DySENSE: Dynamic Spatiotemporal ENergy Synthesis and Emulation with Satellite Imagery
DySENSE is a joint multimodal GenAI framework that synthesizes static urban satellite imagery together with spatially aligned, dynamic monthly building energy consumption maps. It decouples long-term urban morphology from short-term energy dynamics, allowing generation of new static urban imagery with aligned energy maps or dynamic monthly energy maps under climate prompts while keeping morphology fixed. Experiments across New York, Washington, Norfolk, and Busan reported high-fidelity synthetic energy data using less than 25% labelled data, with 8% mIoU and 8.7% accuracy improvements and 9%-18% CVRMSE reductions. The authors provide data and code links for urban energy-efficient planning, peak-load analysis, building simulation, and demand response.