*2.3. Barcelona Semi-Distributed 1D*/*2D USM*

After the two first investigations concerning the development and calibration of a detailed 1D/2D USM in Barcelona, covering approximately half of the administrative land of the city (more than 50 km2) [20,26], a new SD model has been developed and calibrated in the framework of this work [2] and the new drainage master plan of the city (PDISBA, from the acronym in Spanish) [11,27].

The large amount of effort related to the analysis of the deficit of surface drainage systems through the last two drainage master plans of the city and the consequent implementation of thousands of inlets in all the urban catchments allow the assumption that stormwater could be quickly introduced into the sewer system by avoiding uncontrolled runoff circulation and aiming to develop an SD 1D/2D USM (referred to as 1D/2D USM in the following).

Moreover, the new 1D/2D USM presents two relevant improvements with respect to the previous ones: the model includes the main and secondary sewer network, reaching a total length of 1650 km of pipes, and covers the whole hydrological area of the city (administrative land and upstream surfaces), exceeding 120 km<sup>2</sup> of model domain.

The model, with more than 85,000 nodes and a discretized 2D domain in an unstructured mesh of more than 1,360,000 cells, was developed through the Infoworks ICM (Integrated Catchment Modeling) software (www.innovyze.com) [25] and was calibrated and validated using the historical data recorded by more than 100 flow depth gauges located in the city's sewage network and more than 20 rain gauges distributed in the analyzed domain using Thiessen polygons [11,27]. The average size of the 2D cells for overland flow modeling is in the range of 25–100 m2.

The model required high-quality topographic information (physical data from the network, digital terrain model 2 × 2 m with a resolution in height of approximately 15 cm) and phenomenological information (rainfall data and flow level for the calibration phase), in addition to an adequate hardware configuration to reduce computation time during numerical simulations [20].

The new 1D/2D USM allowed the estimation of flow variables (flow depth, flow velocity and flood extension) on the surface prone areas by several numerical simulations of historic events and synthetic storm hyetographs for current and future scenarios. These values were used for the flood hazard and intangible risk assessment (concerning pedestrian and vehicular circulation) and the evaluation of tangible direct and indirect impacts. The same outputs were also used to feed other integrated models of critical urban services to assess the cascading effects of floods in these sectors.
