MELBOURNE — PORT PHILLIP, THE BAY BEHIND THE HEADS

The basics, fast

Most bays are a piece of the ocean with a shore around it. Port Phillip is closer to a reservoir that the ocean is allowed to top up twice a day through a gap it can barely get through, and it behaves accordingly.

• The Heads — the entrance between Point Lonsdale and Point Nepean. Everything the ocean does to the bay has to fit through it, and the tide arrives at Melbourne late, small and reshaped as a result.
• The bay itself — broad, shallow and remarkably featureless. Across the 24,113 seabed cells our terrain raster holds inside the bay, the median depth is 17 m and the median ruggedness is 0.19 m. There is not much down there to fish to, which is why the state has been building reefs.
• Western Port — the next bay east, and the instructive comparison. Its tide station sits 53.9 km from Point Lonsdale, almost exactly as far as Melbourne's 54.3 km, and its spring range is 3.9 times larger. Same ocean, same distance, a completely different fishery.

The practical upshot: in Western Port you plan around the tide, and in Port Phillip you mostly cannot, because there is not enough of one to plan around. What you plan around instead is wind, and the last sections of this guide are about that.

What this guide will not tell you

Worth stating before anything else, because it shapes what is and is not below.

• No bag limits, no size limits, no closure dates. Victorian recreational fishing rules belong to the Victorian Fisheries Authority, and they change. A number printed in a guide is a number the guide eventually gets wrong, and the infringement is yours rather than ours. Our Victorian regulations pages carry the current values with the date each was last verified against the source, which is a different promise from an article repeating a figure from memory.
• No marine-park or sanctuary-zone boundaries. Port Phillip contains several, and where you may fish at all is a separate instrument from what you may keep. Check the map that governs, not a blog.
• No marks. The four artificial reefs below have official published coordinates and those are the VFA's to publish, not ours to relay as if they were our discovery. Everything else here is reasoning, not numbers off somebody's sounder.
• No claim that our heatmaps solve this water. They do not, and there is a section saying exactly which parts of the bay's fishery our models have nothing to say about.

The door: what the Heads do to the ocean tide

A tide is a sum of constituents — separate waves, each with a fixed period, each arriving at a given place with its own amplitude and its own phase. Comparing those constants between two stations is the cleanest way to see what a piece of coastline does to the water passing through it.

Our tide engine holds eight constituents for each Victorian reference station. The one that matters most is M2, the principal lunar semidiurnal tide — the twice-a-day wave that does most of the work almost everywhere. Here is M2 at six stations, starting outside the Heads and working up the bay, with the phase difference converted into the thing an angler actually wants, which is minutes.

Two things in that table are worth slowing down for.

The first is where the delay happens. Queenscliff is 4.8 km inside the entrance and the tide gets there 25 minutes behind Point Lonsdale. Hovell Pile is 25.2 km in and it is 3 hours 9 minutes behind. So roughly 2 hours 44 minutes of the total lag is accumulated in the 20 km between them — over the entrance channels and the shallow sand banks immediately inside them — and then almost nothing is added over the remaining 29 km to Melbourne. Williamstown is only two minutes behind Hovell Pile despite being twice as far from the sea. The bay's interior does not delay the tide much. The door does.

The second is that the loss is front-loaded the same way. M2 is already down to 58% at Queenscliff, before the tide has entered the bay proper. It then drifts between 46% and 61% everywhere else we measure, including at the far end. Once the water is through, it is through; what the bay does afterwards is slosh, not decay.

For a fishing plan this has one blunt consequence. A Port Phillip tide time is only a tide time for the part of the bay it was predicted for. Our own Melbourne tide page is predicted for Melbourne Williamstown, so it already carries the full 3 h 11 min; if you read it and then drive to Queenscliff Pier to fish a top of the tide, you will arrive most of three hours after it happened. The Geelong page is its own station for the same reason.

The filter is frequency-selective, and that changes the shape of the day

The entrance does not damp all of the tide equally. It damps the fast part harder than the slow part, and the result is a bay whose tide has a different character from the ocean outside it, not merely a smaller size.

