Norway Building Heights Import

From OpenStreetMap Wiki
Jump to navigation Jump to search

Goal

The goal of this import is to augment existing 2d building outlines with estimated height information.

Context

Norway currently contains over 800 000 buildings, but very few contains any height information. This import aims to augment the existing collection of building outlines in Norway with height estimated from Digital Terrain Models (DTM) and Digital Surface Models (DSM) which has been made available for OpenStreetMap. These datasets cmost of Norway, and will enable more informative 3d rendering of all buildings in Norway.

A similar import was completed in Nice, France in 2017.

Schedule

The import will start in the Autumn of 2019, with no defined end date. The data files will be updated periodically as new elevation data becomes available.


Data source

The dataset is derived from data available at hoydedata.no.

Permission: https://lists.nuug.no/pipermail/kart/2014-August/004831.html


Data preperation

The OSM files have been generated by measuring the maximum difference between the DSM and DTM within the building outline. The full script used to generate these files are available here:

https://github.com/geimas5/DTMDEMtoBuildingHeight/blob/master/BuildingHeightGenerator.py


The script downloads the following ways:

  • building
  • building:part
  • man_made=silo
  • man_made=chimney
  • man_made=communications_tower
  • man_made=lighthouse
  • man_made=storage_tank
  • man_made=tower
  • man_made=works


It then calculates the highest point above the ground within the area formed by the way, and places a node on the highest point. If there are multiple points of equal height, it tries to select the node closest to the center of the area.


OSM Data Files

OSM datafiles are available at: https://drive.google.com/open?id=1s2HKON6IlbVJwngPz-Agb9YQcgJuzmim


The height files are divided into 25 regions. The area covered by each region can be seen here: https://hoydedata.no/LaserInnsyn/Images/NHM1_gruppering_33.png

The following table lists which data date was used to generate the height estimates, and the date the updated height data was generated. The generated date will correspond to the OSM data date.

Region DTM date DSM date Generated Notes
10-10 2018-12-21 2018-12-28 2019-07-28
11-10 2019-05-17 2019-05-17 2019-07-27 Nodes not at highest point, only contains building and building:part ways
12-10 2019-02-15 2019-02-15 2019-07-28
10-11 2019-05-24 2019-05-24 Planned
11-11 2019-05-24 Planned
1213-11 2019-02-15 2019-02-15 2019-07-28
10-12
11-13
1213-13
11-14
12-14
13-14
1213-15
1214-16
13-1718
14-17
14-1819
15-1718
14-17
14-1819
15-1718
16-18
15-19
16-1920
1718-1820

The zip-file for a region contains a number of files with each file named as follows:

  • Main files: 33-x-y.osm
  • Difference file: 33-x-y_difference.osm


The 33 refers to how the source data is encoded in UTM 33N, and the x and y corresponds to a grid within the UTM 33N coordinate system. The last part indicates weather the file is a main file, which contains all height estimates, or a difference file, which only contains nodes for which existing height information is either missing, not correctly encoded, or deviates from the estimate by more than 10 cm. The difference file will be useful scope the work, once a larger quantity of buildings becomes augmented with height data.

Import workflow

For simple buildings, traced with high accuracy from NIB imagery, without trees or other buildings close by, the height information can be safely used. But there are a number of situations where more care is necessary:

  • The tallest point of the building is not contained within the building outline
  • The building outline contains trees which might be higher than the building itself.
  • The building is close to other buildings which might be taller.
  • The building is very small


If the tallest point of the building is not within the building outline, then the correct course of action is to improve the building outline, and wait until the next time the data files are updated.

Whether the next two cases have caused a problem for a specific building, it is to see where the node containing the height information is located. The script tries to place the node on the highest point of the building, and if the node is located on the neighboring building or the overhanging tree, the height estimate is unlikely to be correct. There is, unfortunately, no general work-around for these situations, except in the cases when it’s natural to map the highest point using a building: part. note: Because the height estimate is based on UTM1, which has a resolution of 1m, the node position can appear a small ways off from the highest point, even though there is no error.

The last case requires a judgement by the user of whether the height estimate is likely to be correct. It might be safer to not import the height of buildings which are both small and low for now.

Tagging Plans

On each created changeset there are 2 tags :

  • source: Estimation derived from data provided by Kartverket
  • comment: Import of height information for buildings in Norway.

Quality

The quality of the data have been manually verified by spot-checking many buildings, without finding significant deviation.