One of the underappreciated assets in Australian hydrology is the calibrated URBS model dataset compiled by Terry Malone — a collection of models spanning catchments across the continent, each with peak ratios (PR) derived from PQH files comparing gauge observations to modelled outputs.
At around 55,000 print locations, this is a substantial cross-section of Australian hydrology. I wanted to see what the data said if you looked at it differently: not model-by-model, but as a cloud of observations from which a frontier relationship might emerge.
The Question
When you plot rainfall, catchment area and peak flow together across 16,437 calibrated URBS locations, does a physically meaningful envelope appear? And if so, does it correspond to known regional flood estimation relationships?
What the Frontier Shows
The log-log scatter of area versus peak flow does produce a frontier — a bounding envelope that the dataset approaches but doesn’t exceed. The form of this frontier is:
ORT relationship: Q = 7 × Area^0.644
This aligns closely with the Queensland Regional Flood Estimation (QRT) relationship published by TMR, which is reassuring — it suggests the frontier is capturing a physically real upper bound on peak flow per unit catchment area, not a statistical artefact.
What This Might Mean
As we approach the frontier, the implication is that catchments are approaching the maximum runoff efficiency — close to a runoff coefficient of 1.0 for the design event. Events near or on the frontier correspond to conditions where:
- Soils are fully saturated (antecedent moisture at maximum)
- Rainfall intensity is high relative to catchment storage capacity
- The relationship between area and peak flow holds most cleanly
The corollary is that data points well below the frontier indicate either: sub-saturated antecedent conditions, significant attenuation through storage or floodplain routing, or lower rainfall intensity relative to capacity.
Where to Take This Analysis
This was exploratory work in progress. Some directions worth pursuing:
-
Climate zone stratification — does the frontier coefficient vary systematically by Köppen classification or BOM climate zone? The Monsoonal North versus the southeast coastal zone should behave differently.
-
Event conditioning — selecting only events with high antecedent moisture (soil moisture > 90th percentile from AWRA-L) should push points closer to the frontier.
-
Regional calibration — can the frontier be used as a sanity check on new URBS calibrations? A model producing peak flows well above the regional frontier for a given area warrants scrutiny.
-
Comparison to RFFE — how does the frontier relationship compare to the ARR Regional Flood Frequency Estimation outputs for the same catchments?
The dataset exists and is well-curated. The analysis is straightforward. It’s the kind of investigation that rarely gets done because it sits between individual project deliverables rather than within them.
Dataset: Terry Malone’s calibrated Australian URBS models (PQH output files). Analysis: WRM Water & Environment — Lindsay Millard, 2023. Originally shared on LinkedIn — 75 reactions.