Compare how much of each constituent survives the trip to Hovell Pile in the middle of the bay:

The twice-daily wave loses more than half of itself; the once-daily waves keep about two thirds. That is what a constricted, shallow, frictional entrance does — the faster the flow has to reverse, the more of it is dissipated getting in and out.

Tidal science has a single number for the balance between those families, the form factor: the diurnal constituents divided by the semidiurnal ones. Outside the Heads it is 0.434. At Hovell Pile it is 0.659. The same water, on the other side of a 3 km gap, is measurably more of a once-a-day tide than it was in the open sea.

What that means in practice is that Port Phillip is not reliably a two-tides-a-day fishery in the way the coast outside it is. Some days one of the two highs is so much weaker than the other that only one of them is worth the drive, and which one it is moves through the month. This is the reason a Melbourne tide graph so often looks lopsided next to a Sydney one, and it is not a plotting artefact.

Western Port, the same distance from the same ocean

The strongest evidence that the entrance is doing all this is next door, where an almost identical distance from the same open water produces the opposite result.

Western Port does not merely keep the ocean's tide — it amplifies it, to twice the amplitude measured outside Port Phillip's entrance, and it does so while staying closer to the ocean's own semidiurnal character and running less than an hour behind it. Two bays, side by side, equally far from the same water: one with a 2.24 m spring range and one with 0.58 m, a factor of 3.9.

This is why advice does not travel between them, and why anglers who learn one and then fish the other tend to be wrong in a specific direction. Western Port runs hard, drains vast flats, and rewards fishing the stages of the tide; the fish there move on the water. Port Phillip's water barely moves at all by comparison, and the fish are found by covering ground, reading wind and light, and finding the few pieces of structure that exist — not by timing a run that is not coming.

The floor: a bay with almost nothing on it

The other half of what makes Port Phillip difficult is the seabed, and here we can be unusually precise, because the bay happens to be the best-mapped water in our entire eastern Australian model.

Our snapper habitat layer is built from a static seabed-terrain raster on a 250 m lattice, covering the east coast from 39.5°S to 28°S. Each cell records the depth, a detrended ruggedness figure, a slope, a mosaic-edge value that measures how abruptly the terrain changes, and — this is the part that matters here — the resolution of the bathymetry the cell was computed from.

Inside the box that contains Port Phillip and Corio Bay, our raster holds 24,113 cells within its 5–250 m working depth band, and every single one of them was computed from 30 m source bathymetry, against the 250 m national base grid that underlies, for example, the water off Sydney. Whatever else is uncertain about the bay, we are not guessing about its shape.

Nor, any longer, about its substrate. Victoria's statewide marine habitat map covers 93.7 % of those cells, and it reports the bay floor as 1.0 % hard ground on average. That is a survey, not an inference from shape, and it says the same thing the depth soundings do.

And what that fine-resolution data shows is a floor with essentially nothing on it:

Half the bay floor varies by under twenty centimetres over a 250 m cell. The 95th percentile — the roughest one cell in twenty, the pick of the whole bay — is three quarters of a metre. For scale, a single cell sampled at the entrance, in 6 m of water, returns a mosaic edge of 15.9 m and a bathymetric position 12 m above its surroundings: eight times the bay's 95th percentile for edge, in one cell — and the habitat map calls that cell 100 % hard ground, where the average inside the bay is 1 %. The most dramatic seabed in Port Phillip is the doorway, and after that it is a car park.

This is the honest explanation for something every Melbourne angler is told and few are told the reason for: that the bay is fished by covering water rather than by running between marks. There are very few marks. The structure that exists — the channels and banks near the entrance, the drop into the shipping channels, the artificial reefs below, the piers — is a small number of places in a very large, very flat basin.

The artificial reefs, and the thing our own model cannot see

Victoria's answer to a featureless bay has been to build structure into it. We hold four of those reefs as named spots, and running them through our own seabed model produces the most useful negative result in this guide.

The four in Port Phillip are Rhys Reef off Aspendale, Tedesco Reef off Seaford, Yakka Reef, and Kayaker's Reef, a shellfish reef within kayak range of Kerferd Road Pier. They are real, they are official, and the VFA publishes exact coordinates for the boat-based ones.

