4. DISCUSSION
4.1 Influencing factors
Urban hydrological processes have altered with urban development, and thus distribution pattern of flooding points has changed (Chang et al., 2021; Zhao et al., 2014). In this study, we integrated a series of driving factors (i.e., topography, building metrics) and identified the major driving factors of urban flooding by adopting BRT method. The reasons for urban flooding are complex, including natural and anthropogenic factors. Through BRT analysis, land cover configuration has the greatest impact on flooding. The fragmentation of green spaces would increase occurrence of urban flooding events on a given land cover composition (Zhang et al., 2020). Higher fragmentation at landscape level might lead to serious urban flooding, and its role varied across different megacities in China (Li et al., 2022). In our study, it suggests that the optimal landscape shape index was nearly 2.8. The results were not consistent with previous studies, the difference was caused by the response variable selected, explanatory factors selected, analysis units, and analysis methods. Similarly, contribution of patch density was only 2.54%, which reflects the patch density was not important.
Through BRT analysis, land cover composition has important impact on flooding. Increasing the green space and waterbody could mitigate the flooding risks. It was also suggested that increase the impervious surfaces would be inevitably increase urban flooding risks. This is attributed to large difference between impervious surfaces and green-blue space (green space and waterbody). Specifically, it is difficult for waterflow to penetrate into the impervious surfaces. Additionally, less surface roughness of the impervious surface promotes rainwater flow and accumulation. On the contrary, urban green space can reduce runoff and regulate rainwater storage. This finding is consistent with case study in other cities (Zhang et al., 2020). Zhang et al. (2020) found that land composition plays a dominant role in determining urban floods with analysis scale range from 1 km to 5 km in metropolitan coastal cities of Guangzhou, China (Zhang et al., 2020).
Fewer studies explored the impact of the building metrics on urban flooding, and the importance of building has not yet been agreed. For example, Lin et al. (2022) found that 3D building metrics exert the greatest influence on urban flooding. Li et al. (2022) revealed 3D building metrics had slight impact on urban flooding. Our results suggested that 3D building morphology played an important role in urban flooding, and the contribution of 3D building morphology was equal to that of land cover composition. In fact, contributions of building congestion degree and building density reached at 6.35% and 4.81%, respectively. However, the current study also found that the importance of three-dimensional fractal and building height was relatively low? The possible reason was that green space is correlated with impervious surfaces, and impervious surfaces is highly correlated with building coverage ratio. Meanwhile, buildings with higher three-dimensional fractal generally occupy larger base area and taller building height. The green space, building height, and 3D fractal had occlusion effects on urban flooding and thereby the importance of building coverage ratio and building height became weaker.