Discussion:
STURLA captured urban structure and characterized the physical property of ST in Philadelphia as previously done in NYC (Hamstead et al., 2016) and Berlin (Kremer et al., 2018), despite variation in size, demography, and historical planning. This suggests that urban areas may be subject to similar processes that result in between city-redundant spatial organizations (Votsis & Haavisto, 2019). Likewise, STURLA may be suited for understanding urban biogeography, environmental justice, and city planning for a sustainable future. Global analyses of cities may also identify clusters of urban areas that would benefit from similar management practices. Likewise, STURLA offers a computationally inexpensive alternative to network analyses of urban structure (Zhong et al., 2014).
One of the main limitations of STURLA classification is the binary nature of class assignment. If the STURLA grid were shifted it would change the relative proportions of the within class elements (e.g. trees decrease and other elements increase). Despite this variation, STURLA classes are a discrete countable number and have a Poisson distribution. Thus, the ranked order abundances of different STURLA classes should not vary in the most frequent classes. For example, since ‘tgpl’ is common in Philadelphia, a reduction in a large number of ‘tgpl’ classes in the city would be relatively less influential than additions/reductions of less common class.