Empowering water utilities: Crafting an end-user-friendly reliability ranking for evaluating satellite remote sensing advances through literature insights

Water is a fundamental resource closely linked to economic development, environmental sustainability, and social well-being. Ensuring access to an adequate quantity and quality of water is essential for sustaining various aspects of our lives, including food production, safe drinking water, sanitation services, and overall economic growth. However, the availability and quality of water are increasingly threatened by a range of factors such as climate change and extreme climatic events, resource depletion, land use changes, population growth, and natural disasters. These challenges pose significant obstacles to achieving long-term and sustainable access to sufficient quantities of high-quality water in many regions across the globe (Zamyadi 2014; Food & Agriculture Organization 2020; World Bank 2022; United Nations Environment Programme 2023).

Satellite remote sensing, also known as Earth observation (EO), has emerged as a promising solution to meet the growing demand for comprehensive data and information needed for effective water management (Malthus et al. 2019; Agnoli et al. 2023; Dube et al. 2023). Most remote sensing methods operate on the basis of the measurement of a latent property of an earth surface feature based on its interaction with electromagnetic radiation. The most useful forms of electromagnetic radiation in satellite detection cover the optical (400–2,500 nm), thermal (3–20 μm), and microwave (1–30 cm) spectral regions. Sources of electromagnetic radiation include the sun (so-called ‘passive’ approach) or may be actively emitted by the sensor itself (‘active’ approach). Interactions of the radiation with earth's surface features may include reflection, absorption, scattering, and transmission and by measuring these interactions satellite sensors can gather information about a wide range of physical properties of those surfaces. For example, satellite altimeters use sensor-generated microwaves to measure the height of the earth's surface, relative to a geoid reference, via the time delay between emission and reflection back from the earth's surface. Whilst mostly used in oceanographic applications for sea level and circulation studies (Nelli et al. 2023), the method has also been used to determine river stage height and hence flow in large remote rivers (Liu et al. 2018). Satellite-based remote sensing approaches with the most utility for water-related studies include:

• Passive optical – These systems operate in the visible and short-wave infrared wavelength regions of the spectrum and measure reflected and emitted radiation off surface features naturally illuminated with light from the sun. Optical approaches are used to measure many properties of land and water surfaces and rely on the concept of the spectral signature – that different types of objects on the earth's surface reflect and absorb radiation in different wavelengths and thus have different spectral reflectance signatures. Thus, given sufficient spectral detail, the (bio-)physical properties of different earth surface features can be discriminated and quantified (for example, vegetation properties and water quality; Hunter et al. 2008; Botha et al. 2013; Zolfaghari et al. 2022). The strength and the spectral characteristics of the signal measured by the optical sensors depend on the characteristics of the object(s) reflecting the light and the properties of the atmosphere through which the radiation must pass, including cloud cover. Well-known sensors in this category include the Landsat series of satellites and the more recent Sentinel 2 and 3 series of sensors.

• Passive thermal – Objects with a temperature above absolute zero emit thermal radiation. Thermal remote sensing approaches are based on the principle that the radiant energy detected from an object is proportional to its internal temperature and emissive properties (Elachi & Van Zyl 2021). The temperature of an object might be affected by its colour, roughness, and moisture content. This method is used for land and water temperature measurement and active fire detection during the day or at night. The method also provides the basis for determining evapotranspiration and the passive determination of soil moisture. Thermal remote sensing is also undertaken in areas of low atmospheric influence and is affected by the presence of clouds. Thermal sensors include the Advanced Very High Resolution Radiometer, the Moderate Resolution Imaging Spectroradiometer (MODIS), and Landsat.

• Passive microwave – Radiation that is thermally emitted from earth surface objects can also be re-emitted at longer wavelengths in the microwave region of the electromagnetic spectrum, albeit at lower energy levels. Passive microwave radiometers measure the intensity of this radiation at different frequencies. The intensity of emitted microwave radiation is influenced by the temperature and moisture content of different surface features and may be used to map surface soil moisture, vegetation water content, and properties of snow and ice (Weng 2017). Passive microwave remote sensing is particularly useful for monitoring changes in the Earth's system over time, as the measurements are not affected by cloud cover or daylight. Sensors used for passive microwaves include CIMR and SSMI/S.

