From heat resilience to sustainable co-benefits: Adaptive urban morphology generation based on multimodal data fusion and a novel generative framework

Abstract

Rapid urbanization and global climate change have intensified the Urban Heat Island (UHI) effect. However, practical implementation is often constrained by limitations in data availability and computational capacity, overlooking the influence of socioeconomic factors and spatial heterogeneity. This study proposed an end-to-end urban 3D morphology generation framework that leveraged multimodal datasets, including Local Climate Zones (LCZ), Land Surface Temperature (LST), and Population Density (POPH) through a novel CycleGAN-Pix2pix (CP-GAN) model chain. Using six representative LCZ areas in Guangzhou as case studies, the research evaluated the Urban Morphology Indicators (UMI), Land Use and Land Cover Change (LUCC), and Points of Interest (POI) across various responsive generation scenarios to identify urban morphologies that balanced cooling effects with socioeconomic and ecological benefits. The results showed that:(1) The CP-GAN achieved robust performance in urban morphology generation, demonstrating stable convergence and high precision, with an average structural similarity index exceeding 0.811, along with high signal-to-noise ratios and low error metrics. (2) Rising temperatures reshaped urban morphology, with every 3°C increase reducing green space by 5.47% while raising commercial activity and impervious surfaces by 2.38% and 2.84%, respectively; (3) Population density drove POI clustering but exhibited weaker morphological control than temperature gradients. (4) LCZ4, LCZ5, and LCZ6 exhibited spatial heterogeneity in UMI, LUCC, and POI responses to temperature and population density variations, necessitating LCZ-specific adaptive strategies. This generative system offers fine-grained 3D morphological solutions to mitigate UHI effects while establishing a transformative framework for sustainable urban development.