Annually modelling built-settlements between remotely-sensed observations using relative changes in subnational...

Area of Science:

• Geographic Information Science
• Remote Sensing
• Urban Planning

Background:

• Mapping urban and built-settlement extents annually is crucial for demography, public health, and sustainable development.
• Existing remotely-sensed datasets suffer from limitations in spatial-temporal coverage, atmospheric interference, and cost, hindering the creation of comprehensive annual urban extent data.
• High-resolution global datasets for key years often have temporal discrepancies, making it difficult to achieve high spatial resolution and temporal coverage simultaneously.

Purpose of the Study:

• To develop and present an interpolative and flexible modeling framework for generating annual built-settlement extents.
• To address the challenges of limited temporal coverage and spatial resolution in existing urban extent datasets.
• To provide a method for creating annual binary built-settlement datasets at 100m resolution.

Main Methods:

• Utilized a combined approach of random forest and spatio-temporal dasymetric modeling.
• Integrated open-source subnational data for modeling.
• Produced annual 100m x 100m resolution binary built-settlement datasets for four diverse test countries over five-year intervals.

Main Results:

• The model correctly identified 85-99% of pixels transitioning to built-settlement in the majority of years across all study areas.
• The developed model significantly outperformed a baseline model where all pixels had an equal chance of transitioning.
• The framework demonstrated robust performance in varying environmental and developmental contexts.

Conclusions:

• The proposed modeling framework shows significant promise for filling data gaps in remotely-sensed urban and built-settlement datasets.
• This approach provides a foundation for creating future urban extent projections and exploring urban-population dynamics.
• The method offers a flexible and accurate solution for annual mapping of human settlement expansion.