Periodic Reporting for period 2 - PrimeWater (Delivering Advanced Predictive Tools form Medium to Seasonal Range for Water Dependent Industries Exploiting the Cross-Cutting Potential of EO and Hydro-Ecological Modeling)
Período documentado: 2021-06-01 hasta 2023-05-31
Advances in Earth Observation can be a catalyst to promote technological innovation in Integrated Water Resources Management, by providing a sustainable and effective way for monitoring inland waters, but also by increasing adaptive capacity and building resilience to hydro-ecological hazards when combined with modelling applications.
The EU funded PrimeWater project will add value to EO data and other Copernicus sectoral services, through cross-cutting research, using state of the art process-based and data-driven modelling for predicting hydrological extremes and Harmful Algae Bloom events in lakes and reservoirs. The aim is to improve the skill of hydro-ecological forecasts and their related impact at different spatial and temporal scales, by integrating satellite imaging spectrometry into the modelling chain through advanced Data Assimilation techniques and Machine Learning algorithms.
By providing increased situational intelligence and enhanced early warning capabilities, PrimeWater enables informed decision making and allows water managers to optimize the operation of their downstream water services and to better prepare against forthcoming critical changes in water quantity and quality.
PrimeWater has identified 3 Strategic Objectives related to the technologies, products and services that establish a complete value chain from EO to the water business sector: 1) to exploit the possibilities of multiple-mission water quality products for delivering enhanced monitoring of inland waters, 2) to advance research and innovation on Copernicus Data through cross-cutting, data-driven applications, and 3) to exploit the potential and opportunities of the data-driven economy in the water sector.
The EU funded PrimeWater project will add value to EO data and other Copernicus sectoral services, through cross-cutting research, using state of the art process-based and data-driven modelling for predicting hydrological extremes and Harmful Algae Bloom events in lakes and reservoirs. The aim is to improve the skill of hydro-ecological forecasts and their related impact at different spatial and temporal scales, by integrating satellite imaging spectrometry into the modelling chain through advanced Data Assimilation techniques and Machine Learning algorithms.
By providing increased situational intelligence and enhanced early warning capabilities, PrimeWater enables informed decision making and allows water managers to optimize the operation of their downstream water services and to better prepare against forthcoming critical changes in water quantity and quality.
PrimeWater has identified 3 Strategic Objectives related to the technologies, products and services that establish a complete value chain from EO to the water business sector: 1) to exploit the possibilities of multiple-mission water quality products for delivering enhanced monitoring of inland waters, 2) to advance research and innovation on Copernicus Data through cross-cutting, data-driven applications, and 3) to exploit the potential and opportunities of the data-driven economy in the water sector.
PrimeWater explores the scientific frontiers of EO observation technology by exploiting the potential of multi- and hyper-spectral sensors in water quality monitoring and demonstrating their added value in hydro-ecological forecasting services.
PrimeWater has formulated the Earth Observations Science Virtual Lab which aims to investigate major scientific questions related to EO science and Water Resources Modelling, and contains 9 scientific experiments grouped in three scientific domains: A) remote sensing; B) process-based hydro-ecological modelling; and C) data driven water quality modelling. In (A) hyperspectral images from PRISMA and DESIS missions have been collected and processed using existing and new water quality algorithms for estimation of cyanobacterial concentrations. On the same time a series of in situ, airborne and UAV flight campaigns have been performed in Mulargia and Flumendosa reservoirs to explore the usage of concurrent measurements in aquatic ecosystems mapping. In (B) three-dimensional hydrodynamic and water quality models for lake Harsha and Mulargia reservoir have been setup and calibrated. Data assimilation techniques have been developed and used to benchmark the improvement of model performance using historical multispectral water quality products (e.g. chlorophyll-a, temperature, turbidity) from Sentinel 2 and Landsat 8. In lake Hume a one-dimensional process model coupled with a cyanobacteria growth model has been used to replicate the timing and magnitude of cyanobacteria blooms. In Umeälven River the Ensemble Kalman Filter method was used to assimilate EO-derived and in-situ data aiming to quantify their potential for seasonal hydrological forecasting. In (C) an experiment using a Random Forest algorithm was conducted in Mulargia reservoir and Lake Harsha to identify the most significant hydrometeorological drivers for forecasting of chlorophyll-a values. Consequently, two data-driven approaches, i.e. an ensemble learning method and a Gaussian Process Regression model, were trained for short-term forecasting horizons and compared against naïve predicting alternatives.
PrimeWater has also defined the functional and technical requirements of its operational platform that will offer a) EO-based water quality monitoring, b) operational hydrological (short-term and seasonal) and ecological (short-term) forecasting, c) Early warning functionality and d) a water blending optimization tool for managing interconnected reservoirs. A comprehensive consultation procedure has been applied for consolidating user requirements in the development of the proposed PrimeWater services, including the organization of the 1st (out of 3) Multi User Panel workshop, with representatives from various water sectors.
On the same time behavioral approaches have been used to investigate the determinants of adoption of innovative EO and forecasting services. A pilot-survey has been conducted, using a discrete choice experiment to collect information from participants on user preferences towards different monitoring and forecasting earth observation services.
A 1st prototype of the operational PrimeWater platform has been released online at M18, which will be communicated to the end-users to receive their feedback and initiate a round of further customization and improvements of the offered services.
