Long-term monitoring of the built-up area is essential in order to study political indicators and their trends over time, which forms the basis for their future projections and public discussion of sustainable development paths. So far, the joint use of Landsat and Sentinel sensors for long-term built-up surface monitoring was an unsolved task in the state-of-the-art applications of remote sensing. This study introduces an integrated solution for inferring changes on built-up surfaces from Sentinel-2 MSI images, combined with historical Landsat scenes, organized into four epochs 1975, 1990, 2000 and 2014. The objective of this study is two-fold. First, we aim to develop a methodology for estimating multi-temporal global built-up surface and volumes that allows for controlled estimates of built-up change in time in rural and urban areas. Secondly, we aim to deliver the multi-temporal assessment of global built-up surfaces and volumes with greater accuracy than in the previous Global Human Settlement Layer (GHSL) products. Our approach relies on stratified multiple-quantization associative rule learning applied to Earth Observation data, object-oriented image processing, and multiple decision support ensemble modelling. Assessment of our model shows that the built-up surface change predictions of the proposed solution are more accurate than those reported in the previous GHSL data package (R2022A), as well as in other current multi-temporal estimates of built-up surface with worldwide coverage.

PESARESI Martino;  POLITIS Panagiotis;  GOCH Katarzyna;  KEMPER Thomas; 

2024-01-29

Publications Office of the European Union

JRC135315

978-92-68-11541-1 (online),   

1831-9424 (online),   

EUR 31811 EN,    OP KJ-NA-31-811-EN-N (online),   

https://publications.jrc.ec.europa.eu/repository/handle/JRC135315,   

10.2760/664949 (online),