Assessing the perception of urban visual quality: an approach integrating big data and geostatistical techniques
From Firenze University Press Journal: Aestimum
Veronica Alampi Sottini, Department of Agriculture, Food, Environment and Forestry (DAGRI), University of Florence
Elena Barbierato, Department of Agriculture, Food, Environment and Forestry (DAGRI), University of Florence
Irene Capecchi, Department of Agriculture, Food, Environment and Forestry (DAGRI), University of Florence
Tommaso Borghini, Department of Architecture (DIDA), University of Florence
Claudio Saragosa, Department of Architecture (DIDA), University of Florence
It is well known that people live well in environments that they recognize and perceive as pleasant, comfortable, and safe. Human well-being is influenced by the physical characteristics of the surrounding urban space and how they aggregate with each other (Alexander et al., 1977; Lynch, 1960). European cities are generally built in different periods with distinctive architectural styles. The visual quality of the urban space for each era and for each zone of the city is influenced by different variables. In the various zones of the city, the visual quality of urban spaces can be explained by geographical and morphological macro-elements, such as coastlines, waterways or hills. These characteristics influence the visual quality of urban spaces in limited areas and not of the whole urbanized area.According to Radovic (2003) the physical structure of the city implies “a complex set of built elements, space and environment, units and assemblages, which united and connected in an integrated urban system, create the atmosphere and environment for the complex processing of urban life”.
Therefore, visual perception, understood as the subjective presentation of objective reality, has always been a complex and highly sensitive issue in the architectural and urban design process. Resources and visual effects play a dominant role in the identification of cultural, socio-economic, identity and communal values of the built environment, as the value and meaning of the built space is manifested predominantly through the subjective view of that space (Perovic & Folic, 2012).Many studies have focused on researching visual perception using photo-graphic data shared on social platforms (Alampi Sottini et al., 2018; Dunkel, 2015; Quercia et al., 2014; Zhou et al., 2015), others have used indicators to obtain information on urban visual quality using panoramic images from the Google Street View (GSW) web service (Yin and Wang, 2016), but, at the urban level, there are no studies that correlate perceptions of visual quality detected by social media platforms with spatial geographic characteristics through geostatistical models.Therefore, this paper proposes a geostatistical approach using Geographical Random Forest regression on the Tuscan city of Livorno. this has been analysed city because allows us to assess the visual quality of urban space in very diverse geographical areas. In fact, despite its relatively small size, the city of Livorno consists of a rather heterogeneous mosaic of neighbourhoods with peculiar characteristics due to different construction periods. For this reason, it is an appropri-ate study area to test a first version of the model to assess the visual quality of urban spaces.
The proposed methodological approach consists of 3 macro-phases: the first one aims at obtaining the indices that compose the urban visual quality perceived by users using photos shared on Flickr; the second one involves calculating the indicators that constitute the urban visual quality using data from both Google Street View, LiDAR data and geographic data; the last one consists in applying two geostatistical models a global random forest and a geographic random forest to differentiate the results for each neighbourhood of the city.The objective of the study is to test the proposed methodological approach by understanding its strengths and weaknesses and to understand what methodological aspects are needed for the spatial component in the regression models used. The final goal is to provide useful information not only to researchers but also to public and private sectors to develop projects, standards and guidelines to improve the visual quality of urban design in cities.
DOI: https://doi.org/10.36253/aestim-12093
Read Full Text: https://oaj.fupress.net/index.php/ceset/article/view/12093
