Global Impervious Surface Area Dynamics from Landsat Satellites, 1972-2019

A global impervious surface area (GISA) dataset maps human-built surfaces from 1972 to 2019 using over three million Landsat satellite images. Developed by Xin Huang at Wuhan University, it was validated with 120,777 independent reference sites from 270 cities worldwide, achieving an F-score of 0.954. The dataset provides the longest time span for global ISA mapping and was extracted from the entire global land area to reduce underestimation.

Use Cases

• Analyzing global urban expansion patterns based on the 47-year time series of impervious surface area.
• Validating and comparing urban mapping methods based on the described semi-automatic sample collection and locally adaptive classification strategy.
• Modeling environmental impacts of urbanization based on the global, high-accuracy impervious surface maps.
• Studying historical land use change before 1985 based on the provided global ISA maps for that earlier period.

Strengths

• Longest time span for a global ISA dataset, covering 47 years from 1972 to 2019.
• High validation accuracy with an F-score of 0.954 based on 120,777 independent reference sites.
• Covers the entire global land area, not just urban masks, which reduces underestimation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

Wuhan University

Collection Method

Derived from analysis of over three million Landsat satellite images using a semi-automatic sample collection and locally adaptive classification strategy.

Time Range

1972-2019

Geography

Global