Street space form is an important factor affecting urban air temperature and humidity, solar radiation and thermal comfort. Sunshine duration is not only the most intuitive expression of solar radiation intensity, but also one of the important indicators for characterizing climate change. Quantification of sunshine at street level can be of great significance for optimizing solar energy utilization, regulating human psychological state, improving urban microclimate environment.
However, due to the inadequate accuracy, manual processing and time-consuming of the existing measurement methods, and simulation software has many problems, such as long time-consuming, low accuracy, difficult to simulate street elements of tree canopy and so on,
In this paper, a new big data method of calculating sunshine duration of urban street space by using street view photographs is proposed.
The panorama is semantically segmented into fisheye image composed of three elements: sky, buildings and trees. And the sunshine hours are estimated by matching the fisheye image with the corresponding solar path.
Taking the Gulou Campus of Nanjing University and its surrounding areas as research object, the distribution of daily average sunshine hours in summer is measured, and the influence of various building types and tree elements on it is quantitatively analyzed.
It is proved that the method has high recognition accuracy, short time-consuming and full-process batch processing. It can easily calculate the sunshine hours at the street level, which is important for optimizing solar energy utilization, regulating human psychological state, and improving urban microclimate environment.