The identification and analysis of regional building stock characteristics using map based data

Building energy and Carbon emission calculation methods for regions are of limited use if appropriate input data cannot be economically generated. To enable a wider uptake of regional modelling methods an automated analysis system is required to replace or assist time-consuming and expensive manual surveys of building stock. Building age is an important parameter in estimating energy use and Carbon emissions. In this paper a number of methods to extract information about the built environment from digital maps and use that information to infer building age have been tested against a database of a known large urban region. The methods include different types of shape recognition of plan form and of identification of contextual geography; e.g. distance from entrance to the nearest road. Tested against samples containing several thousands of domestic buildings from a known region, it was found that the different methods were able to cluster buildings into different form “styles”, and that those styles had some correlation to built age.  Victorian (pre-1919) age housing was detected with the greatest accuracy, with over 90% in the sample tested correctly identified.  This is useful as those older buildings are often the least energy efficient. Success in identification of other eras was less pronounced; although the results are promising, further development of the methods are required.