The Bellinge data set: open data and models for community-wide urban drainage systems research

Abstract

Abstract. This paper describes a comprehensive and unique open-access data set for research within hydrological and hydraulic modelling of urban drainage systems. The data come from a mainly combined urban drainage system covering a 1.7 km2 area in the town of Bellinge, a suburb of the city of Odense, Denmark. The data set consists of up to 10 years of observations (2010–2020) from 13 level meters, 1 flow meter, 1 position sensor and 4 power sensors in the system, along with rainfall data from three rain gauges and two weather radars (X- and C-band), and meteorological data from a nearby weather station. The system characteristics of the urban drainage system (information about manholes, pipes, etc.) can be found in the data set along with characteristics of the surface area (contour lines, surface description, etc.). Two detailed hydrodynamic, distributed urban drainage models of the system are provided in the software systems MIKE URBAN and EPA Storm Water Management Model (SWMM). The two simulation models generally show similar responses, but systematic differences are present since the models have not been calibrated. With this data set we provide a useful case that will enable independent testing and replication of results from future scientific developments and innovation within urban hydrology and urban drainage systems research. The data set can be downloaded from https://doi.org/10.11583/DTU.c.5029124 (Pedersen et al., 2021a).

Exploring the Performance of Ensemble Smoothers to Calibrate Urban Drainage Models

“…For instance, at the calibration stage for rainfall event 4 as shown in Figure 13c, the deviations of the peak water depth at site 2 estimated by ES‐MDA and ES‐ILU (0.05 and 0.07 m respectively) are much lower than those obtained by Pedersen et al. (2021) (0.55 m). For the validation of the model predictions for rainfall event 3 at site 2 as shown in Figure 12c, the peak times estimated by ES‐MDA and ES‐ILU are 4 and 3 min earlier than the observations, respectively, while it is 15 min earlier for the estimation by Pedersen et al.…”

“…Specifically, for the real‐world case, both the ES‐MDA and ES‐ILU methods provide better calibration results than the best‐known solution manually obtained by Pedersen et al. (2021). It is found that a good overall fit to the observed data does not necessarily guarantee the good accuracy of estimations for all model states.…”

“…For the validation of the model predictions for rainfall event 3 at site 2 as shown in Figure 12c, the peak times estimated by ES‐MDA and ES‐ILU are 4 and 3 min earlier than the observations, respectively, while it is 15 min earlier for the estimation by Pedersen et al. (2021). This apparent advantage of ES‐MDA and ES‐ILU can be useful for state prediction and real‐time control in practical applications.…”

“…Case 1 (Figure 2a) is taken from the first illustrative example in the SWMM software manual that consists of 8 sub‐catchments, 14 nodes and 13 linkes as shown in Table 1. Both Cases 2 and 3 originate from an open‐access SWMM model of the real urban drainage system in Bellinge, Denmark (Figure 2c, Pedersen et al., 2021), which is termed as the Bellinge model in this paper. Case 2 (Figure 2b) is taken from a part of the Bellinge model (i.e., the part circled by the dot dash line in Figure 2c), where a free outfall has been added to this model as the terminal.…”

“…In this study, five real rainfall events provided in Pedersen et al. (2021) are used for model inputs. Details of these rainfall events are presented in Figure 3 and Table 2.…”