Posa et al. (2026) A spatiotemporal analysis of hydro-meteorological factors driving floods

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2026
• Date: 2026-01-06
• Authors: Poornima Chandra Lekha Posa, Conrad Wasko, Wenyan Wu, S. Sreedevi, Rajarshi Das Bhowmik
• DOI: 10.1016/j.ejrh.2025.103060

Research Groups

• Interdisciplinary Centre for Water Research, Indian Institute of Science, Bengaluru, India
• School of Civil Engineering, The University of Sydney, Sydney, New South Wales, Australia
• Department of Infrastructure Engineering, The University of Melbourne, Victoria, Australia
• Central Water and Power Research Station, Pune, India

Short Summary

This study investigates the spatiotemporal hydro-meteorological drivers of Annual Maximum Floods (AMF) across major Indian river basins, revealing that AMFs are frequently caused by heavy rainfall events smaller than Annual Maximum Rainfall (AMR) and are significantly amplified by antecedent catchment conditions like high baseflow and soil moisture, challenging traditional rainfall-centric flood estimation.

Objective

• To examine the relative role of rainfall, soil moisture, evapotranspiration, and baseflow in generating Annual Maximum Floods (AMF) across major Indian river basins, specifically investigating instances where Annual Maximum Rainfall (AMR) does and does not drive AMF, and identifying dominant spatiotemporal hydrometeorological clusters contributing to flooding.

Study Configuration

• Spatial Scale: Seven sub-basins across six major river basins in India (Kollegal, Yadgiri, Perur, Polavaram, Basantpur, Barman, Sarangkheda). Catchment sizes range from approximately 21,000 square kilometers to 307,800 square kilometers. Gridded data resolutions are 0.25° x 0.25° for rainfall and 0.1° x 0.1° for other reanalysis variables.
• Temporal Scale: Daily data spanning 44 to 54 years, depending on streamflow data availability for each basin. Analysis includes 1-day, 3-day, 5-day, and 7-day accumulation windows for hydrometeorological fluxes.

Methodology and Data

• Models used:
  • Temporal composite analysis
  • Bayesian multi-linear regression (BMLR)
  • Moran’s I spatial autocorrelation (using Local Indicator of Spatial Association - LISA functions in Python)
  • Block Maxima approach for identifying Annual Maximum Flood (AMF) and Annual Maximum Rainfall (AMR) events.
  • Digital filter method for baseflow separation.
• Data sources:
  • Streamflow (Q): Gauged daily streamflow records from the Central Water Commission (CWC), India.
  • Rainfall (P): Daily gridded rainfall data (0.25° x 0.25°) from the India Meteorological Department (IMD).
  • Evapotranspiration (E), Surface Soil Moisture (SMsurf), Root-zone Soil Moisture (SMroot): Daily gridded data (0.1° x 0.1°) from the ERA5 reanalysis dataset (European Centre for Medium-Range Weather Forecasts - ECMWF). SMsurf is the sum of soil moisture from 0 to 100 centimeters depth, and SMroot is from 100 to 289 centimeters depth.

Main Results

• Annual Maximum Floods (AMF) are more frequently caused by heavy rainfall events (smaller than Annual Maximum Rainfall - AMR) than by AMR events across all studied basins.
• While AMR-driven AMF events generally exhibit higher streamflow magnitudes, in the Kollegal and Yadgiri basins, heavy rainfall-driven AMF events can produce higher streamflow magnitudes.
• Bayesian multi-linear regression consistently identifies rainfall and baseflow as dominant, positive drivers of extreme streamflow across all basins and event types.
• Evapotranspiration, surface soil moisture, and root-zone soil moisture show variable and often statistically insignificant influences in temporal analyses. Root-zone soil moisture generally exhibits negative associations, while surface soil moisture tends to have positive associations, particularly for AMR-driven events.
• Spatial autocorrelation analysis (Moran's I) reveals stronger and more coherent clustering for soil moisture (Global Moran’s I typically 0.55–0.9) compared to rainfall (Global Moran’s I typically 0.2–0.57), indicating that antecedent saturation patterns are more spatially consistent and critical for AMF generation.
• Floods are significantly amplified when heavy rainfall spatially coincides with low evapotranspiration and high soil moisture, highlighting the importance of compound clustering effects.
• The Kollegal basin exhibits unique flood dynamics, with AMR not being a statistically significant flood driver and heavy rainfall-driven AMFs being more intense, influenced by post-monsoon rainfall, steep topography, and clay-loam soils.
• Basin size, despite a 15-fold variation (21,000 to 307,800 square kilometers), primarily shapes the degree of internal variability but does not alter the overall ranking of dominant hydrometeorological flood drivers.

Contributions

• Provides a novel integrated spatiotemporal framework, combining Bayesian multi-linear regression for temporal analysis and Moran’s I spatial autocorrelation for spatial analysis, to disentangle complex flood drivers.
• Challenges the traditional assumption that Annual Maximum Rainfall (AMR) is the primary driver of Annual Maximum Floods (AMF), demonstrating that AMFs are frequently generated by less intense, yet heavy, rainfall events under specific antecedent conditions.
• Emphasizes the critical role of antecedent hydrometeorological conditions (baseflow, soil moisture, evapotranspiration) and their spatial clustering in modulating flood generation, advocating for a shift beyond rainfall-centric flood estimation.
• Identifies localized spatial hotspots of hydrometeorological variables that are crucial for flood generation, an aspect often obscured by basin-wide averaging in purely temporal studies.
• Offers significant implications for improving flood forecasting, risk assessment, and climate adaptation planning by highlighting the need to incorporate multiple, spatially distributed hydro-meteorological factors and their dependencies.

Funding

Not explicitly mentioned in the provided paper text.

Citation

@article{Posa2026spatiotemporal,
  author = {Posa, Poornima Chandra Lekha and Wasko, Conrad and Wu, Wenyan and Sreedevi, S. and Bhowmik, Rajarshi Das},
  title = {A spatiotemporal analysis of hydro-meteorological factors driving floods},
  journal = {Journal of Hydrology Regional Studies},
  year = {2026},
  doi = {10.1016/j.ejrh.2025.103060},
  url = {https://doi.org/10.1016/j.ejrh.2025.103060}
}