Drinking waters are one of the most vulnerable resources on the planet and are endangered by recent climate change effects including increase of inland waters temperature and higher frequency of algae blooms. These algae blooms can be formed by toxin building cyanobacteria species, such as Microcystis, Anabaena or Planktothrix. Accordingly, the production of drinking water from surface water needs appropriate risk control measures to avoid impacts of toxic cyanobacteria to human health. Nevertheless, the information that water managers have on algal blooms and cyanobacteria evolution in the catchment is generally limited to some in-situ sampling every few weeks. A potential solution is offered by Earth Observation, offering a cost-effective way of frequently monitoring all the water bodies of interest. These data can then be integrated with in situ data to develop clear, robust, and proactive risk management protocols for drinking water plant managers.
Based on spectral information gathered by multi- and hyperspectral satellite missions, these water quality feature changes can be detected by applying state-of-the-art physics-based retrieval algorithms such as the Modular Inversion and Processing MIP developed by EOMAP. One essential MIP output water quality product for the drinking water application is the harmful algal bloom indicator eoHAB, which is sensitive to the appearance of cyanobacteria related pigments, i.e. phycocyanin and phycoerythrin. The product identifies reflectance and absorption discrepancies between the 550nm and 650nm wavelength bands, and a qualitative classification is provided. Further, the MIP architecture systematically manages the independent properties of sensor parameters and specific optical properties, as well as the radiative transfer relationships at 1 nm spectral resolution. This enables MIP to follow a sensor-agnostic approach with the capability to incorporate various satellite missions in a harmonized way. The usage of multiple satellite data sources is an important benefit when applying satellite-based monitoring in emergency settings.
Within the H2020 funded WQeMS project, tested the application of high resolution water quality products calculated by MIP, derived from Sentinel-2 A/B in 10m resolution as well as 2m resolution products from commercial distributed WorldView-2 data. This application was performed by EOMAP in collaboration with local water utilities and water managers (WQeMS partners CETAQUA, EMUASA and HIDROGEA, stakeholder River Basin Agency of Segura). The in-situ monitoring data gathered in past algae bloom events in the reservoir of Ojós and El Judío have been used to validate the sensitivity and suitability of the approach, especially looking at the retrieved harmful algae bloom indicator (eoHAB) and chlorophyll-a concentrations.
By adding hyperspectral PRISMA data to the analysis, the potentials of differentiation of algae groups are further tested. Additionally, frequency improvements by using the temporal higher resolved PlanetDoves fleet with over 180 satellites in space are examined. Data will be made available on the WQeMS platform with user oriented visualization, developed in close cooperation with the stakeholders.
The use of satellite images will bring more frequent and complete information on the state of the different reservoirs and ponds and will therefore allow a cost-efficient management of the risks induced by water quality changes. Indeed, a weekly monitoring of the system reservoirs and ponds would allow to alert early changes in water quality, and to plan in-situ analyses more efficiently. This can lead to a better understanding of the system to develop water quality forecasts.
This will allow for a better understanding of the processes at stake, serving as the link between physical and chemical parameters of importance for water treatment. Additionally, frequently updated information leads to faster decision making and tuning the treatment accordingly towards a better environmental footprint by adjusting doses of chemical reactive agents, thereby leading to a more economical use of resources and reducing costs.
Other benefits concern risk management through a better control and forecast of the hazards, e.g. algal blooms, reduction of the vulnerability of the plan by improving decision-making processes using new information in order to tune or change treatment and reduction of exposure of the treatment plants.
WQeMS project has received funding from the European Union’s Horizon 2020 Research and Innovation Action programme under Grant Agreement No 101004157.

An integrated combination of satellites, in situ sensors, and advanced modelling is key to monitoring and forecasting ecosystem changes: notably, the impacts of pressures such as population growth, extreme events, climate change and industry impacts on the health of our inland, coastal and marine ecosystems. The AquaWatch Australia Mission, (See LPS’22 abstract of Dekker et al.), proposes such a step change in monitoring technologies to support the scales and speeds at which our modelling is now required and to safeguard water bodies. We are developing an integrated nationwide ground-to-space national monitoring platform incorporating satellite and in situ sensor observations together with a dedicated data analysis platform to accomplish this ambitious goal.

