Outlining where humans live -- The World Settlement Footprint 2015

Abstract: Human settlements are the cause and consequence of most environmental and societal changes on Earth; however, their location and extent is still under debate. We provide here a new 10m resolution (0.32 arc sec) global map of human settlements on Earth for the year 2015, namely the World Settlement Footprint 2015 (WSF2015). The raster dataset has been generated by means of an advanced classification system which, for the first time, jointly exploits open-and-free optical and radar satellite imagery. The WSF2015 has been validated against 900,000 samples labelled by crowdsourcing photointerpretation of very high resolution Google Earth imagery and outperforms all other similar existing layers; in particular, it considerably improves the detection of very small settlements in rural regions and better outlines scattered suburban areas. The dataset can be used at any scale of observation in support to all applications requiring detailed and accurate information on human presence (e.g., socioeconomic development, population distribution, risks assessment, etc.).

• The paper presents a novel 10-meter resolution global settlement map integrating both optical and radar imagery.
• The methodology utilizes advanced machine learning and SVM classifiers combined with a majority voting system for enhanced accuracy.
• The study validates the dataset with 900,000 samples, achieving an average accuracy of 86.37% against existing settlement layers.

Analyzing Human Habitation: The World Settlement Footprint 2015

The paper introduces the World Settlement Footprint 2015 (WSF2015), a 10-meter resolution global map detailing human settlements in the year 2015. This dataset marks a significant improvement over previous attempts at global settlement mapping by integrating optical and radar satellite imagery from Sentinel-1 and Landsat-8, offering a nuanced understanding of settlement extents. The map's development is underpinned by a novel methodology that employs both multitemporal data and advanced machine learning classification techniques, which result in robust and detailed delineation of settlements.

Methodology and Data Integration

The methodology stands out by leveraging complementary data from both radar and optical sources. Radar imagery, sourced from Sentinel-1, is known for its ability to penetrate cloud cover and provide geospatial data that highlights built-up areas due to double-bounce reflection. Meanwhile, optical data from Landsat-8 offers spectral indices sensitive to vegetation and built-up areas, aiding in differentiating settlement land cover from non-settlement classes such as agricultural or naturally vegetated lands.

The classification process utilizes Support Vector Machines (SVM) for separate evaluations of the optical and radar datasets, followed by a majority voting system among multiple classifiers to enhance reliability and reduce errors. This dual-modality approach successfully mitigates the limitations associated with single-source data, particularly with respect to defining smaller and more dispersed settlement patterns.

Validation and Results

Validation of the WSF2015 map was performed using a comprehensive set of 900,000 samples drawn from various global regions. The dataset was compared against existing global settlement layers such as the Global Urban Footprint (GUF), Global Human Settlement Layer (GHSL), and GLOBELAND30, demonstrating superior performance in terms of accuracy and reliability. The validation exercise revealed an average accuracy (AA) percentage of 86.37 with a Kappa coefficient indicating substantial agreement.

These impressive metrics highlight the dataset's efficacy in identifying settlements down to minute scales, particularly the smaller villages and suburban areas previously underrepresented in other datasets. Notably, WSF2015 outperformed in areas prone to misclassification due to complex topography or arid environments where previous datasets falter.

Implications and Future Directions

The introduction of WSF2015 provides a crucial tool for urban planning, socioeconomic studies, risk assessment, and environmental monitoring, offering granular insights into human settlement patterns that can inform policy and research at both global and local scales. The enhanced detail allows for a better understanding of spatial human dynamics, potentially improving models related to urbanization, population distribution, and infrastructure development.

Future advancements could build on the WSF2015 framework by integrating newer data sources and enhanced machine learning models further, refining settlement classification accuracy. Additionally, the periodic updating of such datasets with newer satellite missions could enable tracking changes over time, providing a dynamic perspective on global settlement evolution.

The development of the WSF2015 underscores the transformative potential of combining multiple remote sensing technologies to extract meaningful insights from Earth observation data, setting a precedent for future geospatial research methodologies and applications.