Li et al. (2026) Uncovering the Spatiotemporal Evolution and Driving Factors of Flash Flood in the Qinghai–Tibet Plateau

Identification

• Journal: Remote Sensing
• Year: 2026
• Date: 2026-03-26
• Authors: Chaoyue Li, Xun Feng, Guotao Zhang, Zhonggen Wang, Wen Jin, C. Li
• DOI: 10.3390/rs18070996

Research Groups

• Key Laboratory of Mountain Hazards and Engineering Resilience, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China
• College of Urban and Environmental Science, Northwest University, Xi’an, China
• National Institute of Natural Disaster Prevention, Ministry of Emergency Management, Beijing, China
• National Disaster Reduction Center of China, Ministry of Emergency Management, Beijing, China
• College of Geography and Environment, Shandong Normal University, Jinan, China

Short Summary

This study investigates the spatiotemporal evolution and driving factors of flash floods in the Qinghai–Tibet Plateau from 1950 to 2015, revealing an exponential increase in flood frequency primarily driven by soil moisture and intensified human activities.

Objective

• To systematically examine the spatiotemporal evolution and integrated driving mechanisms of flash floods across the Qinghai–Tibet Plateau (QTP) using multi-source remote sensing data, historical disaster records, and interpretable machine learning models.

Study Configuration

• Spatial Scale: Qinghai–Tibet Plateau (QTP), covering approximately 2.5 × 10^6 square kilometers, analyzed at a 0.25° spatial resolution.
• Temporal Scale: Flash flood events from 1950 to 2015.

Methodology and Data

• Models used: Gravity model, Standard Deviation Ellipse (SDE), Spatial Autocorrelation (Global Moran's I, Local Spatial Autocorrelation), Random Forest regression model, Shapley Additive Explanation (SHAP) model.
• Data sources:
  • Flash flood event data: National Flash Flood Investigation and Evaluation Project database (1950–2015).
  • Runoff data: China Natural Runoff Grid Dataset CNRD v1.0 (1961–2018).
  • Soil moisture data: Monthly gap-filled CCI soil moisture over China.
  • Elevation data: SRTM DEM (~30 meter spatial resolution).
  • Land cover data: From Huang (1985–2015).
  • Historical water intake data: National Tibetan Plateau/Third Pole Environment Data Center.
  • Derived data: Runoff Concentration Index (QCI), Precipitation Concentration Index (PCI), snowfall fraction, slope, curvature, Human Activity Intensity Index (HAII).
  • All data resampled to a unified spatial resolution of 0.25°.

Main Results

• The frequency of flash floods on the QTP exhibited an exponential increase from 1950 to 2015.
• Flash flood events are highly concentrated between April and September, with a dominant peak during June–August, accounting for 86.4% of the annual total (July alone contributing 40.1%).
• The seasonal barycenter of flash floods migrated approximately 1003.33 kilometers from April to September, showing a predominant southwest–northeast alignment (standard deviation ellipses directions ranged from 56° to 74°).
• High-density flash flood areas are concentrated in the headwaters of the Yarlung Zangbo, Jinsha, Lancang, Nu, and Yellow rivers.
• A statistically significant positive spatial autocorrelation (Moran’s I = 0.164, Z-score = 3.203, p = 0.01) indicates spatial clustering, predominantly "high–high" and "low–low" aggregation patterns.
• Flash flood occurrences showed increasing trends across the QTP, with rapid increases observed in the South Tibet Valley and the Sichuan–Tibet Alpine Valley (16.48% of the total area).
• Soil moisture and human activity intensity (HAII) are the predominant driving factors, followed by Runoff Concentration Index (QCI), Precipitation Concentration Index (PCI), elevation, snow fraction, slope, and curvature.
• Soil moisture exhibits a nonlinear relationship with flash flood density, with its positive contribution peaking when moisture levels are between 20% and 40%.
• The Random Forest model demonstrated good predictive ability for flash flood occurrence, with an RMSE of 15.3, an MAE of 11.2, and an R^2 of 0.72 on the test set.

Contributions

• Provides a comprehensive, region-wide assessment of flash flood spatiotemporal evolution and driving mechanisms across the data-scarce Qinghai–Tibet Plateau.
• Integrates multi-source remote sensing data, historical records, and interpretable machine learning (Random Forest, SHAP) to quantify both natural and anthropogenic drivers.
• Highlights the increasing prominence of human activities as a key contributor and amplifying factor to flash flood frequency.
• Offers a practical and reproducible blueprint for investigating flood dynamics, advancing monitoring and early-warning research, and informing disaster prevention and mitigation strategies in High Mountain Asia.
• Supports the advancement of water-related targets under the United Nations’ Sustainable Development Goals.

Funding

• Deep Earth Probe and Mineral Resources Exploration-National Science and Technology Major Project (2024ZD1000500)
• China Postdoctoral Science Foundation (GZC20241687; 2024M763228)
• Xinjiang Key Laboratory of Water Cycle and Utilization in Arid Zone (XJYS0907-2024-zd-07)
• Integrated Research on Disaster Risk (IRDR) program

Citation

@article{Li2026Uncovering,
  author = {Li, Chaoyue and Feng, Xun and Zhang, Guotao and Wang, Zhonggen and Jin, Wen and Li, C.},
  title = {Uncovering the Spatiotemporal Evolution and Driving Factors of Flash Flood in the Qinghai–Tibet Plateau},
  journal = {Remote Sensing},
  year = {2026},
  doi = {10.3390/rs18070996},
  url = {https://doi.org/10.3390/rs18070996}
}