The rollout of a nationwide in situ water quality monitoring sensor network requires sensor nodes that are cost-effective to construct and operate, easy to maintain, and deliver timely, robust and credible data which complement satellite observations for appropriate decision making. Sensors are required to meet three key roles in support of such a mission: 1) continuous observations of optical water quality parameters at all times, including under cloud cover, 2) satellite calibration and validation, and 3) observations of water quality parameters not measured by optical satellites.

An Internet of Things (IoT) solution is seen as the most cost-effective approach to meeting the goals of ubiquitous, autonomous sensing in both spatial and temporal domains across the Australian continent. However, central to the IoT concept are low-cost sensors; current suitable Commercial-Off-the-Shelf (COTS) water quality sensors are expensive, poorly adapted to IoT and are economically unviable for water quality management at scale. Reliable, cost-effective water quality sensors suitable for IoT adoption are still largely in the research domain. New sensors will need to be innovatively and robustly constructed for IoT systems that are characterised by resource constraints: in communication capabilities, energy, processing capabilities and limited data storage. Each of these constraints will influence sensor design, degree of maintenance and calibration, operation mode and sampling rate, on-board processing and type and rate of communication.

The paper will address the challenges we face in the development of a nationwide water quality network in support of satellite monitoring and modelling efforts. New thinking will be required to cost-effectively address the means of water quality parameter detection, their reliability, robustness and maintenance even in remote areas, hardware robustness, water resistance, biofouling, powering options, communication requirements and security/privacy procedures. Clear definition of the system requirements, along with new standards protocols will need to be developed to support the technology development, as will the challenge to integrate and analyse real-time data generated from a highly distributed and heterogenous sensor network.

Waterborne diseases are a major source of mortality in the world with more than 2.2 million deaths per year and even more morbidity cases every day. Some of the most serious infections such as cholera provoke humanitarian crises in regions most in need of health resources and clean water supplies. However, deaths attributed to waterborne diseases are also occurring in countries with modern health care systems. For example, in 2000, in Canada, the contamination of Walkerton town water system by Escherichia coli and Campylobacter jejuni led to the exposure of at least 2300 people, and the death of 7.

Canada harbours about 20% of the globe’s renewable freshwater and 50% of the world’s lakes, providing vast services for several societal practices including agriculture and pasture, drinking water and industries, and for recreative activities. Monitoring these resources over Canada’s enormous land mass in a changing climate with increasing human pressure on all natural resources is a major challenge. Increased availability of wide-scale spatial environmental data through satellite technologies, meteorological models and governmental census enable new opportunities to improve management and decision-making tools for public health authorities. Moreover, as bio-optical algorithms are expanded to inland waters, they offer more accurate water quality datasets. Light properties reveal interesting patterns with microbial species that are sensitive to ultraviolets.

Here, we used a pluridisciplinary approach for the modelling and mapping of potentially pathogenic microorganisms at the continental scale. Through the NSERC Canadian Lake Pulse Network, we sampled 664 lakes within 10 ecozones in Canada for 3 summers (June-September) from 2017 to 2019 and obtained surface water environmental DNA and bio-optical measurements from the lakes. In parallel, public environmental data from governmental agricultural census and scientific works, scaled to the lake watersheds were acquired. Potentially pathogenic genera (PPG) were extracted from our 16S and 18S rRNA gene amplicon datasets which include bacteria, fungi and protists, using the ePathogen Public Health Agency of Canada database.

We used a boosted regression tree (BRT) model on each identified PPG to determine the relative influence of environmental and bio-optical variables on their occurrences and relative abundances. As BRT do not provide a standardized way to test significance, we also trained 1000 bootstrap samples with replacement to provide 95% confidence interval estimates of each PPG prediction and the individual relative influence of each variable.

Our results provide the occurrences and geographical distributions for a range of health-relevant PPG found in Canadian lakes. Predictive maps are also presented for the most populated ecozones in Canada. By focusing solely on remotely or census derivable predictive variables, the approach should be applicable to many tens of thousands of lakes in Canada once inland water algorithms have improved sufficiently.

