Scholarlink

<p>The current study examined the multi-temporal remote sensing (RS) data from 2001 to 2025 in Google Earth Engine (GEE) and used a machine learning (ML) based random forest (RF) model to assess the spatiotemporal changes in land use/land cover (LULC), while land surface temperature (LST) and the key spectral indices (SI) including normalized difference vegetation index (NDVI), normalized difference built-up index (NDBI), normalized difference water index (NDWI), and bare soil index (BSI) were derived to characterize surface dynamics in the arid heartland of Pakistan. Key results show a 480% expansion of water bodies due to recurrent flooding, accompanied by net vegetation loss (− 2.89%) and barren land reduction (− 4.20%) between 2001 and 2025. LST analysis indicates a 21% reduction in thermal range declining maximum temperatures, rising minimum temperatures, and persistent urban heat island intensification linked to impervious surface expansion. The spatiotemporal SI results show an expansion of NDVI extremes (− 0.56 to 0.86) indicating intensified irrigation-driven greening alongside expanding non-vegetated surfaces, while NDBI peaks increased to 0.34, reflecting progressive urban densification. NDWI maxima rose from ~ 0.35 (2001) to 0.56 (2025), confirming flood-induced expansion of surface water and soil saturation, and BSI maxima increased to 0.29, indicating greater exposure of bare and disturbed land surfaces. Overall, the results reveal strong land-climate interactions in arid, flood-prone environment and confirm the effectiveness of GEE-based ML analytics for long term environmental monitoring and climate adaptive land management in South Asia arid regions.</p> Graphical Abstract <p></p>

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

Integrating Google Earth Engine and Machine Learning To Map LULC and LST Dynamics in the Arid Heartland of Pakistan

  • 期刊论文

  • 发表日期：

    2026年2月4日

查看全文
  • Neyha Rubab Syed,
  • Nazia bibi,
  • Laila battool,
  • Aleeza Noureen,
  • Siham Acharki,
  • Ram L. Ray,
  • Mansour Almazroui,
  • Ali Raza

摘要

The current study examined the multi-temporal remote sensing (RS) data from 2001 to 2025 in Google Earth Engine (GEE) and used a machine learning (ML) based random forest (RF) model to assess the spatiotemporal changes in land use/land cover (LULC), while land surface temperature (LST) and the key spectral indices (SI) including normalized difference vegetation index (NDVI), normalized difference built-up index (NDBI), normalized difference water index (NDWI), and bare soil index (BSI) were derived to characterize surface dynamics in the arid heartland of Pakistan. Key results show a 480% expansion of water bodies due to recurrent flooding, accompanied by net vegetation loss (− 2.89%) and barren land reduction (− 4.20%) between 2001 and 2025. LST analysis indicates a 21% reduction in thermal range declining maximum temperatures, rising minimum temperatures, and persistent urban heat island intensification linked to impervious surface expansion. The spatiotemporal SI results show an expansion of NDVI extremes (− 0.56 to 0.86) indicating intensified irrigation-driven greening alongside expanding non-vegetated surfaces, while NDBI peaks increased to 0.34, reflecting progressive urban densification. NDWI maxima rose from ~ 0.35 (2001) to 0.56 (2025), confirming flood-induced expansion of surface water and soil saturation, and BSI maxima increased to 0.29, indicating greater exposure of bare and disturbed land surfaces. Overall, the results reveal strong land-climate interactions in arid, flood-prone environment and confirm the effectiveness of GEE-based ML analytics for long term environmental monitoring and climate adaptive land management in South Asia arid regions.

Graphical Abstract