In recent years, the demand for 3D spatial information and 3D city models has increased, as they support and allow many different applications, e.g. noise simulations, energy demand estimations, and shadow analysis. Constructing a city model with 3D buildings requires elevation data (such as LiDAR or Digital Terrain Models), but unfortunately, data of sufficient quality is often unavailable. This thesis focuses on the use of machine learning methods to estimate the height of building footprints and thus bypassing the use of elevation data completely. Three different methods are tested and compared: Random Forest Regression (RFR), Multiple Linear Regression (MLR), and Support Vector Regression (SVR). A case study is performed for the conterminous United States of America (USA) because of its availability of a nation-wide building dataset, containing roughly 125 million building footprints. The high diversity in urban layouts is considered, where a distinction is made between Central Business Districts (CBDs) in cities and all other regions (e.g. suburbs and rural areas). All building footprints are characterised by nine features derived from their geometry, which are then used (in several combinations) in the model training and predicting stages. Furthermore, the influence of additional features – including census and cadastral data – on the results of the building height predictions is analysed for the city of Denver, Colorado. The experiments show that it is feasible to predict the height for all buildings in the conterminous USA in under 6 minutes. Both the MLR and SVR method even accomplish it in under 30 seconds. The height prediction results show that the different prediction models struggle to accurately estimate the height for buildings in CBDs. The lowest achieved Mean Absolute Error (MAE) is 31.81m, whereas for the suburban and rural areas it is 1.41m. Adding additional, non-geometric features (e.g. census data) to the prediction models for one city (Denver) proved to be successful; the RFR method reduced its MAE from 1.35m to 0.96m for the suburbs, achieving sub-metre accuracy. The CBDs, however, are still problematic with an MAE of 16.87m. These results show that for the suburban and rural areas, the accuracy recommendations from the CityGML specifications for LOD1 models can be met (5m limit). For the CBDs, improvement is required. The experiments also proved that the proposed methodology can be used to generate 3D city models of very large datasets if no elevation data is available. Moreover, the method is, in theory, generic enough to be applied outside the USA.

In 2021, noise pollution monitoring will be mandatory in the Netherlands, which requires data on traffic that can be re-used for air quality estimation models. One of the important input parameters for the latter is the street type, which is required by the dilution parametrisation used within the air quality model. The goal of this project is to show whether automatic street classification for air quality estimation is feasible and reliable, considering the geo-spatial data currently available in The Netherlands. The motivation for this project originates from the common data used in noise and air quality monitoring tools by the Dutch National Institute for Public Health and the Environment, (RIVM). Currently, street classification is performed manually by many municipalities. The larger municipalities are legally obliged to monitor air quality levels, which makes use of the street types. Automating the process by using existing datasets can save a lot of time, costs, and resources, while providing standardised results in comparison to manual classification. In addition, our method is extendable to the whole of the Netherlands. Consequently, our method can have a large societal impact, since it allows the provision of air quality estimations for all municipalities; even those that are not yet required to do so. To our knowledge, no similar work has been conducted in this field, which made it even a bigger challenge. The implementation of the automatic classification algorithm, which is thoroughly explained in this re- port, shows very promising results. We first tested the approaches in a small area, the Weesperstraat in Amsterdam, where we have success rates from 76.7% to 83.3% for the different classification methods when compared to the NSL classification. After evaluating the performance of each of the methods, the optimal approach has been tested on larger areas where visual inspection shows a priori promising results as well. In addition to the automatic classification algorithm, air quality measurements with new Flow sensors from Plume Labs were performed in the city of Amsterdam. The goal was to investigate whether different street types can be identified through the use of small air quality sensors. The limited measurements did not provide distinct patterns for the different street types, and therefore identification based on pollutant concentrations was not possible within the project. We hope that the results of this project will motivate public bodies and agencies in the Netherlands to invest in automated workflows using currently available and high accuracy geo-spatial data. This can potentially improve their efficiency, while creating a more standardised and scalable framework.