Method
On the Western Australian south coast, most estuaries are bar-built and seasonally closed: a sandbar shuts the mouth for much of the year and is breached, often artificially, when levels rise. That makes the downstream boundary condition of any hydraulic model a variable rather than a constant. On the lower reaches of these rivers the tailwater assumption typically has more influence on the modelled flood level than the rainfall input does — and it is the assumption most commonly carried in unexamined from methodology calibrated somewhere else.
What "bar-built and seasonally closed" does to a model
Wilson Inlet at Denmark is the clearest local case. The sandbar closes the inlet from around late January through to August, and in most years it is opened artificially at Ocean Beach once the water level reaches about a metre above mean sea level, rather than breaching on its own. The estuary covers about 48 square kilometres with an average depth of 1.8 metres below mean sea level. So the level at the mouth of the Denmark River during a flood depends on the date, on how much water has already accumulated behind the bar, and on whether anyone has opened it.
Torbay Inlet behaves the same way, on a catchment of about 330 square kilometres roughly 55 per cent cleared, fed by a swampland system — Lake Powell, Lake Manarup, Five Mile Swamp, Seven Mile Swamp and Marbelup Brook — plus a network of agricultural drains. The sandbar sits across the mouth for most of the year and is breached after significant rain.
A model that adopts a single fixed tailwater on either system is not making a conservative simplification. It is answering a different question — and depending on which level was chosen, it can be badly non-conservative in exactly the scenario that matters.
The exception that proves the rule
Walpole and Nornalup are permanently open. The mouth sits at the extreme western end of Bellanger Beach, sheltered from the south-westerly swell by a high granite headland, and in that protected position tidal exchange and river flow together keep the bar open year round. The estuary is always tidal.
That does not make the boundary condition simple, it makes it different. The entrance channel narrows to about twenty metres where limestone and granite constrain it, and the inlets behind are small and shallow — 13.2 square kilometres combined. The result is that water level variation inside the inlets runs at only about 40 per cent of the ocean range measured at Albany. Apply an open-coast tide directly at the inlet and the tidal signal is overstated by more than a factor of two. The tide also reaches a long way inland: the Frankland is tidal for about twelve kilometres and the Deep is estuarine for about six.
The second mechanism nobody models
Much of the land around Torbay Inlet sits below one metre elevation. Under certain tide and wind combinations, seawater pushes back up the estuary and inundates land from the seaward side — flooding with no rainfall component at all. A barrage with floodgates was built there in 1912 to control it and was ultimately unsuccessful.
An assessment that models catchment runoff alone is modelling one of the two mechanisms that actually put water on the ground. On low-lying south coast land the coincident case — a catchment flood arriving while the bar is shut and the sea is pushing in — is the one that sets the design level, and it is not the case a rainfall-driven model produces by default.
And the third: what is in the soil
Draining the lower swampland around Torbay for agriculture exposed pyrite, which oxidised and left acid sulfate conditions across parts of the lower catchment. DWER classifies acid sulfate soil risk against the likelihood of the material occurring within three metres of the natural surface, precisely because that is the depth ordinary earthworks and drainage reach.
That constrains what a drainage strategy can reasonably propose. A detention basin excavated into the wrong material substitutes a water quality problem for the water quantity problem it was built to solve — and does so after construction, when the options are expensive.
What a defensible south coast model does differently
It states the outlet condition as an assumption and tests it, rather than adopting a level and moving on. It represents the bar state explicitly — open, closed, and the level at which breaching occurs — and reports the sensitivity of the flood level to that choice. It considers the seaward mechanism alongside the catchment one on low-lying land. And it checks the soil profile before proposing an excavation.
None of that is exotic. It is the ordinary discipline of stating what the model assumes and showing what happens when the assumption is wrong. On this coast it is also the difference between a set of numbers that survives review and a set that does not.