Innovation dynamics in social and technological systems are strongly linked to urban systems and their multi-scale properties. Understanding underlying processes is crucial for sustainable territorial planning. We introduce a multi-scalar model for innovation dynamics in systems of cities, coupling a macroscopic innovation diffusion and urban dynamics model with mesoscopic models for local innovation clusters. The model parameter space is explored, and we apply a bi-objective optimisation algorithm with objectives across scales. Implementing indicators for downward causation, we finally investigate with a diversity search algorithm the diverse regimes of emergence the model can produce. This suggests strong emergence is captured, confirming the relevance of multi-scale approaches to artificial societies and urban simulation.