ETHOS.BUILDA: Residential Building TABULA Archetype Dataset Germany

Introduction

This dataset contains all residential buildings in Germany with their construction year, size class, refurbishment state, and TABULA archetype. It is a partial dump of the ETHOS.BUILDA database (version v8_20240916). ETHOS.BUILDA is a database containing building-level data for the German building stock. It is based on various data sources that are combined and enriched with machine learning approaches to generate one consistent and complete building dataset. 

ETHOS.BUILDA is made available under the Open Database License (ODbL). The licenses of the contents of the database depend on the data source. The sources of the building attributes and information on the type of processing that was done to assign the information from the raw data to the building in ETHOS.BUILDA are provided for each individual data point.

Data structure and file overview

Building data is provided per federal state, the files are named according to the NUTS-1 region names. The building data has the following fields:

field name
description

ID
unique identifier of the building

position
location of building centroid in WKT-format, EPSG:3035

construction_year

value: construction year, 

source: source of the construction year data,

lineage: construction year assignment method

size_class

value: size class of the building, 

source: source of the size class data,

lineage: size class assignment method

refurbishment_state

value: refurbishment state of the building, 

source: source of the refurbishment state data,

lineage: refurbishment state assignment method

tabula_type

value: TABULA type of the building, 

source: source of the TABULA data,

lineage: TABULA type assignment method

A mapping of the abbreviations of "source" and "lineage" of individual data points to the descriptions is provided in sources.csv and lineages.csv. There is no entry for the source "v3_model.json", as it refers to the internally trained machine learning model for the respective attribute and not to an external data source.

The full footprint polygons from which the centroids are derived and the sources of the footprints are found in the related dataset linked as "is supplemented by".

Acknowledgements

This work was supported by the Helmholtz Association under the program "Energy System Design".