2001). Even though it requires large amount of in situ measured data limits its temporal and spatial applicability, semi-empirical models are more generalizable than completely empirical ones, and it is often used to assess parameters like Chl-α, TSM, CDOM, SDD and TUR (Hunter et al., 2010; Yang et al., 2022).
Analytical Methods
The analytical mode (AM) uses bio-optical models and radiation transmission models to simulate the propagation of light in the atmosphere and water bodies to describe the relationship between water quality components and the radiance or reflection spectrum of off-water radiation (Yang et al., 2022). The analytical method is also called physical methods require theoretical analyses of spectral data not statistical analyses like empirical and semi-empirical methods (Batina and Krtalic, 2023). The analytical method’s physical mechanism can simultaneously identify all water parameters using well-established parameter properties and large in situ data (Gholizadeh et al., 2016) and portability is good, but it requires a highly accurate measuring instrument, high application costs, and challenges to widespread adoption (Keller, 2001). For purely analytical models, the inverse equation is parameterized based purely on light physics; however, these are rarely used for optically complex waters where the interactions of numerous water quality constituents become difficult to model.
As results, semi-analytical models, which incorporate in situ measurements to parameterize the inverse equation, are the primary form of physics-based algorithms developed for inland water quality remote sensing retrievals (Matthews, 2011). This modelling approach evolved from the reflectance approximation developed by Morel and Prieur (1977) who studied turbidity and chlorophyll in ocean waters (Batina and Krtalic, 2023). This type of algorithms combines radiative transfer models with empirical relationships to estimate water quality parameters. These algorithms consider the inherent optical properties of water and use physical models to simulate the light interaction with the water column. Semi-analytical algorithms use measured remote sensing data to retrieve the optical properties and then relate them to the desired water quality parameter using empirical relationships. The NASA MODIS algorithm for chlorophyll-a retrieval is an example of a semi-analytical algorithm. Semi-analytical and analytical methods are used to retrieve mainly optically active parameters, such as chl-α, TSM, CDOM, and SDD (Wang et al.,, 2019). Dekker (1991) has developed applications of semi-analytical models to examining chl-a, TSS, and CDOM across large spatiotemporal scales.
Machine Learning Models
In recent years, increases in computational capacity and available data have created opportunities for novel approaches to data analysis. As a result, Machine learning algorithms, such as support vector machines, random forests, and neural networks, have gained popularity in water quality retrieval and have shown promise in accurately estimating water quality parameters across a variety of spatiotemporal scales (Chang et al., 2014; Lary et al., 2016; Lin et al., 2018; Hafeez et al.,2019). Machine learning techniques can handle complex relationships and non-linearities in the data, offering potentially improved accuracy and robustness. These approaches use large datasets of remote sensing and in situ measurements to train models that can predict water quality parameters.
To avoid over fitting, machine learning methods require the provision of separate training and testing datasets that contain representative samples of the parameters of interest. The power and scalability of most machine learning algorithms is dependent on the quality and range of the training and testing data. Given the proper inputs, these algorithms can produce generalizable models that capture complex, non-linear relationships between remotely sensed reflectance and bio-geophysical parameters. Song (2011) has found reductions in root mean square error of 76% and 65%, when comparing traditional regression techniques to artificial neural networks, respectively, while modelling chl-a and turbidity in Lake Chagan, China. Similarly, Xiang et al. (2021) found a 20% increase in trophic state classification accuracy when using machine learning compared to multivariate regression.