• Active microwave – Active microwave instruments emit their own illuminating microwave radiation at fixed long wavelengths and measure the amount of energy that is backscattered from the earth's surface. They perform these measurements by a method known as ‘synthetic aperture radar’, or SAR, where images are very much constructed from information on the phase, amplitude, intensity, and polarisation of the returned wavelengths (Elachi & Van Zyl 2021). The amount of reflected energy depends on the wavelength used and the properties of the surface, such as its roughness, moisture content, and composition. Thus, the physical properties of the earth's surface features are measured via interactions with its structural properties and moisture content; SAR can provide information properties relevant to this study, such as topography, soil moisture, and vegetation (Gupta et al. 2022). Furthermore, the method is largely unaffected by atmospheric conditions, including clouds, and can be undertaken at night. The commonly used active microwave sensors include Sentinel-1 and RadarSAT.

• Satellite altimetry – Satellite altimeters use sensor-generated microwaves to measure the height of the earth's surface, relative to a geoid reference, via the time delay between emission and reflection back from the earth's surface. Whilst mostly used in oceanographic applications for sea level and circulation studies, the method has also been used to determine river stage height and hence flow in large remote rivers (e.g. Liu et al. 2018). Important existing satellite altimeters include Jason-3 and Sentinel-6 POS4.

Other methods providing some utility to water-related issues, but still under considerable technological development, include light detection and ranging (LiDAR) and satellite gravimetric missions. LiDAR is an active method relying on the use of a laser pulse (e.g. at 900 nm) to determine surface height after the pulse is backscattered from the terrain. Using short bursts and a high burst rate ‘point clouds’ of returned pulses of individual points in space can be used to derive images of earth surface height, vegetation properties, and bathymetry. Whilst well developed for airborne approaches, which may be expensive to deploy, there are few examples of satellite-based LiDAR instruments. Satellite gravimetric missions measure the precise distance between two satellites flown in tandem, allowing the determination of the gravitation field below. In water-related studies, differences in the density and mass of water on or beneath the earth's surface may impact the gravitational field, and thus satellite gravimetry has found utility in groundwater monitoring, determinations of terrestrial water storage, and drought monitoring (Tian et al. 2017).

Over the past decade, the field of satellite EO technologies has seen remarkable advancements, fundamentally transforming the capabilities available for water utilities and agencies (Malthus et al. 2019; Derkani et al. 2021; Alberello et al. 2022; Agnoli et al. 2023; Dube et al. 2023; Nelli et al. 2023). These developments have been characterized by several key trends. First, satellite capabilities have significantly improved, with modern satellites offering higher resolution imagery, increased frequency of observations, and the ability to monitor a broader spectrum of environmental variables. This has made it possible to track changes in water bodies with unprecedented detail and timeliness, which is crucial for managing water resources effectively. In addition to technical advancements, there has been a notable increase in data accessibility (Malthus et al. 2019; Agnoli et al. 2023; Dube et al. 2023; Nelli et al. 2023; Rubbens et al. 2023). Enhanced data access have been facilitated by initiatives from both governmental and international organizations, which have started to provide open access to satellite data. This democratization of data has enabled a wider range of stakeholders to utilize satellite observations for water management purposes.

Furthermore, the involvement of the private sector has introduced new dynamics into the EO ecosystem. Private companies have launched their own satellites, offering specialized services that complement public satellite missions. This has expanded the range of available data and services, allowing for more tailored solutions to specific water management challenges. Alongside these developments, there have been significant advancements in data processing and analytics (Nelli et al. 2023; Rubbens et al. 2023). Machine learning (ML) and artificial intelligence (AI) technologies have been increasingly applied to satellite data, enabling the extraction of more nuanced insights and the prediction of future trends with greater accuracy. However, despite these considerable advancements, there remains a gap in the form of a user-friendly reliability ranking tool for water sector practitioners and utility operators. Such a tool would be invaluable for evaluating the numerous satellite remote sensing options, helping water utilities and agencies make informed operational decisions. This gap highlights an ongoing challenge in effectively bridging the sophisticated capabilities of satellite technologies with the practical needs of end users for operational decision-making with utilities.

