EESS talk on "Towards educated decision-making on water resources in the face of uncertainty of climate projections"
Event details
| Date | 14.03.2023 |
| Hour | 12:15 › 13:15 |
| Speaker | Prof. Paolo Reggiani, Department of Civil Engineering, University of Siegen |
| Location | Online |
| Category | Conferences - Seminars |
| Event Language | English |
Abstract:
Ongoing global warming requires to periodically revisit and validate adaptation strategies in water resources planning and management to address resource scarcity challenges in the not-to-distant future. This concerns especially the agricultural sector, which in many parts of the world is heavily dependent on freshwater for irrigation, but can be extended to related areas like renewable energy. The main challenge in proposing mitigating actions lies in the uncertainty of climate projections, especially the non-stationarity of the change process. Continuing to assume stationarity or applying simple trend extrapolation to project current states into the future is not to be considered satisfactory and is therefore entrusted to Earth-system model ensembles, which are generated across international model inter-comparison projects (e.g. CMIP5 and CMIP6).
Decision-making in water resources planning can be supported by stochastic optimization approaches, like for instance stochastic-dynamic programming (SDP), which take uncertain temperature, precipitation or surface runoff signals as input to optimize the management of reservoirs or irrigation systems managed through a set operational rules and/or chosen best cropping practices. A sharp and calibrated specification of the uncertainty affecting climate forcing variables through a predictive probability density or approximated via appropriately derived ensembles constitutes the most important prerequisite for reaching optimal decisions. The ensemble spread provided by multiple raw Earth-system model outputs, which is often interpreted as representing true climate uncertainty, requires further processing to obtain a useful predictive density through Bayesian conditioning on observations, under the assumption of weakly stationary processes.
Through an application to rainfall and seasonal temperature over Northern Italy, we show the difference between the probability distributions derived directly from the projection ensembles and those derived after conditioning raw projections on observations. These distributions, directly or in the form of derived ensembles, will be the basis for rationally implementing any further decision-making process.
Short biography:
P. Reggiani holds of the Chair of Water Resources Management and Climate Impact Research at the University of Siegen, Germany. P. Reggiani graduated in Environmental Engineering at the University of Trento, Italy, in 1994. He completed his Ph.D. at the University of Western Australia. Between 1999 and 2000 P. Reggiani worked at CSIRO Land and Water, where he developed modeling approaches to assess dryland salinity in Australia. After his return to Europe in late 2000 he worked as “Marie Curie Fellow” at the soil science lab LTHE in Grenoble, France. In 2002 he took up an appointment as researcher and scientific consultant at the Dutch Institute Deltares. In 2014 he was appointed Full Professor at the University of Sieger. During his career, P. Reggiani worked in various applied scientific and consulting projects in Asia, Africa and Europe. Am important focal point of his scientific work has been the interface between numerical weather prediction and hydrological forecasting, including the area of forecasting uncertainty, as well as operational water management in data-poor areas. P. Reggiani has published more than 50 papers in different areas of hydrology, climate change impacts on water resources, flood forecasting, channel hydraulics and soil science. He also acted as coordinator for several projects, among which the EU FP5 R&D initiative “European Flood Forecasting System (EFFS)”. P. Reggiani partakes in several scientific boards on water resources issues. He is part of the standing Committee on Hydrological Services (Sercom-HYD) at the World Meteorological Organization.
Ongoing global warming requires to periodically revisit and validate adaptation strategies in water resources planning and management to address resource scarcity challenges in the not-to-distant future. This concerns especially the agricultural sector, which in many parts of the world is heavily dependent on freshwater for irrigation, but can be extended to related areas like renewable energy. The main challenge in proposing mitigating actions lies in the uncertainty of climate projections, especially the non-stationarity of the change process. Continuing to assume stationarity or applying simple trend extrapolation to project current states into the future is not to be considered satisfactory and is therefore entrusted to Earth-system model ensembles, which are generated across international model inter-comparison projects (e.g. CMIP5 and CMIP6).
Decision-making in water resources planning can be supported by stochastic optimization approaches, like for instance stochastic-dynamic programming (SDP), which take uncertain temperature, precipitation or surface runoff signals as input to optimize the management of reservoirs or irrigation systems managed through a set operational rules and/or chosen best cropping practices. A sharp and calibrated specification of the uncertainty affecting climate forcing variables through a predictive probability density or approximated via appropriately derived ensembles constitutes the most important prerequisite for reaching optimal decisions. The ensemble spread provided by multiple raw Earth-system model outputs, which is often interpreted as representing true climate uncertainty, requires further processing to obtain a useful predictive density through Bayesian conditioning on observations, under the assumption of weakly stationary processes.
Through an application to rainfall and seasonal temperature over Northern Italy, we show the difference between the probability distributions derived directly from the projection ensembles and those derived after conditioning raw projections on observations. These distributions, directly or in the form of derived ensembles, will be the basis for rationally implementing any further decision-making process.
Short biography:
P. Reggiani holds of the Chair of Water Resources Management and Climate Impact Research at the University of Siegen, Germany. P. Reggiani graduated in Environmental Engineering at the University of Trento, Italy, in 1994. He completed his Ph.D. at the University of Western Australia. Between 1999 and 2000 P. Reggiani worked at CSIRO Land and Water, where he developed modeling approaches to assess dryland salinity in Australia. After his return to Europe in late 2000 he worked as “Marie Curie Fellow” at the soil science lab LTHE in Grenoble, France. In 2002 he took up an appointment as researcher and scientific consultant at the Dutch Institute Deltares. In 2014 he was appointed Full Professor at the University of Sieger. During his career, P. Reggiani worked in various applied scientific and consulting projects in Asia, Africa and Europe. Am important focal point of his scientific work has been the interface between numerical weather prediction and hydrological forecasting, including the area of forecasting uncertainty, as well as operational water management in data-poor areas. P. Reggiani has published more than 50 papers in different areas of hydrology, climate change impacts on water resources, flood forecasting, channel hydraulics and soil science. He also acted as coordinator for several projects, among which the EU FP5 R&D initiative “European Flood Forecasting System (EFFS)”. P. Reggiani partakes in several scientific boards on water resources issues. He is part of the standing Committee on Hydrological Services (Sercom-HYD) at the World Meteorological Organization.
Practical information
- General public
- Free
- This event is internal
Organizer
- EESS - IIE
Contact
- Prof. Andrew Barry, ECOL