Table 4: Satellite sensors and water quality retrieval algorisms
Satellite/remote sensing data Water quality parameters involved Algorithm used Sources
MODIS/Aqua salinity, Chl-a, temperature, CDOM Polynomial regression was reported significant Wouthuyzen al., 2020
MERIS Chl-a The MLP outperforms the SVM regression to capture satellite Chl-a Martinez et al., 2020
GEE DO, temperature, salinity, Chl-a, and Ph RF bestowed significant accuracy Yniguez and Ottong, 2020
Landsat-8 Chl-a, TP, TN The ensemble of ANN, SVM, RF and KNN increased the inversion water quality parameter results Yniguez and Ottong, 2020
Sentinel-2 TP, TN, COD ANN exhibited the best performance followed by RF and SVM regression Guo et al., 2021
Sentinel-3/OLCI Chl-a Hierarchical Bayesian Spatio-temporal modelling shows high performance with low Deviance Information Criterion (DIC) value Myer et al., 2020
MODIS/Aqua Chl-a LR was used to compare retrieval algorithm, OC3M Abbas et al., 2019
Landsat 8/OLI Chl-a SVM regression showed slightly superior performance than ANN Peterson et al., 2020
SMOS temperature, salinity RF outperformed SVM regression Ruescas et al., 2018
Sentinel-2A Chl-a, suspended solids RF bestowed significant accuracy Qasem et al., 2022
Landsat-8, Sentinel-2 BGA, Chl-a, fDOM, DO, SC, and turbidity The proposed method, SVM regression, MLR and ELM regression performed significantly on particular parameters Peterson et al., 2020
Sentinel-2 Microphytobenthos RF was reported significant Martinez et al., 2020
VIIRS Chl-a In comparison to in-situ data, RF has a greater accuracy on satellite observations Park et al., 2020
GEE DO, temperature, salinity, Chl-a, and p RF bestowed significant accuracy Martinez et al., 2020
CZCS, SeaWiFS Chl-a SVM has high efficiency for the study
Landsat-5, Landsat-7, Landsat-8 suspended solids, Chl-a, turbidity ANN outperforms SVM, RF and Cubist Hafeez et al., 2020
MODIS Chl-a The RF regression ensemble produced promising outcomes Chen et al., 2019
MODIS/Terra Turbidity, temperature ANN Chen et al., 2019
SeaWiFS, MERIS, Chl-a, temperature ET model shows better performance than the RF model Park et al., 2019
Landsat 8/OLI Chl-a, suspending solids, TP, TN Multiple Regression was reported significant Lim and Choi, 2015
Sentinel-3/OLCI Chl-a, CDOM, suspended solids SVM, RF, KRR and GPR are very efficient except for RLR Ruescas et al., 2018
MODIS/Aqua Temperature ANN produced promising outcomes Sunder and Ramakrishnan, 2017
MODIS/Aqua Chl-a, TP, TN, SDD ANN strongly demonstrates effectiveness and reliability Chang et al., 2017
MODIS Chl-a SVM outperforms the classical approach Wattelez et al., 2016
GOCI Phytoplankton, suspended solids, CDOM SVM produced promising outcomes
MERIS, MODIS Chl-a SVM combine with Linear, polynomial, RBF, sigmoid regression analysis improves the precision of the algorithm Davila and Zaremba, 2016
VIIRS temperature, salinity, Chl-a MLP method produced promising outcomes Park et al., 2019
Landsat-5/TM suspended solids SVM method is used for the iterative classification process Park et al., 2019
MODIS TP ANN strongly demonstrates effectiveness and reliability Chang et al., 2017
MERIS Suspended solids, Chl-a SVM use for estimation Tang et al., 2019
SeaWiFS orthophosphate, silicate, salinity, temperature In terms of determining temporal and spatial variability, MLR performed well Green and Gould, 2008
SeaWiFS CDOM, suspended solids, temperature, salinity MLR produced promising outcomes Green and Gould, 2008
MODIS Chl-a CNN performs better than SVM regression Yu B et al., 2020
Landsat-8, GEE, Sentinel-2 turbidity, suspending solids, TP, TN SVM provided higher accuracy than ANN Govedari and Yakovlev, 2019
CONCLUSION
Increasing stresses on aquatic ecosystems all over the world have generated the need for cost effective and quick water monitoring techniques. Hence, with advancement in space science and the increasing use of computer applications, remote sensing-based water quality monitoring have been practiced across the world, and has proven to give better results in both temporal and spatial scale. This review summarizes the space-born and airborne data sources, retrieval algorithms and water quality parameters. And the review revealed that a series of remotely sensed data including multispectral and hyperspectral data are widely used in water quality monitoring and provide flexible and efficient solutions satisfying water quality retrieval with higher temporal, spatial and spectral resolution. Regarding the inversion methods, the application of remotely sensed data for water quality monitoring and evaluation can be performed using different methods, i.e., empirical, analytical, and more advanced machine learning algorithms used for retrieval of water quality parameters and to improve the inversion accuracy of different models. Optically active water quality parameters like SSM, turbidity, Chl–a CDOM, water transparency and water temperature can be estimated using simple empirical and analytical models. In addition, with the rapid development of machine learning algorithms, more new artificial intelligence (AI) is used in remote sensing inversion of non-optical active substances such as DO, COD, BOD, TN and TP. It is recommended that aquatic ecologists, water managers, and authorities make use of remote sensing and its ability to provide a comprehensive view of water bodies for sustainable management of water resources.
