Tayer et al. (2025) Mapping resilience: A framework for analysing surface-water dynamics and persistent pools in non-perennial rivers using remote sensing, rainfall and river discharge data

Identification

• Journal: Journal of Hydrology
• Year: 2025
• Date: 2025-12-12
• Authors: Thiaggo C. Tayer, Leah Beesley, Ben Stewart‐Koster, Nick Bond, Michael M. Douglas, M. Rossi, Glenn B. McGregor, Jonathan C. Marshall
• DOI: 10.1016/j.jhydrol.2025.134750

Research Groups

• School of Agriculture and Environment, The University of Western Australia, Perth, Australia
• Australian Rivers Institute, Griffith University, Nathan, Australia
• Centre for Freshwater Ecosystems, La Trobe University, Wodonga, Australia
• Queensland Department of the Environment, Tourism, Science and Innovation, Dutton Park, Australia

Short Summary

This study developed a scalable, data-driven framework integrating remote sensing, rainfall, and discharge data to analyze multi-decadal surface water and persistent pool dynamics in non-perennial rivers. Applied to the Gilbert River (Australia, 1986–2023), the framework revealed a dynamic mosaic of 29 persistent pools, with increasing rainfall and discharge leading to more numerous, larger, and less fragmented pools over time.

Objective

• To introduce a scalable, data-driven framework integrating remote sensing, rainfall, and discharge data to investigate multi-decadal surface water and persistent pool dynamics across large scales.
• To determine the long-term trends (1986–2023) in rainfall, discharge, and key spatiotemporal surface water metrics (extent, number, size, fragmentation) within the Gilbert River study area.
• To identify how key hydrometeorological factors (rainfall patterns, zero-flow duration) influence the persistence and size of pools at the end of the dry season.
• To characterize the spatial patterns of persistent pools (location, size, persistence, clustering) in the Gilbert River.

Study Configuration

• Spatial Scale: A 100 km section of the Gilbert River, Australia, buffered by 1 km on each side. Remote sensing data had a 30 m spatial resolution, allowing detection of features approximately ≥30 m across.
• Temporal Scale: Multi-decadal analysis from August 1986 to December 2023 (37 hydrological years), with data aggregated into monthly composites.

Methodology and Data

• Models used:
  • Image-labelling algorithms (Scikit-image morphology module) for pool delineation.
  • k-means clustering for climatological baseline definition.
  • Predictive Mean Matching (PMM) for data imputation.
  • iRiverMetrics tool for ecohydrological metric generation.
  • Seasonal-Trend decomposition using Loess (STL) for time series analysis.
  • Seasonal Mann–Kendall test for trend detection.
  • Spearman’s ρ for correlation analysis.
  • Moran’s I for spatial autocorrelation analysis.
• Data sources:
  • Remote Sensing: Water Observations from Space (WOfS) dataset (Landsat-based, 30 m spatial resolution).
  • Rainfall: Monthly cumulative rainfall sums from Green Hills (Station ID 30158) and Georgetown Airport (Station ID 030124) stations, Australian Bureau of Meteorology.
  • River Discharge: Maximum discharge (ML/day) records from Gilbert River gauging station at Rockfields (Station ID 917001D), Queensland’s Department of Local Government, Water and Volunteers.

Main Results

• The framework identified a dynamic mosaic of 29 persistent pools in the Gilbert River, with no single pool remaining permanently inundated throughout the 1986–2023 study period.
• Long-term trends showed statistically significant increases in rainfall and discharge, leading to more numerous, larger, and less fragmented pools.
• Wet-season cumulative rainfall (Spearman ρ = 0.67 for wet area, ρ = 0.48 for pool numbers) and zero-flow duration (Spearman ρ = -0.65 for wet area, ρ = -0.65 for pool numbers) were strongly correlated with pool morphology and persistence.
• Seasonal patterns showed rainfall peaking in January, discharge, total wetted area, and area-weighted mean pool area (AWMPA) peaking in February, and the number of pools peaking in March, all declining sharply to October minimums (e.g., discharge reduction of ~99.9%, wetted area ~99%).
• Lag analysis indicated that lower rainfall (50 mm bin) resulted in longer response times for pool metrics (~3.9–4.1 months), while higher rainfall (>150 mm bin) shortened lags (~0.98–2.0 months). Pool fragmentation exhibited longer lags (4.7–7.0 months) with increasing rainfall.
• Persistent pools ranged in size from 0.36 hectares to 10.69 hectares (mean 3.09 ± 2.57 hectares) and persisted for 3 to 30 years (mean 10.14 ± 8 years).
• Spatially, 83% of pools were clustered, with off-channel pools exhibiting greater average persistence (16.1 years vs. 6.5 years) and larger area (3.92 hectares vs. 2.56 hectares) compared to in-channel pools. No significant broad-scale clustering was found for pool area or persistence (Moran’s I p > 0.05).

Contributions

• Introduces a scalable, cost-effective, and accessible data-driven framework for analyzing multi-decadal surface water and persistent pool dynamics in non-perennial rivers, addressing a gap in existing literature often focused on short-term or field-intensive studies.
• Provides a structured and flexible methodology that is reproducible, modular, and sensor-agnostic, allowing for tailored application across diverse river systems and research objectives.
• Utilizes standardized ecohydrological metrics and dynamic hydrological years, enhancing consistency, comparability, and accuracy in detecting seasonal patterns and long-term trends.
• Offers significant practical utility for ecological and socio-economic management by linking hydrometeorological drivers to pool responses, enabling targeted conservation, water-allocation decisions, and supporting climate-resilience goals (UN SDGs 6 and 15).
• Demonstrates the framework's technical feasibility and analytical power through a case study on the Gilbert River, revealing dynamic shifts in dry-season refuge availability and the importance of both older off-channel and newer in-channel pools.

Funding

• Resilient Landscapes Hub (Project 5.6) through the National Environmental Science Program (NESP).

Citation

@article{Tayer2025Mapping,
  author = {Tayer, Thiaggo C. and Beesley, Leah and Stewart‐Koster, Ben and Bond, Nick and Douglas, Michael M. and Rossi, M. and McGregor, Glenn B. and Marshall, Jonathan C.},
  title = {Mapping resilience: A framework for analysing surface-water dynamics and persistent pools in non-perennial rivers using remote sensing, rainfall and river discharge data},
  journal = {Journal of Hydrology},
  year = {2025},
  doi = {10.1016/j.jhydrol.2025.134750},
  url = {https://doi.org/10.1016/j.jhydrol.2025.134750}
}