ENHANCING WATER QUALITY MONITORING AND GOVERNANCE THROUGH HELOISA: AN EO APPROACH TO AQUATIC SYSTEM MANAGEMENT

ENHANCING WATER QUALITY MONITORING AND GOVERNANCE THROUGH HELOISA: AN EO APPROACH TO AQUATIC SYSTEM MANAGEMENT Konstantinos Vlachos, Konstantinos Karystinakis, Anastasia Moumtzidou, Ilias Gialampoukidis, Stefanos Vrochidis, Ioannis Kompatsiaris CDXi Solutions P.C. ABSTRACT Hellenic Operational Integrated Service for Aquatic systems (HELOISA) is one of the projects of the Greek National SmallSat Programme which implements the Water Monitoring Service. The project builds upon three pillars; Water Quantity, Water Quality and Maritime Surveillance. This work focuses on the Water Quality module and specifically on the core products and technical approach that encompasses. It utilizes Copernicus data, as well as sensors of the Greek SmallSat constellation that provide optical and thermal data. The module covers various water body types, offering maps of water quality proxy variables such as water temperature, chlorophyll-a and water pollutants. Validation and evaluation activities include exploitation of existing historical and newly acquired data ensuring generated product quality. Overall, the HELOISA system is scheduled to be operational in mid-2026, consistently providing water quality products in the Greek territory supporting authorities in informed decision-making and policy implementation. Index Terms— Earth Observation, Small Satellite, Artificial Intelligence, Water Quality, Copernicus, Environmental Monitoring 1. INTRODUCTION Water resources are of vital importance to ecosystems, human health, and economic prosperity. In Greece, a country characterized by a complex network of inland, coastal and marine water bodies, effective water monitoring systems are essential for sustainable resource management and environmental protection. Inland waters are especially significant as they provide drinking water to large populations including Athens and Thessaloniki, Greece's largest cities, home to over half of the country's population, as well as they support irrigation in agriculture, hydropower generation and biodiversity conservation. In addition, coastal and marine waters in Greece are vital to the country's economy, environment and security, supporting tourism activity, sustaining rich marine biodiversity, and enabling key sectors like fisheries, aquaculture and maritime, among others. Recognizing those needs and also understanding that space is a key enabler for digital transformation, the Hellenic Ministry of Digital Governance and the Hellenic Space Center, with the assistance of the European Space Agency (ESA), have initiated the Greek National Satellite Space Project. This project is an important step for the materialisation of the strategy of Greece for the utilisation of space technologies and applications and their uptake in the National economy. It includes the development and launch of a small satellite constellation that will cater applications for telecommunications and earth observation for their use in governmental satellite services, cartography, inland, coastal and marine water monitoring, precision agriculture, land and forest monitoring, as well as border security. The project consists of three Axes. Axis 1 (1.1, 1.2) and Axis 2 comprise the space components responsible for the development and launch of the smallsats. Axis 1.1 will provide thermal data with two spectral bands in about 200m spatial resolution. Axis 1.2 will provide SLC and GRD Synthetic Aperture Radar (SAR) data in various imaging modes (e.g., Scan, Strip, Spot etc.) and spatial resolutions ranging from 0.25m to 15m. Axis 2 is dedicated to multispectral and hyperspectral data in high and very high resolutions reaching about 0.9m (panchromatic band), 18m (VNIR hyperspectral bands), and 3.3m and 19.3m for VNIR and SWIR bands for the multispectral sensor, respectively. On the other hand, Axis 3 is dedicated to the ground components that include the hardware and software facilities, as well as the end-user thematic applications of the different Earth Observation (EO) services that will be delivered alongside the satellite constellations, by mid-June 2026. The thematic applications are categorized in Land, Water, Forest, Agriculture and Security. 