3D input data for noise studies (experimental) [Nederlandse versie]

Feedback Session on 12 June 2019

On the Wednesday 12th of June 2019 we are organising a feedback session at the offices of the Kadaster in Rotterdam (13:30-16:00). During this session we will present our approach and we will recieve your feedback for further development. You can register using this form.

What is 3D noise in the Netherlands? An introduction

The past 2 years, we have investigated to automatically reconstruct 3D noise data in a collaboration of Rijkswaterstaat, RIVM, Kadaster and the 3D geoinformation research group from TU Delft. In this collaboration, we have investigated how 3D data on noise sources and the environment, as required in legally prescribed noise studies, can be automatically generated for the whole of the Netherlands from existing data such as the Key Register Addresses and Buildings (BAG), the Basic Register for Large-Scale Topography (BGT) and the airborne LiDAR point cloud covering the whole of The Netherlands (AHN).

This is an ongoing project that was started in 2017. A more detailed project description can be found here.

On this site we publish an example dataset that is generated using the current 0.2 version of our method. With this sample dataset interested parties have the possibility to review our results and send us feedback.

Description test data version 0.2

Our method aims to achieve high detail and accuracy, while keeping the resulting files small and adhering to the requirements and limitations of the currently available noise simulation software on the market.

With version 0.2 we deliver three input layers for 3D noise studies, namely

  1. building models (gebouwen),
  2. ground types with noise reflection/absorption factors (bodemvlakken), and
  3. terrain (hoogtelijnen).

These three input layers were generated fully automatically from the public BAG, BGT and AHN3 datasets and are explained in more detail below. We also investigated noise barriers (geluidsschermen) and bridges (bruggen), however these are not part of the v0.2 example dataset.

The study area of the sample dataset spans the 37ez2, 37fz1, 37gn2, and 37hn1 AHN tiles nearby the city of Rotterdam as illustrated below. We provide this test data in order to collect feedback that we will use for future developments where we plan to upscale the results to the whole of the Netherlands. For version 0.2 we have made initial choices regarding the simplification of the data , height-differences between connecting building parts, minimimum sizes of details etc. Based on the feedback we can adjust these parameters in a next release.

Sample area v0.2

The input data can directly be used as input in software that implements the Standaard Rekenmethode II van het RMG2012 (SRM2) (i.e. the Dutch legally prescribed calculation method), like GeoMilieu and WinHavik.

Building models

For the buildings, the BAG is used as basis. The heights of buildings have been calculated using the AHN-point cloud and the software [3dfier]{https://github.com/tudelft3d/3dfier}. This software converts 2D footprints into 3D block models. The heights of buildings are calculated as the 95-percentile of all AHN-points that fall within one building polygon. This is a good estimation of the roof ridge both for flat roofs and slanted roofs. Outliers such as points that fall on chimneys will be filtered out. As additonal information, we also provide the 75-percentile.

In the current noise simulation practice each building, regardless of its roof shape, is modelled with a single height level. The resulting block-shaped building representation is called LoD1.0. Modelling a building with only a single height can lead to large errors in the modelled height in case the building in reality consists of different parts that each have a very different height. Therefore, we have investigated how to automatically create building models in which multiple height levels are possible, i.e. using the LoD1.3 representation.

We offer three alternative data sets with building models, LoD1.0, LoD1.3 and LoD1.3 experimental (see graphic below).

  1. In the LoD1.0 data set each building is modelled with a single height value.
  2. In the LoD1.3 data set the buildings with only horizontal roofparts are modelled with multiple heights. To achieve this the footprint is subdivided into several roofparts and each roofpart is assigned its own height. Buildings with slanted roofs are modelled with a single height.
  3. In the LoD1.3 experimental dataset, also the buildings with slanted roofparts are modelled with multiple heights. These buildings are the most difficult to model in LoD1.3, and these results are merely indicative of the current status of our LoD1.3 building reconstruction method. We intend to further improve this method in the near future.

The height of each roofpart is computed by taking a percentile of the elevations points it contains. We offer both a 75th percentile and a 95th percentile variant.

Sample area v0.2

More details about our LoD1.3 reconstruction method can be found in the following slideshow.

