Chen et al. (2026) Long-term flood frequency assessment for wetland vegetationin the Murray–Darling Basin, 1988-2024

Full description

This study provides an important Basin-scale assessment of long-term flood frequency and extent (1988–2024) for wetland vegetation. We used the Basin-wide wetland vegetation classification (Chen et al. 2025) as the spatial framework.To determine water requirements for each wetland vegetation class, we used the dominant species in each class as surrogates and compiled their water requirements based on frequency, duration and seasonality and variability of flooding for maintenance or regeneration (Roberts and Marston 2011; Casanova 2015). We used a dataset of two-monthly surface water observations (Ticehurst et al. 2023) for a 36-year period (1988-2024) generated from Landsat surface reflectance remote-sensing data via Digital Earth Australia, and complemented with Water Observation from Space (WOfS) to minimise gaps due to cloud cover (Ticehurst et al. 2022). The original dataset of two-monthly observations consisted of GeoTIFF files for each observation, at 30 meters resolution. Each pixel was assigned one of three values: ‘3’ indicating water present, ‘2’ indicating dry conditions and ‘0’ indicating null values for areas affected by cloud cover or steep topography with no valid Landsat data for water. We processed the two-monthly datasets in ArcGIS Pro and projected them into the consistent co-ordinate system Geocentric Datum 2020/MDA 2020 Zone 55. The dataset was reclassified into binary values, where 1 indicates ‘wet’ and 0 indicates ‘dry’. The two-monthly datasets were then aggregated into annual water-year layers (from 1st of July to next year end of June). We derived the maximum flood coverage for each year and used the annual maximum flooded extent to measure flood frequency over the 36 years period. We assessed water requirements for each wetland vegetation class by identifying areas meeting its flood frequency requirements. Flood intervals were defined as floods occurring every year, every 1-3 years, every 3-7 years and every 10-20 years. Annual flood extent was analysed to identify those pixels that met these specific intervals between flood events. Only pixels meeting each interval criterion were retained and a new spatial layer was produced representing the areas meeting the flood frequency requirements. We used this approach because long-term vegetation condition is influenced by the sequence of flooding events rather than an average inundation pattern or a single ecologically effective flood (Casanova 2015). To identify the extent and distribution of optimally watered vegetation, we intersected the spatial layers of each wetland vegetation class with the layer corresponding to its specific flood frequency requirement. Each wetland vegetation class was evaluated only against its flood-frequency threshold, and the overlapped areas were classified as optimally watered. To identify sub-optimally watered vegetation, we overlayed the wetland vegetation layers with the total flood extent layer and excluded optimally watered areas using the erase function in ArcGIS Pro. We then excluded all flooded areas to derive the remaining wetland vegetation areas that had not been flooded during the 36 year period. We calculated the extent for each flood category and wetland vegetation class using the geometry function.

Notes

163 MB.