The aims of this paper are to (1) provide a state-of-the-art summary of recent operational trends and literature in satellite remote sensing to evaluate the existing capabilities and operationally available tools, and (2) produce an end-user-friendly reliability ranking for evaluating satellite remote sensing advances through literature insights, as they may impact the functions of water utilities and agencies. The study will focus on selected near-to-operational areas of application where remote sensing may impact water quantity and water quality across a range of scales from water bodies to catchments. To the best of the author's knowledge, this paper provides a novel assessment of satellite optical sensors's potential implementation in water-related applications using a new and end-user-friendly reliability ranking process.

This review adheres to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to ensure a comprehensive and transparent evaluation of existing literature on the application of remote sensing technologies in water utility operations. A systematic search was conducted across multiple electronic databases, including Web of Science, Scopus, IEEE Xplore, and Google Scholar, to identify relevant published studies (81% published since 2016). The search strategy combined terms related to ‘remote sensing’ with those associated with ‘water utility operations’. Boolean operators (AND, OR) were used to combine search terms, and filters were applied to exclude non-English articles and those outside the scope of water utilities.

Studies were included if they (1) focused on the application of remote sensing technologies within water utility operations, (2) presented original research findings, and (3) were peer-reviewed articles or conference papers. Exclusion criteria encompassed studies that (1) did not directly relate to water utilities, (2) were review articles or editorial pieces, and (3) lacked empirical data or clear methodology. The authors screened titles, abstracts, and full texts of the identified studies for eligibility. Discrepancies were resolved through discussion. For each included study, data were extracted on the author(s), year of publication, study location, remote sensing technology used, application area within water utilities, main findings, and limitations. The quality of the included studies was assessed using a standardized checklist adapted from the Critical Appraisal Skills Programme, focusing on the clarity of objectives, appropriateness of the methodology, and reliability of the findings.

Given the expected heterogeneity in study designs and outcomes, a narrative synthesis approach was employed. The analysis was structured around key themes identified during the data extraction phase, such as types of remote sensing technologies, application areas (including catchment monitoring, water demand estimation, flood monitoring, water quality monitoring, farm dam monitoring, urbanization trends, drought forecasting, fire spotting, and post-fire water quality impacts), benefits, challenges, and future directions. Trends, patterns, and gaps in the literature were highlighted, and the implications of findings for practice and research were discussed. The study will not address other issues such as greenhouse gas detection, emergency response, and scenario planning, nor an in-depth treatment of other technologies such as ground sensors and drones.

To achieve routine, long-term, and continuous measurements to provide the greatest achievable benefit for a range of envisaged applications, satellite sensor design requires an exercise in trade-offs. Critical trade-off parameters include platform orbit and altitude, the number of spectral bands, ground sampling distance (GSD) (spatial resolution), swath width, data volume, communication bandwidth, satellite size, signal-to-noise and resolution, and cost (Donlon et al. 2012). Where satellites are designed to maximize the benefit to a broad range of applications, careful consideration of a satellite's characteristics, advantages, and limitations for a specific application will be required. The assessment of particular satellite approaches for a particular application is often undertaken using four key characteristics of satellite sensors:

• Spectral resolution: the portion of the electromagnetic spectrum being measured by the sensor, often determined by the number of spectral bands, their locations, and their bandwidths. Generally, higher spectral resolution is achieved with narrower bandwidths and a greater number of spectral bands.

• Spatial resolution: The GSD, which is commonly expressed as the size of a single pixel in an image. A smaller GSD implies higher spatial resolution.

• Temporal resolution: Often given as the repeat coverage (or revisit) time.