Heights of the terrain

For the representation of the height, we use the BGT and AHN as basis. The resulting 3D lines that do not significantly contribute to the terrain have been removed. We also added height lines in case a height-difference was detected in AHN but not represented by the BGT information.

Groundtypes with noise characteristcs

Also for modelling noise-reflection and -absorption values, we use the BGT as basis. Neighbouring polygons with same values have been aggregated, small areas have been eliminated and boundaries have been simplified.

BGT klasse en fysiekVoorkomen attribuutwaarde Akoestische classificatie
OndersteunendWaterdeel (alles) absorberend
OnbegroeidTerreindeel (erf, gesloten verharding, open verharding, half verhard reflecterend
OnbegroeidTerreindeel (onverhard, zand) absorberend
BegroeidTerreindeel absorberend
Pand (alles) reflecterend
Scheiding (alles) reflecterend
Kunstwerkdeel (alles) reflecterend
OverigBouwwerk (alles) reflecterend
Overbruggingsdeel (alles) reflecterend
Wegdeel (anders dan ruiterpad en onverhard) reflecterend
Wegdeel (ruiterpad, onverhard) absorberend
OndersteunendWegdeel (verkeerseiland, gesloten verharding, open verharding, half verhard) reflecterend
OndersteunendWegdeel (berm, onverhard, groenvoorziening) absorberend
Waterdeel (alles) reflecterend


The test data is available as ESRI shape.

The following source data was used:

For the sample area we prepared the following data sets.

Feature Explanation File name Download
ground type Ground types classified by their noise reflecting property. Objects that are smaller than 6 square meters are merged into their neighbour and obtain the neighbour's reflectance property. <tile id>_bodemvlakken [ESRI Shapefile]
Terrain modelled with 3D lines (not contour lines and not breaklines) <tile id>_hoogtelijnen [ESRI Shapefile]
building in LoD1.0 Building footprints with a single height value per building. The height of the building model is computed as the 75th and 95th percentile of the points that are part of the roof. <tile id>_lod10_<percentile> [ESRI Shapefile]
building in LoD1.3 Building footprints with a single height value per building-part. The height of the building model is computed as the 75th and 95th percentile of the points that are part of the roof. If the building has slanted roof surfaces (dak_type is 2), then it is reconstructed in LoD1.0. <tile id>_lod13_<percentile> [ESRI Shapefile]
building in LoD1.3 (experimental version) Bulding footprints with a single height value per building-part. The height of the building model is computed as the 75th and 95th percentile of the points that are part of the roof. If the building has slanted roof surfaces (dak_type is 2), then it is reconstructed in LoD1.3. <tile id>_lod13_<percentile>_experimenteel [ESRI Shapefile]


The table below describes the attributes of the buildings and ground types data sets. The lines in the terrain data set do not have attributes.

Feature Attribute Explanation
building bag_id the identificatie attribute from the BAG
dak_type type of the roof of building
2 – roof with at least one slanted surface
1 – roof with multiple, only horizontal surfaces
0 – roof with a single horizontal surface
-1 – no AHN point was found for the building
-2 – could not detect a roof surface, even though AHN points were found
hoogte_abs hoogte absolute or absolute height of the building (height measure from NAP)
maaiveld_h maaiveld hoogte or absolute ground height of the building
maaiveld_p aantal maaiveld punten or number of AHN points that were used for calculating the ground height. Note that a value below 3 might indicate an unreliable value for maaiveld_h
ahn_geldig or valid height
1 – building was built before the point cloud was collected for the tile
0 – building was built after the point cloud was collected for the area
ahn_datum the date of the acquisition of the point cloud for the tile
ground type uuid unique ID of the object
bodemfactor sound reflectance property of the ground
0 – reflecting
1 – non-reflecting
terrain - -

Feedback Form

In case of questions or comments about the data please fill out our feedback form.


The 3D noise data version 0.2 is only made available for testing purposes. No rights can be obtained from the data. None of the parties involved can be hold responsible for damage that is caused by using the data.

Project partners

RWS logo
RIVM logo
Kadaster logo
IPO logo
ERC logo