PrimeWater has formulated the Earth Observations Science Virtual Lab which aims to investigate major scientific questions related to EO science and Water Resources Modelling, and contains 9 scientific experiments grouped in three scientific domains: A) remote sensing; B) process-based hydro-ecological modelling; and C) data driven water quality modelling. In (A) hyperspectral images from PRISMA and DESIS missions have been collected and processed using existing and new water quality algorithms for estimation of cyanobacterial concentrations. On the same time a series of in situ, airborne and UAV flight campaigns have been performed in Mulargia and Flumendosa reservoirs to explore the usage of concurrent measurements in aquatic ecosystems mapping. In (B) three-dimensional hydrodynamic and water quality models for lake Harsha and Mulargia reservoir have been setup and calibrated. Data assimilation techniques have been developed and used to benchmark the improvement of model performance using historical multispectral water quality products (e.g. chlorophyll-a, temperature, turbidity) from Sentinel 2 and Landsat 8. In lake Hume a one-dimensional process model coupled with a cyanobacteria growth model has been used to replicate the timing and magnitude of cyanobacteria blooms. In Umeälven River the Ensemble Kalman Filter method was used to assimilate EO-derived and in-situ data aiming to quantify their potential for seasonal hydrological forecasting. In (C) an experiment using a Random Forest algorithm was conducted in Mulargia reservoir and Lake Harsha to identify the most significant hydrometeorological drivers for forecasting of chlorophyll-a values. Consequently, two data-driven approaches, i.e. an ensemble learning method and a Gaussian Process Regression model, were trained for short-term forecasting horizons and compared against naïve predicting alternatives.
PrimeWater has also defined the functional and technical requirements of its operational platform that will offer a) EO-based water quality monitoring, b) operational hydrological (short-term and seasonal) and ecological (short-term) forecasting, c) Early warning functionality and d) a water blending optimization tool for managing interconnected reservoirs. A comprehensive consultation procedure has been applied for consolidating user requirements in the development of the proposed PrimeWater services, including the organization of the 1st (out of 3) Multi User Panel workshop, with representatives from various water sectors.
On the same time behavioral approaches have been used to investigate the determinants of adoption of innovative EO and forecasting services. A pilot-survey has been conducted, using a discrete choice experiment to collect information from participants on user preferences towards different monitoring and forecasting earth observation services.
A 1st prototype of the operational PrimeWater platform has been released online at M18, which will be communicated to the end-users to receive their feedback and initiate a round of further customization and improvements of the offered services.
PrimeWater leverages the plethora of available EO sensors for improving inland water quality monitoring capabilities as well as for boosting the performance of ecological models and detect critical changes in lake and reservoir ecosystems. PrimeWater investigates innovative algorithms for retrieving phytoplankton and aquatic vegetation to fully exploit the improved spectral, spatial and temporal capabilities of hyperspectral images from PRISMA, DESIS and HySpex. Various algorithms are under evaluation based on wavelength-dependent peaks/dips, bio-optical modelling inversion and machine learning techniques.
PrimeWater contributes also towards operational EO data-model infusion techniques that allow for improved hydrological and water quality forecasting. In hydrological forecasting the Ensemble Kalman Filter assimilation technique is used to investigate the potential of EO data in improving seasonal forecasting skill. In process-based water quality modelling, advanced sequential and variational assimilation techniques are tested for integrating in-situ end EO-based water quality products into reservoir modelling to efficiently capture the variability of algae-bloom events. Data-driven models are also employed to develop credible models for the prediction of phytoplankton dynamics with results so far indicating their superiority compared to naïve forecasting alternatives.
PrimeWater aspires to impact research in the field of remote sensing and hydro-ecological modelling as well as the innovation in the water sector and downstream sector of EO market. PrimeWater strengthens the research capacity across remote sensing and water resources modelling by offering new algorithms for retrieval of phytoplankton, suspended sediment, CDOM etc. from hyperspectral imagery and new techniques for down streaming EO-based products, in process based, data driven and hybrid modelling. An operational platform that builds on the results of PrimeWater’s scientific experiments, integrates the different EO and modelling components and enhances the adaptive capacity of water-dependent industries. The operational platform offers monitoring, forecasting, and early warning services as well as customized management tools for the 4 case-studies of PrimeWater, reducing the vulnerability of local communities against algae blooms and water shortage events.
PrimeWater contributes also towards operational EO data-model infusion techniques that allow for improved hydrological and water quality forecasting. In hydrological forecasting the Ensemble Kalman Filter assimilation technique is used to investigate the potential of EO data in improving seasonal forecasting skill. In process-based water quality modelling, advanced sequential and variational assimilation techniques are tested for integrating in-situ end EO-based water quality products into reservoir modelling to efficiently capture the variability of algae-bloom events. Data-driven models are also employed to develop credible models for the prediction of phytoplankton dynamics with results so far indicating their superiority compared to naïve forecasting alternatives.
PrimeWater aspires to impact research in the field of remote sensing and hydro-ecological modelling as well as the innovation in the water sector and downstream sector of EO market. PrimeWater strengthens the research capacity across remote sensing and water resources modelling by offering new algorithms for retrieval of phytoplankton, suspended sediment, CDOM etc. from hyperspectral imagery and new techniques for down streaming EO-based products, in process based, data driven and hybrid modelling. An operational platform that builds on the results of PrimeWater’s scientific experiments, integrates the different EO and modelling components and enhances the adaptive capacity of water-dependent industries. The operational platform offers monitoring, forecasting, and early warning services as well as customized management tools for the 4 case-studies of PrimeWater, reducing the vulnerability of local communities against algae blooms and water shortage events.