Multitemporal Analysis of Declassified Keyhole Imagery’ for Landuse Change Detection in China (1960~1984): A Python-Based Spatial Coverage and Automation Workflow

Abstract

Keyhole imagery, acquired between the 1960s and 1980s, offers a unique opportunity to study land use changes prior to the era of modern remote sensing. This study evaluates the potential of free-download Keyhole imagery within China to detect land use changes over five 5-year periods (1960–1984). Using metadata and spatial analysis tools in Python 3.12, we classified images into three resolution categories (meter-level, five-meter-level, and ten-meter-level) and analyzed their spatial distribution and repeated coverage. Results show that 26.5%, 58.9%, and 34.0% of areas were capable of detecting at least one land-use change event for the respective resolution categories. The T3 period (1970–1974) exhibited the greatest diversity of imagery combinations among the five periods. However, uneven spatial and temporal coverage, particularly in western and rural regions, limits the ability of free Keyhole imagery to conduct continuous multi-temporal analysis, and collaboration with paid Keyhole imagery could fill gaps in coverage and improve the accuracy of land use change detection. The study highlights the potential of Keyhole imagery for historical land use research while underscoring the need for methodological refinements to address data limitations. The shared Python scripts and metadata processing techniques could also support other land-use change research using Keyhole imagery globally.