Sub-project: Gathering of data on building features and building classifiers through geo data analysis.
The aim of the project is to fill knowledge gaps regarding non-residential buildings to achieve the same level of data provision as for residential buildings in order to be able to conduct scenario analyses on the entire building sector.
Development of a scientifically validated and practical sample-based data gathering concept that permits the successful implementation of primary data collection on the non-residential building stock in order to answer the specified research questions. The careful selection of data samples is of particular importance as there exists no list or databank of the total research population. For this reason it is necessary beforehand to classify all building footprints in the databank and to seek out suitable contact partners for the building sector. The aim is to construct a databank of buildings and attributes describing various features (e.g. land use, building floor area and other morphological features, number of house coordinates, relationship to neighbouring buildings) as well as a sophisticated classification system of non-residential building types. Here it is necessary to classify all building footprints regarding the probability that these are relevant non-residential buildings.
For the first time, Germany’s stock of non-residential buildings will be the object of a systematic, scientific analysis in order to gather primary data. This will create a databank based on regularly updated official statistics that can be used to develop realistic scenarios on the development of the building stock, thereby helping to realise political goals of energy consumption and climate preservation. Furthermore, this will allow us to underline the relevance of the non-residential building sector to the wider economy up to a sufficient level of detail. Our knowledge will be expanded of the condition, development and investment behaviour in the non-residential building sector, allowing conclusions to be drawn on the motivations of actors in investment decisions and helping to make connections to legal, economic and business framework conditions. The project will help to answer Research Question 1 (knowledge generation from geo-base data) and Research Question 4 (application of the developed process in the provision of scientific services). The project is designed in such a way to offer long-term evaluation of results in the form of an IOER data monitoring.
In recent decades, several attempts have been made to close the knowledge gaps on the scope, structure and energy requirements of non-residential buildings. However, quantitative data on the volume and area of the stock were mainly determined for sub-sectors, using both secondary statistical analyses and isolated unsystematic surveys. However, the methodical quality of these individual surveys does not allow reliable conclusions to be drawn about the totality of all non-residential buildings in Germany. Rather, there is a risk that distorted extrapolation results may be used as a basis for political and legal frameworks due to systematic under-recording of relevant building stocks. The aim of the project is to close the knowledge gaps in non-residential buildings to the same extent as in residential buildings in order to be able to map the building sector as a whole in scenarios. For the first time, methods of sampling theory and geoinformatics were combined. At the beginning of the project, suitable basic geodata were researched and checked for availability. The technical data were examined with regard to the calibration obligations based on the state surveying laws of the federal states. The methods of collection and update rates in ALKIS (Amtliches Liegenschaftskatasterinformationssystem) were also queried. For the later screening and the broad-based survey, it was checked whether owner data of the sample objects could be retrieved from official data area-wide (this was not possible area-wide and was not considered a method). For the selection basis of the sample, the geodata of HU-DE, which describe buildings and building parts, proved to be suitable. 3D building models (LoD1) could also be used in addition to the data of HU-DE. Thus, the building use information (these are derived from the real estate register ALKIS) and the volume information were also available, which led to a significantly better data situation in the project. Country-specific effects and regional differences in the LoD1 data were analysed and processed. After the elimination of small polygons in the HU-DE, the intersection to a uniform data set followed. On the basis of the usage information, the house surrounds were classified as non-residential buildings or not and coded accordingly.
With the more than 40 calculated geometric features, the project partners were able to carry out a binary logistic regression in order to assign a concrete probability to the house surrounds that they are non-residential buildings. The created selection basis was transmitted to the project partners. For the screening there was the practical necessity to cluster the sample spatially, because an equal distribution across Germany would simply not have been financially feasible. Postal code areas or other administrative boundaries were excluded, as these did not allow for a uniform sample per area. For this reason, an approach for automated survey district formation based on minimum conditions was developed and variants for the survey documents and possible route runs of the screeners were tested. Until spring 2017, algorithms and processes of geodata processing were optimized. The results of the pilot phase (carried out with the network partners) were available from summer 2017.
The principle of the geodata-based identification of a non-residential building with the approach used turned out to be viable. At least half of the objects visited were non-residential buildings (the minimum criterion was fulfilled). The findings required slight modifications to the survey districts, threshold value determination and the survey documents. The thresholds defined by the IWU for the sampling of the main housing units at federal state level were included in the formation of the survey districts. The findings from the pilot phase were transferred to the other areas of responsibility. By the end of 2017, an optimized data provision structure for the main phase could be developed and all required data transmitted to the project partners. The central task of the IOER in the joint project was thus successfully completed. This was followed by the main phase of the survey in screening, broad and deep surveys. The practice-oriented phases were carried out by the project partners of IWU and BUW. The IOER was involved in project support, technical questions and separate evaluations. By May 2019 the screening was completed and in September 2019 the broad-based survey. After the screening data had been corrected by comparing them with those of the broad-based survey, the final version was available at the end of 2019. In the research network, the first foundations for the planned research database were laid down as early as 2018 and data flows were designed; these are to be established from 2020.