Achieving climate neutrality in urban areas requires tools capable of representing the interaction between the building envelope, energy systems, and the local microclimate. Although microclimatic phenomena, such as the urban heat island (UHI), are recognized for their influence on cooling loads and system performance, their explicit representation remains limited in energy simulation applications. The Urban Weather Generator (UWG), a model based on a Concentrated Parameter (CP) network, represents a computationally efficient alternative to high-fidelity numerical models, such as CFD simulations, for urban climate analysis. However, its ability to represent spatial heterogeneity remains a critical issue. This thesis analyzes the impact of spatial discretization on model accuracy, comparing a regular 200x200 meter grid-based subdivision of the domain with a discretization based on morphological blocks.