2. APPROACH HELOISA is the project responsible for the delivery of the Water Monitoring Service which aims to develop an advanced monitoring system tailored to the specific needs and requirements of the Greek territory. The Water Monitoring Service builds upon the foundation laid by previous Earth observation initiatives such as the Copernicus program and leverages cutting-edge technology to enhance spatial, temporal, and thematic resolution. By integrating satellite imagery, advanced sensors, and artificial intelligence algorithms, the system aims to provide comprehensive monitoring of water quantity, quality, and maritime surveillance. In this work, we focus on the Water Quality module of the HELOISA project. The water quality module will be delivering Level-3 satellite products utilizing both Copernicus and the Greek SmallSat data. The areas of interest that will be covered include the majority of the Greek lakes and some lagoons, as well as all coastal and marine waters, that go beyond the 12 nautical miles. Some of the products that are associated with sudden natural or human-induced environmental changes will be delivered on a daily basis, while the rest will be delivered with higher latency on a weekly basis, covering a large part of the Greek territory depending on the satellite constellation swath. Aligned with the objectives outlined by the Hellenic Ministry of Digital Governance and the Hellenic Space Center, HELOISA encompasses a multi-phase approach, encompassing system definition, technical specifications, component design, platform integration, testing, and operational implementation. Through close collaboration with stakeholders and adherence to stringent quality standards, the project seeks to address critical water management challenges while ensuring the sustainability and resilience of water resources in Greece. 3. APPLICATIONS 3.1. Ocean Color and Temperature The first water quality application of HELOISA is the Ocean Colour and Temperature delivering products at 10m and about 200m nominal spatial resolution, respectively. In particular, chlorophyll-a and Sea Surface Temperature (SST) are retrieved for the coastal and marine waters of the Greek territory (Fig. 1). Chlorophyll-a retrieval is realized utilizing the Copernicus Sentinel-2 multispectral data. They undergo an atmospheric correction that specifically treats the ubiquitous sunglint effects, utilizing the latest version of Polymer [1] that is adapted to Sentinel-2. Since Sentinel-2 is not specifically designed for marine water applications, several types of noise and various effects exist, such as striping/parallax effect, high frequency noise due to waves, and ship wakes, among others. Those effects need special treatment which is developed on a data-driven basis. The products are foreseen to be delivered weekly including all open waters. On the other hand, the retrieval of the SST is realized utilizing the thermal sensor of the Greek SmallSat data from the Axis 1.1. The constellation is originally aimed for forest fire applications, however HELOISA takes advantage of the provided thermal channels (3.8μm, 11.45μm) and develops a two-channel method for SST retrieval. The top-of-atmosphere observations are translated to water surface reflectances, brightness temperatures and, as a consequence, to SST. This is made possible by generating a Look-Up Table and an atmospheric correction using the Libradtran Radiative Transfer Model (RTM) library [2]. In addition, a dedicated algorithm will be offering cloud masks of different certainty levels. Fig. 1. Marine water quality preliminary products for the Greek territory with chlorophyll-a from Sentinel-2 (left) and preliminary SST from Forest-2 mission (right). 3.2. Muddy Water and Industrial waste The second water quality application of HELOISA is the Muddy water and Industrial waste mapping delivering products at a 10m and less than 5m nominal spatial resolution, for Sentinel-2 and Axis 2 data, respectively. The covered territory is foreseen to be almost all of the Greek lakes, as well as coastal waters less than 12 nautical miles. Concerning muddy waters, there are a number of studies attempting to monitor turbid and sediment-laden waters based on satellite remote sensing. Traditionally, the focus has been put on parameter retrieval of turbidity and total suspended matter, but they are not associated with the potential source origin of the polluting sediment. The source could be natural or human-induced, such as industrial waste. The Muddy water and Industrial waste application of HELOISA aims to give semantic information to the sedimentladen waters. The application builds upon Sentinel-2 data that are annotated based on an ensemble methodology as presented in the MUDDAT dataset [3]. An extension of it is implemented, which adds a list of regions presenting coloured waste waters due to industrial activities around the globe (Fig. 2). A custom