Reddit reviews Agent-Based Models of Geographical Systems

I believe that the most of city simulation games rely on two simulation techniques or their combinations:

• cellular automata; usually fast, easy to implement, and pretty realistic to simulate spatial phenomena; they're good fit to describe rules of land use and zone development on a grid-like map.
  
• multi-agent systems; generally more powerful, can embed non-local interactions and non-spatial phenomena; they're good fit to model social, economics and transport phenomena; or where memory effects (adaptive or learning systems) are important.
  
  Quick web search gave me a couple of books which might be relevant to learn both approaches:
  
  

• Agent-Based Models of Geographical Systems (2011) $141 @ amazon, $0 @ libgen; there's also a chapter on cellular automata in urban spatial modelling.
  
• Cities and Complexity: Understanding Cities with Cellular Automata, Agent-Based Models, and Fractals (2007) $37 @ amazon