Green Benchmarking – An Assessment of 2D and 3D Mapping Approaches to Urban Greenspace Availability in the Netherlands

By Andrew McCabe

Abstract
Modern urban areas face the increasingly larger task of responding to the challenges presented by both climate change and a more urbanised population as more and more people move to urban areas. This demographic shift is visible within the Netherlands where the proportion of total population living in urban areas saw an increase from 87.8% to 92.5% between 2011 and 2021. With this trend showing no signs of slowing it is of utmost importance to Dutch urban planners that they are able to accommodate the balance between artificial structures and green space coverage in the urban environment. Increased urban green space coverage is shown to lower surrounding temperatures during hot periods as well as improving the quality of life for urban residents. Traditionally, urban vegetation was captured and measured with 2D mapping approaches from remote sensing imagery. Recently, research has utilised 3D mapping to capture urban vegetation’s complete spatial structure that is often missed with 2D mapping.

During this thesis, the urban vegetation of two Dutch urban areas were extracted from LiDAR and multispectral imagery. This extraction then served as the basis for three mapping approaches for urban greenspace (UGS) estimation; rasterization, voxelization of a LiDAR point cloud and a multispectral imagery NDVI approach. After estimations of UGS for each mapping approach was processed, the results of these estimations were utilised for a statistical analysis of UGS availability to assess inequality within two Dutch urban areas. Nine population metrics at the neighbourhood level were chosen to be assessed through a spatial GINI statistical test.

The results of this research show both urban areas meet high metrics of greenspace per urban inhabitant and neighbourhood indicating that citizens of these urban areas experience a high standard of availability in UGS in their living environment, yet there are inequalities in the spatial distribution of UGS within the urban areas. The choice of which mapping approach to utilise in measuring UGS availability was found to be largely similar in scope, but dependent upon the landscape profile of the area of study and filtering conditions of the geodata. Furthermore, both 3D estimation methods offer a similar overview of urban vegetation, while the 2D approach is restricted in offering a fair comparison due to dataset filtering and unavailable classification values.