• Radiometric resolution: The dynamic range of radiances measured and the ability of the sensor to measure changes in radiance, determined by the number of bits used across the dynamic range of radiances.

Other factors will include the ability to combine data from different wavelengths (e.g. optical, thermal, and radar), satellite swath width (the strip of earth sampled with each orbit), level of calibration, data availability, and cost. To ensure data collected by a satellite is accurate over the sensor's lifetime, calibration is often performed. Calibration involves ongoing comparison to a known standard (e.g. an onboard calibrator, the moon, or a homogeneous surface on the Earth). Earth-based corner reflectors are used in the case of SAR instruments. These calibration measurements are used to track sensor drift (radiometric, spectral, spatial, and geometric) to ultimately ensure the continuity of an accurate data time series. Validation methods are similarly used to measure the uncertainty of the estimation of earth surface parameters at a higher level in the data processing chain.

As outlined above, the relative utility of a particular satellite sensor for a particular application, such as aiding water utility/agency operations, will depend on the performance characteristics of the sensor itself and of the platform upon which it is carried (Ustin & Middleton 2021). Platform-dependent characteristics include orbit type (e.g. polar or geostationary), orbit altitude, overpass repeat interval, and potential manoeuvrability. Sensor-specific characteristics include viewing mode design, pixel size, swath width, and spectral and radiometric resolutions.

Table 1 provides an overview of the principal characteristics of currently available satellite optical sensors on a range of key attributes, Table 2 provides an assessment of their potential application to the water-related applications under consideration in this study, and the Supplementary Information section provides an example of a decision tool for applying satellite remote sensing in water management. The assessment coefficients (which are 1: highly suitable; 2: suitable; 3: potentially suitable; 4: not suitable) are obtained by combining the data obtained from existing sensors for water-related applications (as shown in García et al. 2016). For each water-related application, the relevant satellite sensor variables are classified according to their potential usefulness. Data are sourced from the Observing Systems Capability Analysis and Review Tooldatabase. The table is also colour-coded according to the reliability of the assessment (green: high confidence, yellow: medium confidence, red: low confidence). Multiple satellite variables have been grouped together to show the suitability of the sensors for more general applications (as shown in the header of the table). A similar procedure is repeated in Tables 3 and 4 for active microwave sensors and Tables 5 and 6 for hyperspectral sensors.

Table 2

Assessment of satellite optical sensors potential application to water-related implementations (1: highly suitable; 2: suitable; 3: potentially suitable; 4: not suitable)

Note. Colour-coded according to the reliability of the assessment (green: high confidence, yellow: medium confidence, red: low confidence).

View Large

Table 4

Assessment of satellite microwave sensors potential application to water-related implementations (1: highly suitable; 2: suitable; 3: potentially suitable; 4: not suitable)

Note. Colour-coded according to the reliability of the assessment (green: high confidence, yellow: medium confidence, red: low confidence).

View Large

Table 6

Assessment of satellite hyperspectral sensors potential application to water-related implementations (1: highly suitable; 2: suitable; 3: potentially suitable; 4: not suitable)

Note. Colour-coded according to the reliability of the assessment (green: high confidence, yellow: medium confidence, red: low confidence)

View Large

This paper focuses primarily on satellite optical and active SAR sensors as providing the most suitable data at finer spatial resolution. However, other sensor types are covered where appropriate in specific applications. Most of the sensors covered are constructed and launched by publicly funded international Space Agencies. For many of these programmes, the data are typically provided under free and open access policies, which have encouraged the uptake of use of such data. Of note amongst these are data acquired by the United States Geological Survey (USGS)/National Aeronautics and Space Administration (NASA) Landsat series of sensors and the European Space Agency's Sentinel series of satellites under the EU Copernicus programme.

