Ecological Indicators (Aug 2024)
Assessing urban river landscape visual quality with extreme learning machines: A case study of the yellow river in ningxia hui autonomous region, china
Abstract
A picturesque on-water landscape can yield favorable outcomes in terms of improving urban quality and boosting the tourism sector. The vantage point from the water offers a distinctive recreational experience and serves as a bridge connecting the urban environment with nature. Nevertheless, existing research pertaining to landscape management and utilization strategies has predominantly centered on the perspective from the riverbanks, with relatively few studies dedicated to assessing visual quality from the unique viewpoint of being on the water. In this research, a system for measuring landscape characteristics numerically was introduced, employing a deep convolutional generative adversarial network (DCGAN) for semantic segmentation, to assess human visual perception. An extreme learning machine model was employed to investigate the non-linear relationships between quantitative metrics and public ratings. This approach led to the development of an effective testing and forecasting model for evaluating the aesthetic appeal of the waterfront landscape in an urban river. Our research was centered on a case study of the Yellow River in Ningxia Hui Autonomous Region, China. The outcomes we obtained indicate that the approach we introduced yielded the strong predictive precision allowed us to establish a hierarchy of the importance of different influencing factors. Additionally, we found that the aesthetic appeal of the waterfront landscape can be significantly impacted by factors such as the degree of urban construction, the destruction level index, the visibility of hard revetments, and the index of green clarity. It’s worth noting that, except for the index of green clarity, the other three factors exhibited a negative correlation with visual quality. Moreover, our suggested method provides a highly efficient way to evaluate how on-water landscapes are visually perceived and their overall quality. It also holds the potential to serve as a valuable method for evaluating upcoming waterscapes from a fresh and innovative perspective.
