Shpuza; Urban Shapes and Urban Grids: A Comparative Study of Adriatic and Ionian Coastal Cities
and Visual Fragmentation for actual shapes and shape hulls correlate strongly and significantly, Fconvex vs. Fvisual r=0.962, p<0.0001 and
Fconvex() vs. Fvisual measures are () r=0.949, p<0.0001, indicating that the two closely interrelated. Interestingly, the relative
fragmentation of shapes due to the depth of overlapping convex areas is almost identical to the relative fragmentation of shapes due to the visual depth of overlapping isovists. Since the sample includes urban shapes, it is not possible to conclude whether the two above correlations indicate a generic mathematical relation between the two fragmentation measures or whether they describe a specific property of urban form.
The study of office building floorplates demonstrated that the
correlation r=0.625, p<0.0001 between Drelative
of a property of the office building type since the two measures are not related mathematically (Shpuza 2006, Shpuza and Peponis forthcoming). In the case of urban shapes, the correlations between compactness and fragmentation are even weaker, Drelative vs. Fconvex r=0.567, p<0.0001; Drelative vs. Fvisual r=0.487, p=0.0003, showing that urban shapes exhibit less regularity than office floorplates (figure 6). However, these correlations improve noticeably when urban shape
hulls are considered: Drelative() vs. Fconvex() r=0.844, p<0.0001; Drelative
() vs. Fvisual
p<0.0001. The correlations between compactness and fragmentation are descriptive of a unique characteristic of urban form where shape hulls overcome geographical constraints and occupy a narrow sector in the spectrum of potential shapes. The extension of radial wings and the connection with peripheral orbital roads (Doxiadis 1968) are some of the dynamic processes that lead to a striking balance between compactness and fragmentation across urban shape hulls. It is possible to speculate that while cities grow, the dialogue between centrality and visibility (Hillier 1996) inside street networks causes urban shape hulls to maintain a tight relationship between compactness and fragmentation, hence expressed as a manifestation of urban emergence and essential urban dynamic (Hillier 1993, Hillier 1999).
Analysis of Urban Grids
Without exception, all cities in the sample have a denser core of street network located near the coast, which includes the historical center, and sparser area of larger urban blocks which are spread towards the hinterland. Urban grids include a combination of radial and orthogonal grids illustrating the urban global near-variant (Hillier 1999) and serpentine roads developed along hilly terrains (figures 2 and 4).
Linear maps of cities are analyzed according to the standard syntactic measures of Integration vii, Connectivity and Line Length as well as their statistical distributions Integration Skewness, Connectivity Skewness and Line Length Skewness (table 2). The Integration in the sample varies between 0.210 and 1.861 and has a mean of 0.800. The most notable correlations among axial measures for the sample are: Connectivity vs. Line Length r=0.758, p<0.0001; Connectivity Skewness vs. Length Skewness r=0.821, p<0.0001; Integration vs. Line Length r=0.664, p<0.0001; and Integration vs. Connectivity r=0.845, p<0.0001. In general, Integration in urban grids increases due to both the elongation of streets, i.e. greater Line Length, and the densification, i.e. greater Connectivity. At this point, it is unclear whether the increase of Connectivity and Length results from the effect of a few better connected and longer lines or whether it results from the combined effect of most of lines in urban grids. This will be clarified at a later part of the discussion.
Proceedings, 6th International Space Syntax Symposium, stanbul, 2007