Both SAR and common optical satellites are launched into near-polar low-earth orbits, which allows for image acquisition at the same local time for each overpass. Spatial resolution and swath of the image acquired mainly dictate the revisit frequency – the time taken to return to reacquire the same point on the Earth's surface. Most optical satellite sensors are multispectral imagers, measuring from 4 to 20 or so spectral bands (Ustin & Middleton 2021). Two recent developments in optical remote sensing include the increasing availability of data from satellites developed and launched by the for-profit sector, particularly characterised by the ‘niche’ of sensors offering high spatial resolution and high temporal frequency achieved via satellite constellation clusters. Two key companies here include Planet and Maxar. On the other hand, trial hyperspectral satellites were recently launched, which are capable of measuring in a greater number of optical bands. These provide more detailed information about the reflectance properties of earth surface features which in turn potentially allow superior discrimination of features and bio-physical variables. Until now hyperspectral satellites have been limited by limited swath width and low temporal coverage.

An important benefit of satellite remote sensing is the availability of repeatable and objective observations of the earth's surface. When considering sensors for operational use, an important issue is thus the assured continuity of the time series such that operational application may be sustained over the longer term. For this reason, the polar-orbiting multispectral high-resolution Landsat programme (with its 30-year time series and continued operations until 2030 and beyond) and the complementary Sentinel-2 programme (2016–2035 and beyond) will provide the background of observations with the widest utility to earth monitoring applications in the water domain, covering the monitoring of vegetation, land use, soil, and water.

In the past, significant barriers hindered the adoption of EO methods in water-related studies. These barriers included different data sources, the cost of computing infrastructure, tools, and time required for the high level of data processing, as well as the need for specialist personnel for both pre-processing and analysis. Additionally, a lack of data interoperability (Mahecha et al. 2020) exacerbated these challenges, presenting major obstacles for organizations seeking to integrate satellite EO into their business practices and thus slowing the adoption of such methods.

As mentioned in the previous section, EO in the modern era can be characterised by a proliferation of satellite data of different sensor types, leading to rapidly increasing volumes of data and, as time goes on, datasets of increasingly longer time series upon which to analyse trends and changes. For small organisations in particular, this increased access to satellite data challenges storage capacity and processing power to integrate and analyse data. The challenge over recent years has thus not so much been the lack of EO data but its transformation from raw form into meaningful information (Coetzee et al. 2020). Surmounting this hurdle requires new solutions, new means of integration and new analytical methods (Ustin & Middleton 2021).

In recent years, a number of key developments have seen the barriers to entry into EO significantly lowered (Table 7). The implications of these trends are such that the integration of remote sensing data into operational workflows has become both simpler and more straightforward. This has seen the user base of remotely sensed data significantly broadened and greater adoption by non-expert user groups (Sudmanns et al. 2020).

Whilst these developments are already changing the paradigm for the uptake and integration of EO approaches into the workflows of companies, key challenges still remain. These challenges may include the long-term viability of cloud services and continuity of service, the frequency of disruptions, and downtime and, of particular concern to business models, questions of data and algorithm security, legal considerations over data ownership, and the reproducibility/trustworthiness of analyses based on internal cloud algorithms (Gomes et al. 2020). Not all the data required may be located on one cloud environment, raising questions about the interoperability and portability of data and methods across different data cubes/cloud environments (Sudmanns et al. 2020). There are additional implications in training and education, including the need for the development of training materials and the need for user-friendly solutions that provide high-level analytical tools (Gomes et al. 2020). Standards still require tighter definitions, including a commonly agreed definition of analysis ready data (ARD), web-based processing, the application of algorithms across multiple data cube instances, time-series analysis, uncertainty and quality assessment (Di & Ramapriyan 2010). Further development of techniques in visualisation and communication techniques and model-data integration is still needed.

Nevertheless, in comparisons of different capabilities, the open data cube concept outperformed other approaches (Gomes et al. 2020), presents an easier means to integrate data from multiple imaging sensors (Ustin & Middleton 2021), and serves as a tool to both implement and execute complex workflows across multiple variables and spatial and temporal scales (Mahecha et al. 2020).