deep learning framework based on U-Net is trained after performing data preparation such as augmentation and other techniques to adjust for the inherent class imbalance. The products undergo post-processing steps such as land-sea masking using the Copernicus 10m Digital Elevation Model (DEM), and filtering to account for systematic and occasional noise effects. Finally, transfer learning is applied so as to generate products using Axis 2 as input data. This is made possible by exploiting the satellite specification similarities but also adjusting for the differences such as different number of spectral bands and pixel size. Fig. 2. Industrial waste mapping with a custom U-Net model with the True Color Composite (left) and binary mask (right). 3.3. Oil spills and Surface formations The third water quality application of HELOISA is the Oil spills and Surface formations mapping, delivering products at a 10m nominal spatial resolution utilizing the Copernicus Sentinel-2 data. The covered territory consists of a list of the largest Greek lakes. In particular, oil spills are largely identified through radar data, which however present limitations when it comes to inland waters, since the latter demonstrate significant look-alikes due to lake morphology and topography, and low surface roughness due to inconsistent wind conditions. The application fills this gap by offering mapping of oil spill and other suspicious formations at the surface of inland water bodies utilizing multispectral data (Fig. 3). The technological foundation of the approach builds upon the only public multispectral dataset, i.e. Marine Debris and Oil Spill (MADOS) [4] that includes oil spills. A state-of-the-art model (i.e., MariNeXt) which shows high performance is combined with a unique Hydro Foundation Model (https://github.com/isaaccorley/hydro-foundation-model), which gives higher generalizing power. To make this happen, special modifications of the two models are necessary to adjust for the different number of bands and preprocessing specifications. Finally, the product comes with relevant quality flags and masks including sensor viewing and sun geometries to assist the user interpretation 3.4. Water Quality Features The fourth, and last, water quality application of HELOISA is the Water Quality Features delivering products at 10m nominal spatial resolution. In particular, the application generates essential water quality variables for inland waters comprising the majority of Greek lakes. This not only includes artificial and natural reservoirs, but also some lagoons. The focus has been put on variables that offer complementary information to the Muddy water and Industrial waste application, which are chlorophyll-a and turbidity (Fig. 4). The estimation of chlorophyll-a and turbidity concentrations in water bodies serves as a major indicator of algal blooms, agricultural practices and pollution. Their retrieval is realized utilizing the Copernicus Sentinel-2 multispectral data after employing the C2RCC [5] atmospheric correction that has been proved to perform well in inland water applications, considering complex Case 2 waters, and also treats adjacency effects, among others. This is possible by using auxiliary information, such as land elevation (e.g., SRTM 30m DEM), air temperature and pressure (e.g., ERA5), total ozone column, and water salinity. The chl-a retrieval is based on the incorporation of red-edge and near-infrared spectral regions, as well as for turbidity [6]. In addition, auxiliary data will be delivered such as Trophic State Index [7], which indicates the eutrophic state of the waters. Finally, products undergo necessary post-processing and offer relevant quality flag layers. Fig. 3. Oil spill mapping utilizing a custom hybrid deep learning framework with the True Color Composite (left) and the respective binary mask (right). Fig. 4. Inland water quality products for the Greek territory with chlorophyll-a (left) and turbidity (right). 