Assessment of the flow predictions from a large urban drainage model forced by distinct spatial resolution weather forecasts

Abstract

This paper assesses calibration and verification results of the Alcântara catchment model (Lisbon) and of the flow model forecasts forced by two different resolutions in Weather Forecasting: the 9 km WRF model from the Windguru website, available with just one forecast point in the catchment area, and the 5 km WRF model of the University of Aveiro, with an orthogonal grid of six forecast points. The accuracy of urban drainage model results was assessed combining statistical and graphical techniques and is substantially affected by the uncertainty in spatial rainfall distribution. Therefore, data from more rain gauges or from satellite sources would certainly improve model parameterisation and results. This study also highlighted the importance of modelling rainfall-derived infiltration and inflow (RDII) for achieving more balanced and accurate results. In forecast mode, the sewer model is significantly more accurate when forced by the University of Aveiro model forecasts than when forced by the Windguru website forecasts. Differences in results are partially attributed to the number of forecast points considered in each model and on differences in spatial resolution of the models. The comparison between predicted and observed storms was sometimes difficult and subjective due to the time lags between them.