Daiman et al. (2025) Assessing the link between changes in landscape and desertification in the chambal river basin using machine learning and remote sensing

Identification

• Journal: ENVIRONMENTAL SYSTEMS RESEARCH
• Year: 2025
• Date: 2025-11-29
• Authors: Amit Daiman, Sajid Pareeth, Biswa Bhattacharya
• DOI: 10.1186/s40068-025-00425-3

Research Groups

• IHE Delft Institute for Water Education, Delft, Netherlands

Short Summary

This study analyzed the linkage between landscape changes and desertification in the Chambal River Basin (India) from 1990 to 2020 using machine learning and remote sensing. It found that anthropogenic land alterations, particularly the conversion of vegetation and agricultural lands, amplified the region's vulnerability to drought and desertification.

Objective

• To conduct a comprehensive analysis of Land Use and Land Cover (LULC) changes and Standardised Precipitation-Evapotranspiration Index (SPEI) within the Chambal River Basin (CRB) from 1990 to 2020, specifically focusing on highlighting potential desertification evolution perspectives by identifying landscape changes and their correlation with SPEI.

Study Configuration

• Spatial Scale: Chambal River Basin (CRB), India, with a surface area of 143,219 km², covering parts of Madhya Pradesh and Rajasthan states.
• Temporal Scale: Three decades, from 1990 to 2020. LULC maps were generated for 1990, 1999, 2011, and 2020, and SPEI analysis covered 1990 to 2020 across various accumulation scales (3, 6, 9, and 12 months).

Methodology and Data

• Models used:
  • Machine Learning (ML) based Random Forest (RF) algorithm for LULC classification.
  • Standardised Precipitation-Evapotranspiration Index (SPEI) for drought condition analysis.
  • iCOR (Iterative Radiative Transfer Model Inversion) algorithm for atmospheric correction.
  • CFMask algorithm for cloud masking.
• Data sources:
  • Satellite: Landsat series (Landsat 5 Thematic Mapper, Landsat 8 OLI/TIRS).
  • Platform: Google Earth Engine (GEE) for data processing and analysis.
  • Drought Index: Global SPEI database (Spanish National Research Council - CSIC) with a spatial resolution of 0.5 degrees.
  • Ground truth/Validation: Google Earth imagery, NRSC-ISRO India LULC data, Copernicus LULC Products.

Main Results

• Overall accuracy for LULC classification was 84% (1990), 84% (1999), 82% (2011), and 86% (2020).
• Approximately 21% (57,020 km²) of the land cover area in the CRB changed between 1990 and 2020.
• Significant LULC changes (1990-2020) include:
  • Increase in built-up land (from 0.18% to 2%), cropland (from 39% to 57%), and water bodies (from 1% to 1.5%).
  • Decrease in deciduous forest (from 9% to 6%), scrub/thorn/open forest (from 27% to 22%), and barren rocky/waste/open land (from 24% to 11%).
• Major land transitions within the altered area were scrubland to cropland (27%), barren land to cropland (22%), and deciduous forest to cropland (8%).
• SPEI analysis indicated an increasing trend towards drier conditions over the period, with 1990 showing moderate wet to near-normal conditions (highest SPEI 1.34) and 2018 showing moderate dry to extreme dry conditions (lowest SPEI -2.78).
• Correlation analysis between LULC dynamics and multi-temporal SPEI values revealed moderate negative associations in several periods, particularly in 2018. This suggests that areas experiencing intense land transformation, especially conversion of vegetation and agricultural lands, corresponded with lower moisture availability and higher drought stress, indicating amplified vulnerability due to anthropogenic land alterations.

Contributions

• Provides a comprehensive, long-term (three-decade) analysis of LULC changes and their correlation with drought conditions (SPEI) in the Chambal River Basin, addressing existing knowledge gaps in land degradation for this semi-arid region.
• Demonstrates the effectiveness and efficiency of integrating machine learning (Random Forest) with remote sensing data on the Google Earth Engine platform for large-scale LULC mapping and desertification assessment.
• Offers valuable insights into the evolving dynamics of desertification driven by both climatic variability and anthropogenic factors, establishing a solid foundation for informed land monitoring and management strategies.
• Aligns with Sustainable Development Goal 15 (SDG 15) and the United Nations Convention to Combat Desertification (UNCCD), contributing to global efforts in combating desertification and promoting sustainable land management.

Funding

• This work has not received financial support, and no funding is available.

Citation

@article{Daiman2025Assessing,
  author = {Daiman, Amit and Pareeth, Sajid and Bhattacharya, Biswa},
  title = {Assessing the link between changes in landscape and desertification in the chambal river basin using machine learning and remote sensing},
  journal = {ENVIRONMENTAL SYSTEMS RESEARCH},
  year = {2025},
  doi = {10.1186/s40068-025-00425-3},
  url = {https://doi.org/10.1186/s40068-025-00425-3}
}