Lake ecosystems face severe anthropogenic forcing through changes in land use and climate, which compromise water quality and the provision of ecosystem services. In river-connected lake systems, individual lake responses will be influenced by fluvial processes such as flow through rates, as well as lake characteristics such as average depth, both of which influence water retention time. Thus, water retention time will determine the distribution of upstream eutrophication events along river-connected lake systems and drive ecological coherence, i.e. mechanisms of biological self-organization. Increasing the strength of lake-to-lake connectivity, i.e. enhancing flow through rate and, consequently decreasing the water residence time stimulate coherence with respect to water constituents, phytoplankton communities, primary production and related ecosystem functions. Thereby, increasing lake connectivity allows for faster and more extensive impact propagation through a lake chain, which may increase eutrophication impacts at the regional scale of the lake chain, potentially leading to a higher risk of widespread cyanobacteria blooms. A combination of in-situ sensor networks and airborne remote sensing measurements (aircraft and drone) with high spatial and temporal resolution is necessary to assess ecological coherence.
To test how lake-to-lake connectivity drives ecological coherence and eutrophication impacts along deep lake chains we conducted a controlled experiment in a large-scale enclosure facility - the LakeLab - installed in Lake Stechlin. In August 2019 we set up six experimental circular lake-chains of four mesocosms each to establish two levels of connectivity. The latter were based on typical epilimnetic water residence times of lakes in the region. At the start of the experiment, a storm event with nutrient-runoff was simulated in each first enclosure of the six chains by respectively mixing the epilimnion from 4 to 14 m into the deep layer and adding P and N (in Redfield ratio). Following that, high and low retention times were simulated by pumping epilimnetic water with a different rate from one enclosure into the next one along each circular lake-chain. During the experiment, we monitored temporal coherence of phytoplankton dynamics and several processes related to grazing, production and ecosystem functioning, such as greenhouse gases (CO2 and CH4). We combined high-resolution in-situ multi-sensor profilers measurements (light, temperature, pH, conductivity, oxygen, turbidity, chl-a) with high-throughput picture-based flow cytometry (FlowCam) and multispectral and hyperspectral remote sensing (airborne HySpex-cameras, drone imagery, field spectrometers) allowing ground validation and upscaling to regional scales. Additional aspects of fine-scale lake physics, optical properties and pelagic ecosystem processes were investigated via in situ and remote sensing approaches through participants of the AQUACOSM Transnational Access program (aquacosm.eu).
Our results indicate that after four weeks, a short retention time (30 d) synchronizes the plankton community, whereas a long retention time (300 d) leads to significant differences in phytoplankton community structure. Surprisingly, low phytoplankton biomass developed in the epilimnion after the mixing, while the deep chlorophyll maximum at 10 to 14 m depth, dominated by the cyanobacteria Planktothrix rubescence, re-established quickly. Low epilimnetic chlorophyll-a values make reflectance measurements particularly difficult. By combining in-situ multi-sensors and high throughput plankton analyses with near and far remote sensing, the performance of reflectance-based chlorophyll-a estimates for oligotrophic waters can be improved. These experimental results will help fine-tune remote sensing-based algorithms used for detection of chlorophyll-a and other optical active water constituents, as well as predictions on signal propagation along river-connected lake systems to support future lake monitoring and management. Our results demonstrate the potential of large-scale aquatic experimental facilities like the LakeLab for cross-instrument calibration and offer opportunities for collaboration and participation through the Transnational Access program within the EU-funded AQUACOSM-plus network.