Here is what our terrain raster — the 30 m-sourced, best-mapped-water-we-have raster — reads at three of them:

Every one of them reads as flatter than the bay's own median, and the last column is the part we did not expect: Victoria's own statewide habitat map, which covers these cells at full confidence, reports the ground under all three as 100 % soft. Kayaker's Reef does not even return a value: it sits shallower than the 5 m floor below which the snapper layer deliberately scores nothing, because that model is a boat-and-drift surface and was never meant to describe wash and pier ground.

So two independent instruments — our seabed shape model, and the state's own benthic survey — both look straight at four reefs the state itself built and see sand. That is not a bug in either. A cell on our lattice is 250 m across; a reef-ball field is tens of metres across and a couple of metres proud of the bottom. Averaging a two-metre object over a 250 m cell erases it, and a habitat map built before or independently of the deployment has no reason to carry it. A built reef is precisely the case where seabed shape, mapped substrate and actual fish-holding structure come apart.

Which is why the layer no longer asks them to agree. Every reef in our own gazetteer is now carried as a known structure — a flag on the cell, set from the published mark, not inferred from the terrain — and the snapper layer scores those cells on that flag instead of on the shape it cannot see. It is the honest arrangement: the model says what it measured, and where somebody has told us there is structure, we carry that as a separate fact rather than pretending the morphology found it.

The practical advice is unchanged: for the artificial reefs, use the VFA's published marks and your sounder. The heatmap is useful for asking which broad parts of a coast have snapper-shaped ground; it is the wrong instrument for finding a structure smaller than one of its pixels, and a guide that let you believe otherwise would be selling you something.

What we do model here, stated plainly

Port Phillip falls inside our eastern Australian forecast region, which spans 45°S to 10°S and 142°E to 160°E. That means the map will happily draw layers over the bay, so it is worth knowing which of them mean anything on this water.

The species with habitat models in this region are yellowfin tuna, southern bluefin tuna, mahi mahi, black, blue and striped marlin, yellowtail kingfish, and snapper. Look at that list against what Port Phillip is actually fished for and the gap is obvious: the bay's mainstays — King George whiting, southern calamari, flathead, garfish, Australian salmon — have no habitat model of any kind. The pelagic models are built for shelf-break water that is nowhere near this bay.

That leaves snapper, which is genuinely a Port Phillip fish and genuinely modelled — but it is flagged beta, it is the terrain surface discussed above with all of that surface's limits, and it explicitly scores zero shallower than 5 m, which is to say at every pier in the bay. If you fish the piers, our snapper layer has nothing to tell you by design.

What does work on this water is the ordinary, unglamorous part of the product: the Melbourne forecast with wind, swell and water temperature, the station-specific tide pages, and the spot and ramp records. Given everything above — a bay where the tide is small and the floor is flat — that is arguably the right set of tools anyway. The wind forecast is the one that decides a Port Phillip day.

Ramps: a bay that does not strand you

The tide analysis pays off here in a way that is easy to miss, because it shows up as an absence.

We hold 28 boat ramps on Port Phillip. Exactly one of them carries a tidal restriction in its official access notes: the single-lane Kirk Point ramp, which is noted as mid-to-high tide only. That is what a 0.58 m spring range buys you — with barely half a metre between high and low at the top of the bay, a ramp has to be very marginal indeed for the tide alone to close it.

The contrast with the bay next door is exactly as the tide numbers predict. Western Port has 14 ramps in our data, and the one that carries a restriction, at Grantville, is flagged high tide only — a harder constraint, on a bay whose spring range is nearly four times greater and whose flats drain.

The ramps themselves are generous by Australian standards. Eight of the 28 have four or more lanes, and four are six-lane facilities: Altona, Werribee South, St Leonards and Clifton Springs. On a bay this size, spread around this much shoreline, the practical decision is usually which ramp puts you on the right side of the wind rather than which ramp is deep enough.

One honest caveat about that data. Of the 28 ramps, 15 are explicitly recorded as not tide-dependent and 12 carry no reading at all. A blank is missing information, not a guarantee — it means the official facility record did not state one, and nobody should read an empty field as an assurance that a ramp is usable at any stage. The single restriction above is what the source actually says, not a complete survey of what the water actually does.

