Keywords
Urban storm drainage, Green infrastructure, Simplified models, City-scale modelling
Start Date
15-9-2020 10:40 AM
End Date
15-9-2020 11:00 AM
Abstract
Smart Climate Hydropower Tool (SCHT, part of the H2020 project "CLARA - Climate forecast enabled knowledge service") is an innovative web-cloud-based climate service that makes use of a set of machine learning methods for supporting decision-making in a context of hydropower production. Although highly flexible and with low costs to power ratio, estimates of future hydropower production are strictly linked with the ability to forecast meteorological conditions. Even if tangible results using physically-based (e.g. Collischonn, et al., 2007; Fan, et al., 2015) or machine learning (e.g. Callegari, et al., 2015; De Gregorio et. al 2017) have been achieved, challenges remain for seasonal lead-times and rainfall dominated catchments. Here, we propose a hybrid forecast system by using a combination of physically-based seasonal forecasts (provided by state-of-art seasonal forecasts of meteorological data from the Copernicus Climate Data Store), with a set of different machine learning algorithms (support vector regression – SVD, Gaussian processes – GP, long short-term memory – LSTM, and recursive neural networks – RNN). We test the application of the different machine learning techniques for forecasting seasonal river discharges up to six months in advance for two catchments in Colombia, South America. Seasonal forecasts are performed by making use of available state-of-art seasonal forecasts of meteorological data provided by the Copernicus Climate Data Store (CDS). Each algorithm is trained over past decades datasets of recorded data, and forecast performances are validated and evaluated using separate test sets with reference to benchmarks (historical average of discharge values and simpler multiparametric regressions). Final results are then presented to the users through a user-friendly web interface. The web interface is the result of a tied connection with end-users in an effective co-design process, adding value to energy forecasts and ideally paving the road for highly scalability and replicability (e.g. development of similar services worldwide).
Smart Climate Hydropower Tool: A web-cloud-based climate service for supporting decision-making in hydropower production
Smart Climate Hydropower Tool (SCHT, part of the H2020 project "CLARA - Climate forecast enabled knowledge service") is an innovative web-cloud-based climate service that makes use of a set of machine learning methods for supporting decision-making in a context of hydropower production. Although highly flexible and with low costs to power ratio, estimates of future hydropower production are strictly linked with the ability to forecast meteorological conditions. Even if tangible results using physically-based (e.g. Collischonn, et al., 2007; Fan, et al., 2015) or machine learning (e.g. Callegari, et al., 2015; De Gregorio et. al 2017) have been achieved, challenges remain for seasonal lead-times and rainfall dominated catchments. Here, we propose a hybrid forecast system by using a combination of physically-based seasonal forecasts (provided by state-of-art seasonal forecasts of meteorological data from the Copernicus Climate Data Store), with a set of different machine learning algorithms (support vector regression – SVD, Gaussian processes – GP, long short-term memory – LSTM, and recursive neural networks – RNN). We test the application of the different machine learning techniques for forecasting seasonal river discharges up to six months in advance for two catchments in Colombia, South America. Seasonal forecasts are performed by making use of available state-of-art seasonal forecasts of meteorological data provided by the Copernicus Climate Data Store (CDS). Each algorithm is trained over past decades datasets of recorded data, and forecast performances are validated and evaluated using separate test sets with reference to benchmarks (historical average of discharge values and simpler multiparametric regressions). Final results are then presented to the users through a user-friendly web interface. The web interface is the result of a tied connection with end-users in an effective co-design process, adding value to energy forecasts and ideally paving the road for highly scalability and replicability (e.g. development of similar services worldwide).
Stream and Session
false