Shortcomings include cost, sparse coverage, and delays in information availability. Although they provide data at fixed points, sensor networks offer the acquisition of parameters at higher temporal resolution than either sampling or satellites provide. Profiling systems may also provide information at different levels in the water column to determine vertical stratification in water quality parameters and to assist in the assessment of 3D structure as determined from models. Ground sensing includes sensors deployed in situ (Valente et al. 2011; Pattnaik et al. 2020; Balivada et al. 2022) as well as above-surface optical sensors measuring high spectral resolution reflectance (Hommersom et al. 2012; Brando et al. 2016; Bowling et al. 2018). Key issues associated with all sensors are related to maintenance, biofouling, and calibration (Zamyadi et al. 2016; Cremella et al. 2018; Bertone et al. 2019, p. 244; Choo et al. 2019). Drawbacks of in situ sensor systems include the lack of spatial coverage and associated with this, the expense to both establish and maintain such networks. Other issues may include the range of water quality parameters measured and the openness of the data they deliver upon, which is the basis for decision-making.

While some current commercial off-the-shelf sensors are expensive, lower-cost versions will allow the creation of networks of field-deployed devices for more widespread autonomous sensor networks based on the Internet of Things (IoT). IoT-based networks employ wireless communications across mixed modes (fixed aerial, LoRa, or satellite networks) to both collect sensor data and provide remote control. However, before this can be viably realised progress will need to be made in sensors that are low-cost yet deliver adequate sensitivity, precision, and reliability and meet a functional range to track abnormal water quality events (Coffer et al. 2023). Combined with edge computing, where data processing (and potentially decision-making) is undertaken on the device or ground station close to the location of the data generation to address bandwidth and latency (Wei & Cao 2019; Li et al. 2021; Ren et al. 2022), satellite-based IoT will enable a wide area connectivity and hence a widely dispersed network across without the restrictions in fixed terrestrial communications (Fraire et al. 2019; Centenaro et al. 2021). Issues of energy management and data security will also need to be addressed.

Unoccupied aerial systems (UAS), or drones, offer a means to support and validate satellite surveys at a relatively low cost and have rapidly become an integral tool for local-scale environmental surveys (Gray et al. 2018; Vélez-Nicolás et al. 2021) and a means to lower health and safety risks to field staff. Given their low flying height, they may offer acquisition at a finer resolution than may be possible using satellites and offer rapid response, for instance, during algal bloom events, after storms, or other disturbances. Through the collection of a far greater number of field points than possible with in situ sampling, UAS offers a means of satellite data validation where the acquisition may be closer in time to the satellite data, at higher resolution, and of directly comparable products (reflectance and water colour) and at more appropriate spatial and temporal scales (e.g. Wu et al. 2019).

The choice of the UAS platform and associated technology will depend on budget, sensor selection, and prevailing regulations. Key platform considerations will include size, cost, carrying capacity, operating height, stability, and flight range/duration. The choice of lightweight, miniature sensors for UAS is rapidly expanding: optical sensors (visible to near-infrared) for UAS are the most commonly available and may consist of non-imaging hyperspectral radiometers (Becker et al. 2009; Shang et al. 2017) or imaging spectrometers (red, green and blue (RGC) colors still frame (Arango & Nairn 2019; Cheng et al. 2020; Cillero Castro et al. 2020; Kwon et al. 2020; Rossiter et al. 2020) or RGB video, multispectral, hyperspectral). Water studies will require a sensor with a good signal-to-noise (O'hea & Laney 2020). Other sensors available include thermal and LiDAR. Choice of camera settings (including ground sampling distance, lens, exposure, stabilization, and image quality) is critical, and there is an expanding availability of good UAS control software to help plan flights and automate acquisition (including set pre-programmed flight paths, geopositioning, and degree of overlap).

