A GPT approach for transcending decision-making in urban environment through the integration of multiple earth observation and in-situ data sources

Vast amount of Earth observation data undergoes daily archiving and tasking. Integrating Remote Sensing with In-Situ measurements, and historical data holds unprecedented potential for decision-making in urban environments. However, sourcing, processing, and utilising this data demand heightened computational complexities. This paper introduces a novel approach to seamlessly integrating multiple Earth Observation satellite imageries with various spectrums and resolutions with ground-based sensor data and historical records to train a Deep Learning model and facilitate it with a Generative Pre-Trained Transformer (GPT). This approach offers enhanced actionable insights for urban decision-making, disaster prevention, emission reduction, and more.