REFERENCES
Abbas M TN, Melesse A M, Scinto L J and Rehage J S (2019). Satellite estimation of chlorophyll-a using moderate resolution imaging spectroradiometer (MODIS) sensor in shallow coastal water bodies: Validation and improvement Water (Switzerland) 11 1621.
Alparslan, E.; Coskun, H.G.; Alganci, U. (2010). Water Quality Determination of Küçükçekmece Lake, Turkey by Using Multispectral Satellite Data. The Sci. World J, 9, 1215–1229.
Avdan ZY, Kaplan G, Goncu S and Avdan U (2019). Monitoring the water quality of small water bodies using high-resolution remote sensing data. ISPRS International Journal of Geo-Information 8(12): 553.
Batina A.and Andrija K. (2023). A Review of Remote Sensing Applications for Determining Lake Water Quality, pre review org, doi:10.20944/preprints202309. 0489.v1.
Caballero I. and Navarro G. (2021). Monitoring cyanoHABs and water quality in Laguna Lake (Philippines) with Sentinel-2 satellites during the 2020 Pacific typhoon season. Science of The Total Environment 788:147700.
Cao Z, Ma R, Duan H, Pahlevan N, Melack J, Shen M and Xue K 2020 A machine learning approach to estimate chlorophyll-a from Landsat-8 measurements in inland lakes Remote Sens. Environ. 248 111974.
Chang N Bin, Bai K and Chen C F (2017). Integrating multisensor satellite data merging and image reconstruction in support of machine learning for better water quality management J. Environ. Manage. 201 227–40.
Chang, N.B., Vannah, B.W., Yang, Y.J., Elovitz, M. (2014). Integrated data fusion and mining techniques for monitoring total organic carbon concentrations in a lake. Int. J. Remote Sens, 35, 1064–1093.
Chen J, Zhu WN, Tian YQ and Yu Q (2017). Estimation of coloured dissolved organic matter from Landsat-8 imagery for complex inland water: case study of Lake Huron. IEEE Transactions on Geoscience and Remote Sensing 55(4): 2201–2212.
Chen S, Hu C, Barnes B B, Xie Y, Lin G and Qiu Z (2019). Improving ocean color data coverage through machine learning Remote Sens. Environ. 222 286–302.
Chen S, Hu C, Barnes B B, Xie Y, Lin G and Qiu Z (2019). Improving ocean color data coverage through machine learning Remote Sens. Environ. 222 286–302.
Chen, J.; Quan, W.T.; Cui, T.W.; Song, Q.J. (2015). Estimation of total suspended matter concentration from MODIS data using a neural network model in the China eastern coastal zone. Estuarine Coast. Shelf Sci., 155, 104–113.
Coelho C, Heim B, Foerster S, Brosinsky A and Ara´ujo JC (2017). In situ and satellite observation of CDOM and chlorophyll-a dynamics in small water surface reservoirs in the Brazilian semiarid region. Water 9(12): 913.
Davila J C and Zaremba M B (2016). An iterative learning framework for multimodal chlorophyll-a estimation IEEE Trans. Geosci. Remote Sens. 54 7299–308.
Dekker, A.G.; Seyhan, E.; Malthus, T.J. (1991). Quantitative Modeling of Inland Water Quality for High-Resolution MSS Systems. IEEE Trans. Geosci. Remote Sens., 29, 89–95.
Gholizadeh, M.; Melesse, A.; Reddi, L. (2016). A Comprehensive Review on Water Quality Parameters Estimation Using Remote Sensing Techniques. Sensors, 16, 1298.
Giardino, C.; Bresciani, M.; Cazzaniga, I.; Schenk, K.; Rieger, P.; Braga, F.; Matta, E.; Brando, V.E. (2014). Evaluation of multi-resolution satellite sensors for assessing water quality and bottom depth of lake garda. Sensors, 14, 24116–24131.
