In Luxembourg, economic and demographic growth is very robust and is among the strongest of the member countries of the European Union. This growth has resulted in significant building needs, with higher and higher units responding to increased human activities (eg, residential, industrial and commercial). Due to limited land resources for construction and the need to preserve natural and agricultural areas, densification seems to be an "intelligent solution" to support sustainable urban growth. Urban planners (and other urban actors) need information and monitoring tools to better  manage the development of building stock and building heights and to improve various aspects of urban planning (e.g., water supply, energy efficiency, services and green spaces).

In this context, the Luxembourg Institute of Socio Ecomomic-Reasearch (LISER) recently conducted a study to measure automatically the building heights in Luxembourg-City using LiDAR (Light Detection And Ranging¹, ²) data. A first prototype tool named "Lux-SmartGrowth" was developed based on a high resolution 'normalised' digital surface model (nDSM).

It was created by performing the following operations:

1. Subtracting the influence of the topography on the object height (i.e., the difference between the digital surface model (DSM) and the digital terrain model (DTM) (nDSM= DSM-DTM).
2. Using a normalised difference vegetation index (also called NDVI) to keep only objects without vegetation.
3. Combining the new data (i.e., nDSM) with cadastral administrative data to identify the types of buildings³.

The outcome, resulting from the combination of these data, is a layer of heights by type of buildings. This first study has shown that the tool has the potential to complement the administrative data of the vertical cadastre by creating an automated 3D cadastre of all buildings with accurate estimated building heights compared to reference data (i.e., ground truth).

The use of LiDAR data is complex, and its application raises many challenges, such as: automated feature detection and extraction, handling large amounts of data, LiDAR data classification, and automatic error correction in LiDAR data (e.g., tower crane on a construction site, car in a street, etc.).