Reducing the uncertainty in sewer system modelling
Climate change with its increasing precipitation could cause increased urban flooding in the future. The project COMCORDE investigates, whether microwave links from telecommunication networks can provide better rainfall information and whether better computer model allow for a better exploitation of our sewer infrastructure.
Engineers design and operate sewer systems to provide sanitation and flood protection for our cities. However, climate change will most likely increase volumes and variability of rainfall in the future. Consequently, there is an urgent need for mitigation measures and to better use our existing infrastructure. However, promising solutions, such as optimal operation through real-time control of drainage networks, or the integrated optimization of sewers, wastewater treatment plants and receiving waters are limited by two factors: poor rainfall data and the computational costs of complex simulation models.
To maintain current flood protection levels, we must first provide high-resolution rainfall data beyond the point estimates of traditional rain gauges and the inaccurate information from rainfall radars. Second, we need fast simulation models and novel techniques of uncertainty analysis to
increase the credibility of model predictions. The goals of our project are therefore to reduce the uncertainty in predicted wet weather flows with better rainfall information and to decrease the computational requirements of sewer flow prediction models. Specifically, we will i) use commercial micro
wave links (MWLs) from telecommunication networks as virtual rain gauges and ii) develop algorithms for data assimilation to predict probabilistic rainfall fields that combine various types of rainfall information. In parallel, we will investigate novel methods to iii) identify model structure deficits and quantify associated uncertainty and iv) develop efficient emulators of hydrodynamic sewer flow models that will be used, together with the probabilistic rainfall fields, to reduce the uncertainty of predicted sewer flows.
In the future, such efficient emulators, together with improved rainfall information based on MWLs, will allow for a better operation of urban drainage systems and for improved uncertainty assessment of the involved simulation models. Accurate rainfall data with a spatial resolution of a few hundred square meters and in time steps of few seconds or minutes would advance not only urban hydrology, but will allow for completely new types of analyses in a variety of fields from meteorology to insurances.