Mapping of urban structure types (UST) is widely used for assessing their ecological situation in the frame of urban planning. Quantitative characterization of UST's requires detailed inventories of urban surface materials. In the past, this has mostly been achieved by cost-and time-intensive manual mapping approaches. Recent advances in hyperspectral remote sensing allow automated identification of urban surface materials. In this paper, the potential of hyperspectral remote sensing for automated characterization of USTs is investigated for a study area in Munich, Germany. First, a systematic spatial inventory of urban surface materials has been automatically derived from hyperspectral HyMap data. Second, the resulting 42 spectrally distinct surface materials in combination with height information have been used for calculation of spatial indicators quantifying the urban land cover. These buildingblock related indicator values have been further analyzed regarding their quality and and suitability for statistical characterization of UST's. Accuracy assessment against a reference dataset has shown high reliability of the obtained indicator values with average deviations of less than 11 \% for most of the UST's. Additionally, these results have been compared to indicators estimated by the municipality of Munich. Differences have mainly been attributed to the underlying mapping principles and to some extent caused by real land cover changes. The results show the great potential of the presented hyperspectral remote sensing based approach for effective and up-to-date derivation of quantitative urban land cover indicators which can be used in urban planning practice.