Scholarlink

Urban growth in Astana, Kazakhstan was quantified using multi-temporal Landsat imagery and an unsupervised change-detection workflow on Google Earth Engine. The methodological pipeline combined atmospherically corrected Landsat 5/7/8 imagery, spectral indices (NDVI, NDBI, NDWI), a 9-dimensional feature space, k-means clustering, and post-classification comparison. Three Areas of Interest (AOIs) were processed; quantitative reporting centers on the Central AOI for 2013, 2015, and 2017 using co-registered images with identical spatial footprint for direct comparison. Built-up share increased from 14.56% (2013) to 21.04% (2015) and 32.96% (2017), corresponding to interval gains of +6.48 and + 11.92 percentage points and a net gain of +18.40 percentage points (2013–2017). Annualized absolute change equaled 3.24 percentage points per year (2013–2015) and 2.38 percentage points per year (2015–2017). Spatial patterns indicate concurrent densification and edge-led expansion, consistent with corridor-oriented development and consolidation of peri-urban patches. Robustness arises from co-registered Central AOI imagery; elsewhere in the manuscript, figures can differ in shape due to acquisition geometry and cropping. Results align with documented pre-EXPO 2017 investment ramp-up and subsequent infrastructure and residential build-out through 2017, providing decision-relevant evidence on the pace and morphology of recent urbanization.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Using Machine Learning for Tracking Urbanization Through Space-Based Imagery

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Leila Rzayeva,
  • Zhaksylyk Kozhakhmet,
  • Nursultan Nyssanov,
  • Abiche-Adejor Egah,
  • Katish Kazbekova,
  • Samat Kairambayev

摘要

Urban growth in Astana, Kazakhstan was quantified using multi-temporal Landsat imagery and an unsupervised change-detection workflow on Google Earth Engine. The methodological pipeline combined atmospherically corrected Landsat 5/7/8 imagery, spectral indices (NDVI, NDBI, NDWI), a 9-dimensional feature space, k-means clustering, and post-classification comparison. Three Areas of Interest (AOIs) were processed; quantitative reporting centers on the Central AOI for 2013, 2015, and 2017 using co-registered images with identical spatial footprint for direct comparison. Built-up share increased from 14.56% (2013) to 21.04% (2015) and 32.96% (2017), corresponding to interval gains of +6.48 and + 11.92 percentage points and a net gain of +18.40 percentage points (2013–2017). Annualized absolute change equaled 3.24 percentage points per year (2013–2015) and 2.38 percentage points per year (2015–2017). Spatial patterns indicate concurrent densification and edge-led expansion, consistent with corridor-oriented development and consolidation of peri-urban patches. Robustness arises from co-registered Central AOI imagery; elsewhere in the manuscript, figures can differ in shape due to acquisition geometry and cropping. Results align with documented pre-EXPO 2017 investment ramp-up and subsequent infrastructure and residential build-out through 2017, providing decision-relevant evidence on the pace and morphology of recent urbanization.