Mapping urban areas poses a significant challenge in Earth Observation. Accurate surface information is crucial for various urban planning and climate studies. The diverse use of artificial materials and mixed pixels complicates spectral analysis for multispectral and hyperspectral sensors, requiring higher resolution for identification. Moreover, existing hyperspectral libraries using spectroradiometers, lack the necessary variability required for effective Machine Learning model training. To address these gaps, the urbisphere urban hyperspectral library is under development, leveraging the spectroradiometer Spectral Evolution RS-3500. The library enables satellite image classification without the need for time-consuming on-site data collection. This study showcases the library’s potential to classify Sentinel-2 satellite images, utilizing an image from the city of Heraklion as a case study. The resulting material map was validated using a randomly generated independent dataset.