Sridhar et al. (2025) Land Use and Water Storage Dynamics in the Krishna River Basin: Insights from Satellite Observations and Machine Learning

Identification

• Journal: ISPRS annals of the photogrammetry, remote sensing and spatial information sciences
• Year: 2025
• Date: 2025-12-19
• Authors: Venkataramana Sridhar, K. Satish Kumar, Christopher W. Zobel, Aditya Tyagi, Venkata Reddy Keesara, Myoor Padmanabhan
• DOI: 10.5194/isprs-annals-x-5-w2-2025-657-2025

Research Groups

• Biological Systems Engineering, Virginia Tech, USA
• Department of Civil Engineering, Indian Institute of Science, Bangalore, India
• Business Information Technology, Virginia Tech, USA
• Jacobs North America, Austin, TX, USA
• National Institute of Technology, Warangal, India
• Virginia Tech India, Chennai, India

Short Summary

This study utilizes the XGBoost machine learning algorithm to reconstruct 30 years of Terrestrial Water Storage Anomalies (TWSA) in the Krishna River Basin, identifying 15 major drought events. The findings reveal that while urban and forest areas have expanded, climatic variability remains the primary driver of water storage fluctuations rather than land-use changes.

Objective

• To reconstruct long-term terrestrial water storage anomalies (TWSA) from 1992 to 2022 using machine learning.
• To characterize spatiotemporal drought variability using the GRACE Drought Severity Index (GRACE-DSI).
• To analyze the correlation between land use and land cover (LULC) changes and basin-scale water storage variations.

Study Configuration

• Spatial Scale: Krishna River Basin, India (approximately 258,948 km²).
• Temporal Scale: 1992–2022 (Monthly resolution).

Methodology and Data

• Models used: Extreme Gradient Boosting (XGBoost) for TWSA reconstruction; Optuna (Bayesian optimization) for hyperparameter tuning.
• Indices: GRACE Drought Severity Index (GRACE-DSI) calculated at a 3-month scale.
• Data sources:
  • Satellite: GRACE/GRACE-FO JPL mascon Level 3 (TWSA), ESA-CCI (Land Cover at 300 m resolution).
  • Reanalysis/Models: GLEAM (Evapotranspiration), GLDAS Noah (Root zone soil moisture).
  • Observations: India Meteorological Department (IMD) gridded rainfall (0.25° × 0.25°) and temperature (1° × 1°).

Main Results

• Model Performance: The XGBoost model reconstructed TWSA with high accuracy compared to observed GRACE data ($R^2 = 0.92$, $NSE = 0.92$, and $RMSE = 43.18$ mm).
• Drought Dynamics: 15 major drought events were identified; the 2015–2017 event was the longest (28 months), while the 2018–2019 event was the most intense (minimum DSI = -2.63). Droughts typically recurred every 5–7 years.
• LULC Transitions: Between 1992 and 2022, agricultural land decreased from 86.24% to 84.74%, while urban settlements expanded from 0.25% to 1.05% and forest cover increased from 5.32% to 6.10%.
• Storage Drivers: Correlation analysis showed weak, statistically insignificant associations between LULC classes and water storage (e.g., Agriculture: -0.143; Urban: 0.153), suggesting that climatic factors and groundwater extraction dominate storage trends.

Contributions

• Successfully extended the GRACE satellite record backward by a decade (1992–2002) using a robust machine learning framework.
• Provided a long-term (31-year) high-resolution assessment of hydrological droughts in one of India's most water-stressed basins.
• Quantified the relative impact of land-use changes versus climatic drivers on terrestrial water storage, offering a template for drought governance in semi-arid regions.

Funding

• NASA Grant No. 80NSSC24M0196.

Citation

@article{Sridhar2025Land,
  author = {Sridhar, Venkataramana and Kumar, K. Satish and Zobel, Christopher W. and Tyagi, Aditya and Keesara, Venkata Reddy and Padmanabhan, Myoor},
  title = {Land Use and Water Storage Dynamics in the Krishna River Basin: Insights from Satellite Observations and Machine Learning},
  journal = {ISPRS annals of the photogrammetry, remote sensing and spatial information sciences},
  year = {2025},
  doi = {10.5194/isprs-annals-x-5-w2-2025-657-2025},
  url = {https://doi.org/10.5194/isprs-annals-x-5-w2-2025-657-2025}
}

Generated by BiblioAssistant using gemini-3-flash-preview (Google API)