Govedarica M and Jakovljevic G (2019). Monitoring spatial and temporal variation of water quality parameters using time series of open multispectral data SPIE Proceedings 11174 55.
Gower, J., King, S., Borstad, G., Brown, L. (2005). Detection of intense plankton blooms using the 709 nm band of the MERIS imaging spectrometer. Int. J. Remote Sens., 26, 2005–2012.
Green R E and Gould R W (2008). A predictive model for sateHite-derived phytoplankton absorption over the Louisiana shelf hypoxic zone: Effects of nutrients and physical forcing J. Geophys. Res. Ocean. 113 1–17.
Guo H, Huang J J, Chen B, Guo X and Singh V P (2021). A machine learning-based strategy for estimating non-optically active water quality parameters using Sentinel-2 imagery Int. J. Remote Sens. 42 1841–66.
Hafeez S, Kong H, Wong M S and Nazeer M (2019). Comparison of Machine Learning Algorithms for Retrieval of Water Quality Indicators in Case-II Waters : A Case Study of Hong Kong. Remote Sens. 11 617.
Hafeez S, Kong H, Wong M S and Nazeer M (2019). Comparison of Machine Learning Algorithms for Retrieval of Water Quality Indicators in Case-II Waters : A Case Study of Hong Kong. Remote Sens. 11 617.
Hestir, E.L.; Brando, V.E.; Bresciani, M.; Giardino, C.; Matta, E.; Villa, P.; Dekker, A.G. (2015). Measuring freshwater aquatic ecosystems: The need for a hyperspectral global mapping satellite mission. Remote Sens. Environ. , 167, 181–195.
Hicks, B.J.; Stichbury, G.A.; Brabyn, L.K.; Allan, M.G.; Ashraf, S. (2013). Hindcasting Water Clarity from Landsat Satellite Images of Unmonitored Shallow Lakes in the Waikato Region, New Zealand. Environ. Monit. Assess. 185, 7245–7261.
Hou, X.; Feng, L.; Duan, H.; Chen, X.; Sun, D.; Shi, K. (2017). Fifteen-Year Monitoring of the Turbidity Dynamics in Large Lakes and Reservoirs in the Middle and Lower Basin of the Yangtze River, China. Remote Sens Environ, 190, 107–121.
Hunter, P.D.; Tyler, A.N.; Carvalho, L.; Codd, G.A.; Maberly, S.C. (2010). Hyperspectral Remote Sensing of Cyanobacterial Pigments as Indicators for Cell Populations and Toxins in Eutrophic Lakes. Remote Sens Environ, 114, 2705–2718.
Isenstein, E.M.; Park, M.-H. (2014). Assessment of Nutrient Distributions in Lake Champlain Using Satellite Remote Sensing. J Environ Sci (China), 26, 1831–1836.
Keith, D.J.; Schaeffer, B.A.; Lunetta, R.S.; Gould, R.W.; Rocha, K.; Cobb, D.J. (2014). Remote Sensing of Selected Water-Quality Indicators with the Hyperspectral Imager for the Coastal Ocean (HICO) Sensor. Int J Remote Sens, 35, 2927–2962.
Keller, P.A. (2001). Imaging Spectroscopy of Lake Water Quality Parameters; Remote Sensing Laboratories, Department of Geography, University of Zürich: Zürich, Switzerland.
Kirk, J.T.O. (1994). Light and Photosynthesis in Aquatic Ecosystems; 2nd ed.; Cambridge University Press: Cambridge, UK.
Kutser T, Pierson DC, Kallio KY, Reinart A and Sobek S (2005). Mapping lake CDOM by satellite remote sensing. Remote Sensing of Environment 94(4): 535–540.
Kutser, T. (2009). Passive optical remote sensing of cyanobacteria and other intense phytoplankton blooms in coastal and inland waters. Int. J. Remote Sens., 30, 4401–4425.
Lary, D.J.; Alavi, A.H.; Gandomi, A.H.;Walker, A.L. (2016). Machine learning in geosciences and remote sensing. Geosci. Front, 7, 3–10.
Lee, Z.; Shang, S.; Hu, C.; Du, K.;Weidemann, A.; Hou,W.; Lin, J.; Lin, G. (2015). Secchi disk depth: A new theory and mechanistic model for underwater visibility. Remote Sens. Environ., 169, 139–149.