Regulations impose operational limitations through altitudinal restrictions, visual line-of-sight and regulation of airspaces, and requirements in terms of permissions, licenses, and authorisation. The quality of UAS outputs will be affected by multiple interacting factors. A good UAS survey requires good preparation and planning with attention to good experimental design, desired spectral and spatial characteristics, standardised georeferencing, sensor calibration, radiometric and atmospheric correction, and error reporting (Arroyo-Mora et al. 2019; Tmušić et al. 2020). A number of useful packages exist to assist in the post-processing of unmanned aerial vehicle (UAV) data. Common tools include the ability to generate 2D and 3D models, orthophoto mosaics, point clouds, and terrain analysis. UAV sensor calibration is typically undertaken pre-flight in the laboratory and/or through the use of spectral targets placed within the acquired scene (Aasen et al. 2018; Zarzar et al. 2020). Similarly, quality will also be affected by environmental conditions: sun angle and solar illumination, wind speed and waves, cloudiness, and surface glare (Arroyo-Mora et al. 2019; Ehmann et al. 2019). The uncertainty associated with end products can be high if insufficient attention to planning and acquisition is given.

Despite being a maturing technology, UAS may fill roles in many of the applications considered as part of this report, such as catchment (e.g. land use and topography) and river reach studies, hydrology (water level and storage and flow), flooding, temperature mapping, bathymetry, benthic and macrophytic vegetation, contaminants, and water quality, including algal blooms and turbidity (Ehmann et al. 2019; Wu et al. 2019; Kislik et al. 2020; Bolch et al. 2021; Vélez-Nicolás et al. 2021). The use of UAS for water sampling, through direct samples (Ore et al. 2015; Wu et al. 2019) or adaptive water measurement with sensors (Koparan et al. 2020) has also been explored. In addition to the limitations and challenges covered above, the lack of standardised guidance/protocols for data collection and interpretation is a barrier to the adoption of best practices, particularly for those new to the technology (Tmušić et al. 2020). Before embarking on a UAS-based surveillance programme, a critical assessment should evaluate benefits versus limitations and cost, required duration, considerations of system complexity, technical challenges, maintenance and logistics and safety/risks (McKee 2017; Coffer et al. 2023).

Target application areas and operational uses are categorised in Table 8. Development, notably in the extent of impervious surfaces, significantly alters the hydrology of urban areas through changes to river channels, surface runoff, and the occurrence and extent of floods, groundwater recharge, as well as impacts on water quality and biodiversity (Wentz et al. 2014). Satellite remote sensing has shown the ability to improve hydrologic modelling over in situ measurements with studies focusing on watershed scale impacts, drinking water sources, nutrient loadings, and water quality trends (Kucukmehmetoglu & Geymen 2008; Ebrahimian et al. 2016; Zhu et al. 2019). SAR and LiDAR data have shown value in near real-time mapping of flood dynamics and other terrain-related hydrologic characteristics, related to disaster response and hazard management. Conley et al. (2022) combined hydrological and Landsat-derived Normalized Difference Vegetation Index (NDVI) data from 372 urbanized watersheds in the USA over a 34-year time period to show that urban greenness had a significant impact on downstream flow responses (as total flows, base, peak, and high flow frequencies and flashiness). The study highlighted the co-benefits and value of green stormwater infrastructure development for mitigating urban runoff impacts.

The importance of water as a crucial resource for our economic, environmental, and social well-being requires water agencies to adapt and adopt innovative approaches for effective water management. In this study, we have shown that EO is a fundamentally physically based method, utilising a variety of techniques (optical, thermal, microwave), each offering different perspectives to measure a range of meteorological and biogeophysical properties including quantification of the extent, cover, and amount of surface features relevant to a wide range of application areas of interest to water utilities and agencies. This information is provided with wide spatial coverage at a high rate of repeatability and with a variety of data sources across different spatial and temporal extents and resolutions. Clouds and associated shadows provide a level of interference with optical and thermal methods, potentially limiting temporal resolution, but the availability of harmonised datasets from satellite constellations provides the means to limit their impact.

EOs do not deliver information on all of the desired variables that may be required to understand and manage water resources, as direct measurement of all possible required parameters is not possible. For example, existing guidelines for the monitoring of algal bloom status in water bodies require the monitoring of cell numbers. Similarly, certain key chemical parameters, such as nutrients, have no direct impact on a remotely sensed signal. However, in both these cases related remotely sensed measurements (e.g. chlorophyll concentration) may be used either as an alternative to algal monitoring or as an indirect, or proxy, measurement of the effect of nutrient concentrations (Dekker et al. 2018).