Wind is the planning variable, and the bay's shape is generous about it

If the tide is small and the floor is flat, what is left to plan around is weather — and specifically wind, because a shallow bay builds a short, steep, unpleasant chop far faster than a deep one.

This is where Port Phillip's geometry gives something back. We hold a coastal exposure model for Victoria that samples the shoreline every 200 m and records, for each of 36 compass directions, how much open water lies upwind. Inside the bay that is 1,148 shoreline sample points.

Run every wind direction across all of them and ask how many points have under a kilometre of open water upwind — a working definition of shelter — and the answer never falls far. Even in the least forgiving direction, a south-southeasterly, 497 of the 1,148 points are still sheltered. Forty-three percent of the bay's shoreline is in the lee of something, in the worst wind there is.

That is the single most useful planning fact about Port Phillip, and it follows directly from its being an enclosed basin with a long, convoluted shore facing every direction at once. There is always a sheltered shore somewhere in this bay. The question on a windy Melbourne morning is never whether to go, it is which side to go to — and because it is 54.3 km from the Heads to Williamstown, that decision is worth an hour in the car and is the difference between a day's fishing and a day's punishment.

The practical version: a northerly flattens the northern shore near Altona and the top of the bay while the Mornington Peninsula side gets the fetch; a southerly reverses it and the Bellarine and northern shores come into their own. The Melbourne forecast is the tool for the timing, and the shoreline you pick is the tool for the comfort.

The piers, which are a fishery in their own right

Twenty named Port Phillip spots sit in our gazetteer, and thirteen of them are jetties. That ratio is not an accident of what we have collected — it reflects what the bay is.

A pier in Port Phillip does three things that the open bay does not. It puts structure in a place with almost none, it puts light on the water at night, and it reaches past the shallowest water without a boat. St Kilda, Middle Brighton, Mordialloc, Portarlington, Sorrento and Queenscliff are spread around enough of the shoreline that the shelter logic above can be applied to them directly: in most winds, several of them are in the lee.

The one thing worth carrying over from the tide section is that a pier near the entrance and a pier near the city are on different clocks. Queenscliff and Sorrento sit where the tide is barely half an hour behind the ocean and the water genuinely runs; St Kilda and Williamstown are three hours later, in water that mostly does not. Those are different fisheries with different rhythms on one bay, and the tide page you should be reading is the one for the end of it you are standing on.

Five ways to get a Port Phillip day wrong

• Reading a Melbourne tide time and fishing it at the Heads. You are three hours out. This is the single most common mistake the geography sets up, and the numbers in the first table are the whole story.
• Bringing Western Port habits across. Timing a run that has 0.58 m of spring range behind it will not produce what the same technique produces on 2.24 m. Different bay, different plan.
• Trusting a habitat heatmap to find a built reef. It cannot, as shown above — our own model and the state's habitat map both read all three boat-based reefs as flatter, softer ground than the bay average. Use the published marks.
• Treating a blank tide-dependency field as an all-tide guarantee. Twelve of the 28 ramps have no reading. That is silence, not assurance.
• Fighting the wind instead of driving around it. Forty-three percent of the shoreline is sheltered even in the worst direction. Being on the wrong side of a 50 km bay is a decision, not bad luck.

The honest truth about fishing Port Phillip

It is a hard bay to fish well and an easy one to fish badly, and the reason is that its two most obvious planning inputs have both been taken away from you. The tide, which is the spine of most Australian bay fishing, is here reduced to about half a metre and delayed by three hours by an entrance the water can barely get through. The seabed, which is where most fishing knowledge is stored, is flat enough that half of it varies by less than twenty centimetres over a 250 m cell — flat enough that the state has resorted to building reefs, and flat enough that our own best-resolution model cannot find the reefs after they were built.

What is left is wind, light, season and covering ground — and a bay whose shape means there is nearly always somewhere sheltered to do it. That is a less satisfying answer than a tide chart with a magic hour circled on it. It is also, as far as our own data can show, the true one.