Researchers have developed a genetic algorithm that can predict a city’s vertical growth using economic and historical construction data.
A genetic algorithm is a program inspired by natural selection, the guiding force behind evolution. Much like evolution under natural selection, the evolution and development of cities is driven by a self-organizing process, where order occurs not as a result of being imposed from the top down, but by interaction between local forces. But, while genetic algorithms had previously been used to predict the development of self-organized systems, they had never been used to predict city growth. Ivan Pazos and his team at Spain’s University of A Coruña wanted to change that.
The team looked at Tokyo’s Minato Ward, an area of the city notable for its rapid vertical growth. The researchers started by developing a 3D model of the ward’s current skyline based on photogrammetric data, which included the terrain level as well as any buildings over 15m (49ft). From that, they developed a gradient map of the area to determine the factors that had been used to predict skyscraper growth in the past, in order to use them as inputs into their algorithm. They found that building height was predicted by vertical density, accessibility via public transit, public vs. private land ownership, and financial and real estate trends. Using that algorithm alongside current data on these same factors, they developed a probabilistic allocation map that shows where skyscrapers were likely to be built between 2016 and 2019.
Afterward, the map was compared with actual construction data of high-rise buildings that were planned to be completed in Minato Ward by 2019. The algorithm correctly predicted the total number of buildings and their locations in the city, although it wasn’t accurate about the exact year in which the buildings were constructed, or their height.
Speaking to Science Daily, lead researcher Ivan Pazos said, “The final conclusion of the study is that evolutionary computation seems to be able to find growth patterns that are not obvious in complex urban systems, and by means of its subsequent application, it serves the function of predicting possible scenarios for the evolution of cities.”
Going forward, the researchers want to go beyond building height to see if a genetic algorithm can model for other variables in a city and determine how it self-organizes. They hope that their model will be useful to city planners, enabling them to “test drive” the impacts of building regulations. They concluded their report: “By simulating possible scenarios of vertical urban growth, urban planners, policy makers, and designers will be able to better assess future changes in cities and anticipate the necessary responses for implementing new infrastructure or regulations.”