4. VA L I D AT I O N A N D E VA L U AT I O N In order to ensure the high quality of the generated Level-3 products, HELOISA and the Water Quality module, in particular, at its core, adopts relevant validation and evaluation practices. To this end, the algorithms of the output products are first verified utilizing proxy/simulated data that are provided in the context of the Greek SmallSat Programm, before the advent of the actual Axes data. For instance, in the case of SST, existing Forest-2 mission data ensure the validity of the proposed retrieval algorithm, while in the case of muddy waters, existing Very High Resolution multispectral data are utilized for transfer learning. Furthermore, concerning the water quality variables such as chl-a, turbidity and SST, existing in situ historical data are being exploited (e.g., from ARGO[8]), and new fieldwork campaigns are being conducted for all inland, coastal and open waters. Additionally, a match-up analysis protocol has been determined and followed that ensures transparency and quality of outcomes. Finally, manual photointerpretation and quality control of products is conducted by remote sensing experts, and comparison with established existing open datasets derived from Copernicus and other researchers. 5. IMPACT The use of EO technology for monitoring water quality brings a wave of positive change across societal, scientific-technical, and economic dimensions. HELOISA leverages national infrastructure and provides satellite-driven insights into inland, coastal and open waters thus addressing pressing environmental challenges while unlocking new opportunities for innovation and growth. 5.1. Societal Impact Protecting public health stands at the forefront of societal benefits of this EO-driven approach on water quality, as early detection of harmful algal blooms, bacterial contamination, or chemical pollutants allows authorities to issue timely warnings and mitigate risks to drinking water supplies and recreational users. Beyond health, by leveraging national satellite infrastructure, it strengthens the country’s autonomy in environmental monitoring, ensuring that critical data for decision-making is generated domestically, thus enhancing national capacity and resilience. Moreover, access to high-quality EO data empowers water authorities to enforce regulations more effectively, supporting compliance with major frameworks such as the Water Framework Directive and the Marine Strategy Framework Directive and shifting governance from reactive crisis management toward proactive and evidence-based management. Furthermore, by integrating EO-derived insights into existing monitoring systems, the project improves the transparency and accountability of water management practices. 5.2. Technical and Scientific Impact The integration of EO data into water quality monitoring offers critical technical and scientific advantages for both Regional Water Utility operators and National governance bodies. By providing continuous, wide-scale, and standardized observations, the project enhances the ability of water utilities - who are directly responsible for distributing safe drinking water - to monitor the quality of their source waters more efficiently and with greater spatial and temporal coverage than traditional sampling alone. For Regional Water Utility operators, early identification of emerging threats such as algal blooms, turbidity spikes, or chemical pollutants enables faster, targeted responses that safeguard drinking water treatment processes and distribution networks. Instead of relying solely on periodic field sampling, Water utility operators gain access to near-real-time intelligence, allowing for more proactive management of risks and better protection of public health. At the National level, the project significantly strengthens the technical capacity of Regional Environmental Departments, the Ministry of Environment, and other governmental agencies. With systematic EO data streams, authorities can implement broader surveillance of inland and coastal waters, ensuring regulatory compliance, detecting environmental trends, and evaluating the effectiveness of water protection measures. It provides the scientific backbone for more strategic policymaking, supporting long-term planning for water security, climate adaptation, and biodiversity conservation. The use of national satellite infrastructure also ensures that data sovereignty is maintained, with critical environmental information produced and controlled within the country. This promotes scientific independence and builds national expertise in remote sensing and environmental monitoring technologies. Importantly, the project fosters an integrated approach where EO data is not a replacement for in-situ monitoring but a powerful complement, bridging gaps and optimizing resource allocation. This hybrid monitoring model raises the scientific standard of water quality assessments and offers a replicable framework for future environmental applications, ensuring that both operational management needs and strategic governance priorities are met in a coordinated, technologically advanced manner. 