Li, J.; Chen, X.; Tian, L.; Huang, J.; Feng, L. Improved capabilities of the Chinese high-resolution remote sensing satellite GF-1 for monitoring suspended particulate matter (SPM) in inland waters: Radiometric and spatial considerations. ISPRS J. Photogramm. Remote Sens. 2015, 106, 145–156.
Li, W. (2009). “Method of Water Quality Remote Sensing and Its Application.” Energy and Environment 5 (5): 62– 64. Wu, G. 2015. “Retrieval of Suspended Sediment Concentration in the Yangtze Estuary and Its Spatiotemporal Dynamics Analysis Based on GOCI Image Data.” Master Thesis, Chang’an University.
Lim J and Choi M (2015) Assessment of water quality based on Landsat 8 operational land imager associated with human activities in Korea Environ. Monit. Assess. 187 1–17.
Lin, S.; Novitski, L.N.; Qi, J.; Stevenson, R.J. (2018). Landsat TM/ETM+ and machine-learning algorithms for limnological studies and algal bloom management of inland lakes. J. Appl. Remote Sens12, 1–17.
Liu H, Li Q, Bai Y, Yang C, Wang J, Zhou Q, Hu S, Shi T, Liao X and Wu G 2021 Improving satellite retrieval of oceanic particulate organic carbon concentrations using machine learning methods Remote Sens. Environ. 256 112316.
Liu X, Lee Z, Zhang Y, Lin J, Shi K, Zhou Y, Qin B and Sun Z (2019). Remote sensing of Secchi depth in highly turbid lake waters and its application with MERIS data. Remote Sensing 11(19): 2226.
Liu, H.; Zhou, Q.; Li, Q.; Hu, S.; Shi, T.;Wu, G. (2019). Determining switching threshold for NIR-SWIR combined atmospheric cor-rection algorithm of ocean color remote sensing. ISPRS J. Photogramm., 153, 59–73.
Martinez E, Brini A, Gorgues T, Drumetz L, Roussillon J, Tandeo P, Maze G and Fablet R (2020). Neural network approaches to reconstruct phytoplankton time-series in the global ocean Remote Sens. 12 1–13.
Matthews, M.W.; Bernard, S.; Winter, K. Remote sensing of cyanobacteria-dominant algal blooms and water quality parameters in Zeekoevlei, a small hypertrophic lake, using MERIS. Remote Sens. Environ. 2010, 114, 2070–2087.
Menken, K.D.; Brezonik, P.L.; Bauer, M.E. (2006). Influence of Chlorophyll and Colored Dissolved Organic Matter (CDOM) on Lake Reflectance Spectra: Implications for Measuring Lake Properties by Remote Sensing. Lake Reserv Manag, 22, 179–190.
Morel and Prieur (1977).Bio-optical Models. In Encyclopedia of Ocean Sciences, 1st ed.; Elsevier Ltd.: Amsterdam, The Netherlands, 2001; pp. 385–394.
Myer M H, Urquhart E, Schaeffer B A and Johnston J M (2020) Spatio-temporal modeling for forecasting high-risk freshwater cyanobacterial harmful algal blooms in Florida Front. Environ. Sci. 8 1–13.
Odermatt, D.; Gitelson, A.; Brando, V.E.; Schaepman, M. (2008). Review of Constituent Retrieval in Optically Deep and Complex Waters from Satellite Imagery. Remote Sens Environ, 118, 116–126.
Ouma, Y.O. Waga, J.; Okech, M.; Lavisa, O.; Mbuthia, D. (2018). Estimation of Reservoir Bio-OpticalWater Quality Parameters Using Smartphone Sensor Apps and Landsat ETM+: Review and Comparative Experimental Results. J. Sens., 2018, 1–32.
Ouma, Y.O.; Waga, J.; Okech, M.; Lavisa, O.; Mbuthia, D. (2018). Estimation of Reservoir Bio-Optical Water Quality Parameters Using Smartphone Sensor Apps and Landsat ETM+: Review and Comparative Experimental Results. J. Sensors, 2018, 1–32.
Park J, Kim J H, Kim H C, Kim B K, Bae D, Jo Y H, Jo N and Lee S H (2019). Reconstruction of ocean color data using machine learning techniques in polar regions: Focusing on off Cape Hallett, Ross Sea Remote Sens. 11 1366.
