| dc.description.abstract | In this thesis, a 2D version of Hydrus-2D/3D was used to model runoff from a green and blue-grey test roof at Høvringen. The van Genuchten-Mualem model was used to analytically describe the hydraulic properties of the substrates. The required input parameters were not available prior to this thesis, but found from literature. The parameters were then calibrated against both water content measurements and runoff measurements during the first two weeks of July, before they were validated against runoff measurements. The validation period showed that the most accurate results were obtained with the calibration against runoff measurements. NSE coefficients for the blue-grey roof was generally above 0.75. This indicates that the performance of the model is acceptable. The same coefficients were generally around 0.6 for the green roof, which is also acceptable. However, the green roof was more sensitive for dry periods, where the NSE was negative for a period in August.
Runoff from the blue-grey roof was also modeled with an NSE slightly above 0.2 for both January and February. The calibrated and optimized parameters from the summer calibration was used. The green roof was modeled with an NSE of -6.0 or less in January and February. The inaccuracies during winter time are likely due to limitations in Hydrus to only model unfrozen soils and a simplified snow accumulation module. The blue-grey roof was modeled significantly more accurate than the green roof during winter time. The main reason for this is that the blue-grey substrate is thicker and covered with pavers.
Overall, Hydrus was able to produce acceptable accuracy for both the leca material in the blue-grey roof and the green roof substrate during unfrozen conditions. Difficulties in modelling storage capacity and extremely dry or wet periods were found to be the disadvantages of the model. Laboratory results of the leca material were recieved late in the thesis, and could have provided better results for the blue-grey roof. | |