5.3. Economic Impact By utilizing EO data, the cost of continuous water monitoring is drastically reduced compared to traditional field-based methods, which are labor-intensive, time-consuming, and geographically limited. For Regional Water Utilities, this means that broader and more frequent assessments of source water bodies can be achieved without proportional increases in operational expenses. Early detection of potential risks - such as contamination events or seasonal degradation in water quality - allows utilities to plan interventions more efficiently, avoiding emergency responses that are often costly and disruptive. This contributes to a more stable and predictable operational environment, protecting critical infrastructure and minimizing financial risks associated with treatment failures or public health incidents. On the National level, the availability of standardized, large-scale water quality data supports smarter investment planning. Environmental agencies and ministries can prioritize actions based on comprehensive, evidencebased assessments, ensuring that limited resources are allocated to the most critical areas. Moreover, the integration of national satellite capabilities into operational services can stimulate the growth of value-added industries such as geospatial analytics, environmental consulting, and digital platform development and strengthens the country’s positioning in the rapidly growing global market for Earth Observation applications. By enhancing water resource governance through EO-based monitoring, the project also supports sectors that depend on water quality, such as tourism, fisheries, agriculture, and energy, ensuring their long-term economic viability, acting as a catalyst for economic modernization, resilience, and sustainable growth tied to better water management practices. 6. ACKNOWLEDGMENTS The Project: Small-Satellites (Measure ID 16855) is implemented by the Hellenic Ministry of Digital Governance with the European Space Agency (ESA) Assistance in the Management and Implementation. The project is part of the National Recovery and Resilience Plan ‘Greece 2.0’, which is funded by the Recovery and Resilience Facility (RRF), core programme of the European UnionNextGenerationEU. The HELOISA consortium comprises TotalView as coordinator, Institute of Communication and Computer Systems (ICCS) of the Technical University of Athens, CDXi Solutions P.C., Center of Security Studies (KEMEA), EYATH S.A. and Lever-Development Consultants S.A. CDXi solutions is an incubatee of ESA BIC Greece and a spin-off company of the Information Technologies Institute, part of the Centre for Research and Technology Hellas (CERTH). REFERENCES [1] Steinmetz, F. and Ramon, D., 2018, October. Sentinel-2 MSI and Sentinel-3 OLCI consistent ocean colour products using POLYMER. In Remote sensing of the open and coastal ocean and inland waters (Vol. 10778, pp. 46-55). SPIE. [2] Emde, C., Buras-Schnell, R., Kylling, A., Mayer, B., Gasteiger, J., Hamann, U., Kylling, J., Richter, B., Pause, C., Dowling, T. and Bugliaro, L., 2016. The libRadtran software package for radiative transfer calculations (version 2.0. 1). Geoscientific Model Development, 9(5), pp.1647-1672. [3] Psychalas, C., Vlachos, K., Moumtzidou, A., Gialampoukidis, I., Vrochidis, S. and Kompatsiaris, I., 2024, July. MUDDAT: A Sentinel-2 Image-Based Muddy Water Benchmark Dataset for Environmental Monitoring. In IGARSS 2024-2024 IEEE International Geoscience and Remote Sensing Symposium (pp. 2902-2907). IEEE. [4] Kikaki, K., Kakogeorgiou, I., Hoteit, I. and Karantzalos, K., 2024. Detecting marine pollutants and sea surface features with deep learning in sentinel-2 imagery. ISPRS Journal of Photogrammetry and Remote Sensing, 210, pp.39-54. [5] Brockmann, C., Doerffer, R., Peters, M., Kerstin, S., Embacher, S. and Ruescas, A., 2016, August. Evolution of the C2RCC neural network for Sentinel 2 and 3 for the retrieval of ocean colour products in normal and extreme optically complex waters. In Living Planet Symposium (Vol. 740, p. 54). [6] Nechad, B., Dogliotti, A., Ruddick, K. and Doxaran, D., 2016, May. Particulate backscattering and suspended matter concentration retrieval from remote-sensed turbidity in various coastal and riverine turbid waters. In Living Planet Symposium, Proceedings of the conference held (pp. 9-13). [7] Carlson, R.E., 1977. A trophic state index for lakes 1. Limnology and oceanography, 22(2), pp.361-369. [8] Johnson, G.C., Hosoda, S., Jayne, S.R., Oke, P.R., Riser, S.C., Roemmich, D., Suga, T., Thierry, V., Wijffels, S.E. and Xu, J., 2022. Argo—Two decades: Global oceanography, revolutionized. Annual review of marine science, 14(1), pp.379-403.