Periodic Reporting for period 2 - SPACE-O (Space Assisted Water Quality Forecasting Platform for Optimized Decision Making in Water Supply Services)
Reporting period: 2017-11-01 to 2018-12-31
SPACE-O integrates state-of-the-art satellite technology and in-situ monitoring with advanced modeling and ICT tools. Thanks to an operational coupling of hydrological modelling in the upstream catchments and hydrodynamic and water quality (WQ) modelling inside the reservoir, the operational SPACE-O service line generates forecasts (up to 10-days) of water quantity and quality characteristics in the reservoirs, updated on a daily basis.
The service line is supported by an automated Data Assimilation (DA) workflow which uses EO-based WQ products (e.g. chlorophyll-a, turbidity) and datasets from in-situ sensors for correcting the model state initialization and improving forecasting skill. SPACE-O service line establishes a complete operational value chain from EO to water business sector. The overall objectives of SPACE-O are listed below:
1) Derive new and improved EO surface water quality products by exploiting the full potential of Copernicus data and services
2) Automate near-real-time assimilation of EO-based water quality products into hydro-ecological modelling for improving forecasting skill
3) Develop a risk-based DSS to enable cost-effective and environmental sustainable WTPs operations
4) Commercialize a complete, operational service line for the water industry, exploiting the market potential for advanced EO products and services.
The SPACE-O platform includes 5 different tools: the Water Information System that combines in situ measurements, satellite images and modelled hydrological, hydrodynamic and ecological data to fill in the information gaps in space and time about water quality, and to produce short term forecasts (up to 10 days) with high spatial and temporal resolution; An Early Warning System that indicates incidences of water quality deterioration with potentially high impact on downstream water utility services; A tool for Water Treatment Plant Optimization, which provides specific water treatment options based on forecasted raw water quality and advanced machine-learning algorithms that allow for improving efficiency in both effluent drinking water quality and financial performance; A tool for Catchment Risk Assessment, providing a method for water managers to identify hazards within the upstream catchment area and asses the level of risk to their water systems; and Improve My Water, a citizen science platform to report, administer and analyze local water issues.
The EO and modelling components of SPACE-O service line were tested and validated in the 2 project case studies, in Aposelemis reservoir in Crete (GR) and in Mulargia reservoir in Sardinia (IT) in close collaboration with the end-users. The results from the validation activities were communicated to potential end-users through a series of evaluation workshops which were held in Belgium, Singapore, Sweden and Austria. These workshops offered a hands-on experience to the participants and allowed them to judge the applicability and the usefulness of the SPACE-O products, while at the same time provided in depth information about the scientific background, the strengths and limitations of the service line, thus helping to establish confidence to potential end-users.
SPACE-O uses the latest generation of satellite sensors (i.e. Landsat-8 OLI, Sentinel-2 MSI, Sentinel-3 OLCI) with improved resolutions for estimating WQ products (e.g. chlorophyll-a, turbidity) as well as for generating new (e.g. suspended matter) in an operational environment.
Hydrological modelling in Aposelemis and Mulargia catchments is performed with HYPE model using meteorological forcing from ECMWF and high-resolution local NWP schemes. The re-calibration of model parameters in the 2 case studies indicated the importance of downscaling large scale continental hydrological services into local scale. Scientific experiments in Lake Garda (IT) and Umealven River (SE) demonstrated the power of dynamic EO information (e.g. snow cover, evapotranspiration) in improving streamflow performance.
Hydrodynamic and ecological modelling in reservoirs was performed using the Delft3D models which were calibrated with a metamodeling approach, allowing for an efficient parameter fine-tuning. The ecological modelling chain is supported by an automated workflow for assimilation of EO-based WQ products (e.g. chlorophyll-a, turbidity, water temperature) and datasets from ground-based monitoring stations using three different techniques (i.e. EnKF, DI and WA). DA revealed a great potential for reducing the mean error and safeguarding against extreme errors for chlorophyll-a.
Simulation of WTPs is performed using data-driven models based on Random Forests and Gaussian Process Regression. The models were trained with historical operational datasets and achieved an adequate description of turbidity variations in the pre-ozonation, pre-oxidation and coagulation units. Using an inverse-modeling approach, specific treatments options that reduce energy and chemicals consumption, can be identified.
The SPACE-O platform provides tools that help reservoir managers to increase the informational dataset available for decision making. The 10-days forecasting capacity of SPACE-O enables managers to reduce impacts from water quality outbreaks through the assessment and implementation of proactive strategies (e.g. by transferring water from upstream reservoirs in the case of Mulargia). At the WTP level SPACE-O facilitates operators to achieve efficient performance in terms of drinking water quality and functional costs. The application of data-driven models in Aposelemis (GR) and Simbirizzi (IT) WTPs indicated that an overall coagulant cost reduction in the order of 6% and 10% respectively is possible without deteriorating the efficiency of turbidity removal.