Park J, Kim J H, Kim H C, Kim B K, Bae D, Jo Y H, Jo N and Lee S H (2019). Reconstruction of ocean color data using machine learning techniques in polar regions: Focusing on off Cape Hallett, Ross Sea Remote Sens. 11 1366.
Peterson K T, Sagan V and Sloan J J 2020 Deep learning-based water quality estimation and anomaly detection using Landsat-8/Sentinel-2 virtual constellation and cloud computing GIScience Remote Sens. 57 510–25.
Politi, E., Cutler, M.E.J., Rowan, J.S. (2015). Evaluating the spatial transferability and temporal repeatability of remote-sensing-based lake water quality retrieval algorithms at the European scale: A meta-analysis approach. Int. J. Remote Sens., 36, 2995–3023.
Powell, R.; Brooks, C.; French, N. (2008). Shuchman, R. Remote Sensing of Lake Clarity; Michigan Tech Research Institute: Ann Arbor, MI, USA.
Pyo, J.; Duan, H.; Baek, S.; Kim, M.S.; Jeon, T.; Kwon, Y.S.; Lee, H.; Cho, K.H. (2019). A convolutional neural network regression for quantifying cyanobacteria using hyperspectral imagery. Remote Sens. Environ, 233, 111350.
Quang NH, Sasaki J, Higa H and Huan NH (2017). Spatiotemporal variation of turbidity based on Landsat 8 OLI in Cam Ranh Bay and Thuy Trieu Lagoon, Vietnam. Water 9(8): 570.
Ruescas A B, Mateo-García G, Camps-Valls G and Hieronymi M (2018) Retrieval of case 2 water quality parameters with machine learning Int. Geosci. Remote Sens. Symp. 124–7.
Rundquist, D.C.; Han, L.; Schalles, J.F.; Peake, J.S. (1996). Remote Measurement of Algal Chlorophyll in Surface Waters: The Case for the First Derivative of Reflectance Near 690 nm. Photogramm. Eng. Remote Sens., 62, 195–200.
Sabat-Tomala A, Jaroci´nska AM, Zagajewski B, Magnuszewski AS, Sławik ŁM, Ochtyra A, Raczko E and Lechnio JR (2018). Application of HySpex hyperspectral images for verification of a two-dimensional hydrodynamic model. European Journal of Remote Sensing 51(1): 637–649.
Shang, W.; Jin, S.; He, Y. (2023). Spatial—Temporal Variations of Total Nitrogen and Phosphorus in Poyang, Dongting and Taihu Lakes from Landsat-8 Data. Water, 13, 1704.
Song, K.; Li, L.; Tedesco, L.P.; Li, S.; Duan, H.; Liu, D.; Hall, B.E.; Du, J.; Li, Z.; Shi, K. (2011). Remote Estimation of Chlorophyll-a in Turbid Inland Waters: Three-Band Model versus GA-PLS Model. Remote Sens Environ, 136, 342–357.
Song, K.; Wang, Z.; Blackwell, J.; Zhang, B.; Li, F.; Zhang, Y.; Jiang, G. (2011). Water Quality Monitoring Using Landsat Themate Mapper Data with Empirical Algorithms in Chagan Lake, China. J. Appl. Remote Sens, 5, 5.
Sunder S, Raaj R and Ramakrishnan B (2017). ANN based estimation of daily sea surface temperature over Arabian sea using MODIS data 38th Asian Conf. Remote Sens. - Sp. Appl. Touching Hum. Lives, ACRS 2017.
Svircev, Z.; Simeunovi´c, J.; Subakov-Simi´c, G.; Krsti´c, S.; Panteli´c, D.; Duli´c, T. Cyanobacterial blooms and their toxicity in Vojvodina Lakes, Serbia. Int. J. Environ. Res. 2013, 7, 745–758.
Tang S, Dong Q, Chen C, Liu F and Jin G (2019). Retrieval of suspended sediment concentration in the Pearl River estuary from MERIS using support vector machines Int. Geosci. Remote Sens. Symp. 3 239–42.
Thiemann, S.; Kaufmann, H. (2002).Lake water quality monitoring using hyperspectral airborne data—A semiempirical multi sensor and multitemporal approach for the Mecklenburg Lake District, Germany. Remote Sens. Environ., 81, 228–237.