In the face of Climate Change, climatic extreme events, in particular storms and heat waves, are becoming more frequent, a trend that is projected to continue in the future (IPCC 2014), posing a threat to freshwater bodies. Storms with intensive rainfall, are typically associated with in-flows of large loads of dissolved organic matter (DOM) and nutrients in different proportions depending on the land use of the catchment, which might trigger cyanobacterial blooms. Also heat waves have been shown to increase the frequency of toxic cyanobacterial blooms.
Monitoring, assessing and understanding these events is becoming increasingly important because of the negative impact they can have on the ecosystem services that lakes provide (water for drinking and irrigation, recreational use) and on the local economy (fisheries and tourism). High DOM levels in water result in formation of disinfection by-products (DBPs) such as trihalomethanes (THMs) when water supplies are chlorinated, associated with diseases of the liver, central nervous system, and an increased risk of cancers. Cyanobacterial blooms, boosted by abrupt nutrient loading or heat waves can produce toxins that affect water use for human consumption and for recreation. Both processes can lead to substantial costs for water managers.
In this context, EO technologies can help to better quantify, and thus understand the behavior of lakes after the extreme events by adding a spatially explicit component to the traditional sampling schemes.
We explored the potential of EO for both monitoring the development of the immediate consequences of these extreme events and study the long-term trends with satellites and drones through a mesocosm experiment. We performed a controlled large lake enclosure environment at the IGB-LakeLab infrastructure through support Transnational Access provision of the H2020 EU AQUACOSM project. The IGB-LakeLab facility of the Leibniz Institute of Freshwater Ecology and Inland Fisheries (http://www.igb-berlin.de/en/lakelab) has 24 in situ lake water enclosures, 9 m diameter, thus large enough that it is feasible to use them for some remote sensing ground truthing activities. The design of the experiment was developed to match the basic conditions of an international experiment (JOMEX) conducted in the framework of the EU AQUACOSM project. This JOMEX-CONNECT experiment was based on the addition of phosphorus, nitrogen and HuminFeed (as a browning agent) to 12 mesocosm with 4 extra mesocosm as control samples (no additions) and was run for 4 weeks during July and August 2021.
We performed above-water and in-water radiometry with different methodologies, including a low-cost radiometric option on board a drone in which a STS-VIS Ocean Insight Spectrometer was mounted in a rotative platform to obtain in-flight radiometric point measurements.
The spectral changes of the lake water color, as a subject to different treatments, were studied over time, defining different browning and trophic conditions described through optically active constituents concentrations (DOM and Chl-a). Variations in the measured spectra are analyzed to understand to what extent the extreme events and their recovery can be detected with remote sensing. Although Sentinel-2 was not designed for waterbodies it is the only satellite sensor with sufficient spatial resolution and revisit time to allow monitoring of smaller lakes. Therefore, we resampled hyperspectral reflectance to Multispectral Instrument (MSI) spectral bands in order to assess whether Sentinel-2 can be used in monitoring lake recovery after extreme events. Moreover, the optical variability observed in the mesocosms during the experiment will allow also to make a step forward in developing lake remote sensing algorithms in general.

The increase of industry, agriculture, and urbanization, among others, show severe consequences over the water quality of rivers and lakes. Appropriate management and effective policy development are required to deal with the problems of surface water contamination around the globe. However, spatial, and temporal variations challenge an adequate water quality decision and policy making. In this research we explore how remote sensing may be highly beneficial to understand the spatio-temporal variations in order to guide policy developments. At the same time, this analysis may provide a better understanding of the cause-impact relations in water quality management of river basins.

To conduct this research, a spatio-temporal analysis of satellite images from 2006 to 2018 was applied to the Katari River Basin (KRB), located in the Bolivian Andes. The KRB incorporates the presence of mining, urban, industrial, and agricultural developments. These human developments largely modified the surface water quality of the system with severe consequences for local indigenous communities allocated in the downstream region (Agramont et al., 2021; Archundia et al., 2016; MMAyA et al., 2014). Moreover, this river basin discharges over the Titicaca Lake, the most important water resource in the Andes.

To understand the modifications linked to the water contamination phenomena, this research employed Landsat 7 images to generate land cover (LC) maps for the period of study. Subsequently, a trajectory analysis was performed by intersecting the maps and classifying the 1023 different trajectories obtained into 3 main categories. On the other hand, Normalized Difference Vegetation Index (NDVI), Normalized Difference Aquatic Vegetation Index (NDAVI), Turbidity, and Chlorophyll-a were analyzed at the outlet of the basin which exposed the spread of eutrophication in Lake Titicaca. Finally, a combination of GIS tools and a multi-criteria analysis based on the Analytic Hierarchy Process (AHP) was used to re-design the water quality monitoring system and allocate sampling sites based on anthropic, physiographic and water quality aspects.

The results revealed a 123% increase in urban areas in 2018 compared to 2006. At the same time, important impacts were detected on the lake’s shores. The analysis of the relation between inland LC modifications and NDVI shows a ratio of approximately 1:3 among eutrophicated areas downstream versus urbanized areas upstream. This indicates that for the study period for every 3 km2 of urban built-up area, the extent of eutrophicated areas in Lake Titicaca’s shores increased by 1 km2. This comparison shows a significant influence of the urban growth over the lake contamination, mainly due to the untreated wastewaters and effluents from industries that reach the water course.

Even though there has been interest and resource allocation to this basin from different actors, this research shows that what is being done is not enough. The current monitoring system is not efficient and the quality of the water in the rivers and in the outlet of the basin is deteriorating. Both the sampling site selection and the trajectory analysis are relevant tools to policy-makers, as they display the LC changes in the basin, which makes it possible to identify priority areas to enact decision-making. The understanding of the source-impact relation between urban areas and eutrophication is evident through this type of assessment, and can be used to raise awareness, take actions and allocate resources for effective policy responses.