Field observations are often more direct in nature than EO, providing measurements of variables with fewer uncertainties and assumptions, albeit with limited spatial coverage and greater labour intensity (Guerschman et al. 2016). Field observations require extrapolation to fill in the gaps between observations, which thus increases uncertainty. Operational EOs provide a significant contribution to any monitoring effort in providing a sustained temporal and spatial ‘sampling’ capability to measure landscape and water dynamics over a range of scales (seasonal, annual; across a water body or a catchment, or a state) at resolutions of 10–100 s of metres; when compared to the footprints of ground-based observations, this may make direct comparisons a challenge.

From the accumulated historical datasets that exist now, some representing 35-plus years, key trends and anomalies can be investigated, including the effects of climate change and climatic variability. Satellite remote sensing provides utility in situations where in situ monitoring is impractical; these measurements augment existing field-based methods by capturing spatial measurements where extensive distances, remoteness, and resources limit the frequency of field sampling visits and augment in situ sensor acquisitions by providing the spatial context to wider scales (e.g. to a water body or catchment). However, differences in observation between those from space and those made in the field need to be taken into account when comparisons are made, considering the limitations of both approaches. These differences perhaps present a challenge to the integration of observations from space into existing utility functions, but where there is an increasingly economically viable case for doing so.

Whilst there is value in the integration of space-based and field-based observations, there is similarly also value to be gained when EO data are assimilated with other data in models. EO data provides the means to keep models on track by adding information on spatial variability and address timeliness in the provision of extensive data in near real-time, but models can be predictive and can be used to estimate conditions when observations are not yet available. Observations help test the assumptions of the physical processes and their interactions upon which the models are based, assisting modelling in providing forecasting potential and skill (Jones et al. 2016). The integration of EO data and field-based observations and the integration of observations with modelling (‘satellite enabled’ modelling) are perhaps the areas most in need of further capability development, such that improved early warning tools are the key outcome.

This review has focused on data from those satellites currently capable of providing sustained operational delivery of consistent data. At present, this capability is largely restricted to those provided by public space agencies, where missions, such as the Landsat and the Sentinel series, have been designed from the outset to provide sustained operational monitoring into the foreseeable future with repeated launches of similar satellites to sustain the measurements. Significantly, these data are made available under free and open access policies (presently, it remains to be seen if commercial approaches can provide a sustained operational service niche to match the benefits of data provided by public space agencies). However, additional benefits of the free and open policy have been the development and delivery of ARD datasets (through consistent data pre-processing) and an increased focus on data exploitation, including the development of algorithms, tools, and applications covered in this report. ARD and data cube technologies (such as the DEA and EASI data cubes) have made it considerably easier to apply EOs for monitoring many of the biogeophysical properties of interest reviewed in this report. Custom algorithms made available through open-source allows for simpler integration into commercial Geographic Information System (GIS) and analytical platforms. Off-the-shelf products (e.g. WOfS, land cover classifications) are also increasingly available for incorporation into such platforms. Central to the ethos of these approaches is access to EO data and analysis platforms that are traceable, transparent, interoperable, and open source, facilitating a rapid expansion of user capabilities (Malthus et al. 2019).

Ultimately, these developments offer water utilities and agencies the potential to facilitate improved management of water resources and assets in terms of sustainability, quality, and future availability, as well as improving response effectiveness towards natural disasters and their aftereffects. The wider benefits include reduced risks to staff, potential water savings, improved efficiencies in environmental monitoring and improvements in productivity and information gained, competitive advantage, informed policy and management/regulation, as well as the additional wider economic and environmental benefits. Finally, these findings highlight the need for future research to explore a wider range of remote sensing applications in greater depth.

The authors acknowledge Melbourne Water, Water New South Wales and IconWater funding via Water Research Australia project number 1134. Also, the in-kind contribution from the CSIRO AquaWatch mission and the Civil Engineering Department at Monash University is appreciated.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.