| dc.description.abstract | Green roofs (GR) are useful structures for runoff reduction in urban areas, since its layer design allows for rainfall water storage, which delays runoff start and reduces peak flow. GR are among the most known and monitored nature-based solutions in the world, although few studies have been able to explain their behavior through hydrological modeling. GR modeling allows the analysis of different scenarios, which can be tested and compared, aiming to increase the effectiveness of such studies. In this study, the first Chapter presents a review of the studies made in the last decade regarding green roofs hydrological modeling in order to establish this subject’s state of the art. GR modeling results depend mainly on the model used, on rainfall and climate characteristics and GR constructive settings. This review also found the employment of distributed models in GR simulation are rare and unexplored. Given the knowledge gap regarding distributed models for green roofs, it is presented, in Chapter II, a set of adaptations (Hidropixel-GR) from the original models who compose the distributed rainfall-runoff Hidropixel model – NRCS-TUH, NRCS-TUH+ and DLR – for green roof application. The adaptations are named GR1, GR2, GR3 and GR4. GR1 consists in the addition of a T0 parameter to the flow’s travel time, GR2 consists in the modification of NRCS-TUH+ routing methods by turning travel time into a calibrable parameter, GR3 consists in turning travel time into a calibrable parameter by modifying NRCS-TUH routing methods and GR4 consists in the addition of T0 parameter, originating from the DLR model. Observed rainfall-runoff data from green roofs whose surface area ranged from 1 to 12 m² was utilized in order to assess model quality in representing the hydrological processes occurring in green roofs. The models were validated by using the median value of calibrated parameters. The best performance in both calibration and validation was reached by the models GR4 (0,82 average NSE value during calibration, reaching 0,98 average NSE during validation), GR2 and GR3 (having similar results, they have reached average NSE of 0,71 during calibration and 0,87 during validation). Results obtained by these models were influenced mainly by the Curve-Number (CN) parameter, the auxiliary parameter for GR water storage (β) and travel time (Tt). This study has demonstrated the potential for simulating the hydrographs of linear and modular, scale-varying green roofs using Hidropixel-GR, a distributed rainfall-runoff hydrological model of simple parameterization. | |
