Transforming shores: remote sensing analyses of two decades of coastal change in the Arabian Gulf (2001–2021)

Coastal zones along the Arabian Gulf have experienced extensive urban expansion and land–water transformation over the past two decades, yet comparable Gulf-wide evidence on how these changes vary across national, urban, and local scales remains limited. Building on a previously published regional assessment, this study extends the analysis through normalized comparisons across eight Gulf countries, quantitative assessment of 13 coastal urban centers, and examination of 33 selected 10 × 10 km Focus Zones. Landsat imagery from 2001 to 2021 and Random Forest classification were used to map urbanization, water-to-nonwater transitions, and nonwater-to-water transitions. These transitions are operationally termed reclamation and inundation, respectively, without independently attributing their causes. Regional results indicate approximately 6,827 km 2 of nonurban-to-urban conversion, 1,011 km 2 of water-to-nonwater conversion, and 1,810 km 2 of nonwater-to-water conversion. The added comparisons reveal substantial differences in transformation patterns among countries, coastal urban centers, and selected local settings. Pooled overall classification accuracies were 92.00% for 2001 and 92.96% for 2021. Although mapped nonwater-to-water conversion exceeded the reverse transition regionally, this balance does not independently establish erosion, sea-level rise, permanent land loss, or flood risk. The analysis provides a consistent framework for comparing coastal transformation at regional, national, urban, and local scales, while highlighting the need to distinguish engineered development, natural coastal processes, and acquisition-related differences when interpreting mapped changes. Future work should integrate higher-frequency imagery, tidal normalization, elevation data, and field or very-high-resolution validation to better distinguish engineered, natural, and climate-related coastal processes. This study is therefore framed as a two-date, long-term change assessment based on endpoint comparisons between 2001 and 2021, rather than a continuous time-series analysis of the timing or sequence of change.