Toming, K.; Kutser, T.; Laas, A.; Sepp, M.; Paavel, B.; Nõges, T. (2016). First experiences in mapping lakewater quality parameters with sentinel-2 MSI imagery. Remote Sens., 8, 640.
Torbick, N.; Hession, S.; Hagen, S.; Wiangwang, N.; Becker, B.; Qi, J. (2013). Mapping inland lake water quality across the Lower Peninsula of Michigan using Landsat TM imagery. Int. J. Remote Sens., 34, 7607–7624.
Uudeberg K, Aavaste A, K˜oks K, Ansper A, Uus˜oue M, Kangro K, Ansko I, Ligi M, Toming K and Reinart A (2020). Optical water type guided approach to estimate optical water quality parameters. Remote Sensing 12(6): 931.
Vakili, T.; Amanollahi, J. (2019). Determination of optically inactive water quality variables using Landsat 8 data: A case study in Geshlagh reservoir affected by agricultural land use. J. Clean. Prod., 247, 119134.
Vundo A, Matsushita B, Jiang D, Gondwe M, Hamzah R, Setiawan F and Fukushima T (2019). An overall evaluation of water transparency in Lake Malawi from MERIS data. Remote Sensing 11(3): 279.
Wang, S.; Garcia, M.; Bauer-Gottwein, P.; Jakobsen, J.; Zarco-Tejada, P.J.; Bandini, F.; Paz, V.S.; Ibrom, A. (2019). High spatial res-olution monitoring land surface energy, water and CO2 fluxes from an Unmanned Aerial System. Remote Sens. Environ., 229, 14–31.
Wattelez G, Dupouy C, Mangeas M, Lefèvre J, Touraivane and Frouin R (2016). A statistical algorithm for estimating chlorophyll concentration in the New Caledonian lagoon Remote Sens. 8 1–23.
Wouthuyzen S, Kusmanto E, Fadli M, Harsono G, Salamena G, Lekalette J and Syahailatua A (2020). Ocean color as a proxy to predict sea surface salinity in the Banda Sea IOP Conf. Ser. Earth Environ. Sci. 618 012037.
Wu, C.; Wu, J.; Qi, J.; Zhang, L.; Huang, H.; Lou, L.; Chen, Y. (2010). Empirical estimation of total phosphorus concentration in the mainstream of the Qiantang River in China using Landsat TM data. Int. J. Remote Sens., 31, 2309–2324.
Xiang, R.; Wang, L.; Li, H.; Tian, Z.; Zheng, B. (2021). Water quality variation in tributaries of the Three Gorges Reservoir from 2000 to 2015. Water Res., 195, 116993.
Yang, H.; Kong, J.; Hu, H.; Du, Y.; Gao, M.; Chen, F. (2022). A Review of Remote Sensing forWater Quality Retrieval: Progress and Challenges. Remote Sens., 14, 1770.
Yniguez A T and Ottong Z J (2020). Predicting fish kills and toxic blooms in an intensive mariculture site in the Philippines using a machine learning model Sci. Total Environ. 707 136173.
Yu B, Xu L, Peng J, Hu Z and Wong A (2020). Global chlorophyll-a concentration estimation from moderate resolution imaging spectroradiometer using convolutional neural networks J. Appl. Remote Sens. 14 034520.
Zhang L, Zhang R and He Q (2020). Sea surface salinity retrieval from aquarius in the south china sea using machine learning algorithm Int. Geosci. Remote Sens. Symp. 5643–6.
Zhang Y, Liu X, Qin B, Shi K, Deng J and Zhou Y (2016). Aquatic vegetation in response to increased eutrophication and degraded light climate in Eastern Lake Taihu: Implications for lake ecological restoration. Scientific reports 6(1): 1–12.
Zhou, B., Shang, M., Wang, G., Zhang, S., Feng, L., Liu, X., Wu, L., Shan, K. (2018). Distinguishing two phenotypes of blooms using the normalised di_erence peak-valley index (NDPI) and Cyano-Chlorophyta index (CCI). Sci. Total Environ, 628, 848–857.
Copyright (c) 2024 Abebe Tesfaye
This work is licensed under a Creative Commons Attribution 4.0 International License.
