Developing a Satellite-Based Methodology to Estimate Built-up Biocapacity for Use in the National Ecological Footprint and Biocapacity Accounts: A Comparative Study of Pléiades Neo and Sentinel-2 Imagery, Classification Algorithms, and Segmentation Approaches

Built-up land is an important component of the ecological footprint accounting framework, yet the National Footprint and Biocapacity Accounts (NFBA) assume built-up land has the biocapacity of cropland. The NFBA uses equivalence factors and yield factors to transform raw area data into global hectares (gha), hectares of globally average productivity. This thesis aims to develop a satellite-based methodology for estimating built-up biocapacity. A 2,800-ha study area in The Hague, the Netherlands, was used to compare the effect of spatial resolution, classification algorithm, and segmentation technique on the results. Multispectral imagery from Pléiades Neo (1.2-meter resolution) and Sentinel-2 (10-meter resolution) provided the basis for classification into land cover categories. Identified land cover categories within the built-up fabric were transformed to biocapacity in gha by linking them to the NFBA land types and adapting the yield and equivalence factors from the NFBA to reflect reduced productivity in built-up conditions. Built-up biocapacity for the study area was between 2500 and 4000 gha, far below the 10325 gha using the current NFBA method. Among methodological factors, segmentation had the strongest influence, explaining 27.3% of variation for Sentinel-2 and 9.7% for Pléiades Neo. Pixel-based classifications in both Sentinel-2 and Pléiades Neo yielded remarkably consistent results across algorithms, showing a spread of 3.90% and 2.34%, respectively. However, while pixel-based classifications tended to capture more small, vegetated features, they were more prone to misclassification of shaded infrastructure as dune sands or water. Although Sentinel-2 results captured a lot of urban tree canopy, with a lower BC yield than trees in urban forests, the higher area attributed to this land cover explains the low difference in means between Sentinel-2 and Pléiades Neo results.

Author dc:creator

• Johanna Louise Van Berkum 2001-

Contributors dc:contributor

• Háskóli Íslands

• Umhverfis- og auðlindafræði
• Fjarkönnun
• Þéttbýli
• Mannvirki

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/1946/51551

OAI identifier oai:identifier

oai:skemman.is:1946/51551