Spring watering by city: September, October and November in the model

My heavy clay holds water like a grudge, so spring is when I get most suspicious of watering calendars. The September, October and November figures in this comparison tell different stories depending on the city — with Darwin's tropical comparison needing its own reading.

This is the spring slice of the FAO-56 city comparison. Same method, same stations, same declared assumptions. The irrigation figures below are monthly model estimates, not weekly watering prescriptions. The research pack behind them holds monthly figures. A weekly view is held until publishable weekly estimates exist.

Where the numbers come from

The short version, with the full method in the pillar post: monthly net irrigation is max(0, Kc × ET0 − 0.75 × rainfall), divided by 0.75 sprinkler efficiency to get gross sprinkler supply. ET0 and rainfall are Queensland Government Long Paddock SILO Patched Point series, FAO-56 reference ET0 on BoM-derived data, 1991–2020, one named station per capital. Kc steps through an illustrative warm-season scenario (0.55 in September, 0.60 in October, 0.65 in November for the cities other than Darwin; 0.70 throughout for Darwin). The 0.75 factors and the seasonal Kc mapping are declared scenario assumptions, not measurements of your lawn.

Each month's figure is the mean of that month's balances calculated separately in all thirty years — wet years don't cancel dry ones before the shortfall is floored at zero. That's the primary method; the lower "balance of climate means" variant stays in the pack as a sensitivity and isn't used here.

The spring months, city by city

Rainfall is the long-term monthly mean in mm/month. The irrigation column is the modelled gross sprinkler supply in mm/month for the scenario above — an estimate, rounded to whole mm as in the research pack.

For comparison, the 25 mm-a-week rule comes to about 108–110 mm a month, whatever the city or month.

A zero or small number is not a promise your lawn won't need water this month. The model doesn't see dry spells inside a month, how the rain actually falls, or what your soil can store between waterings. Treat each figure as a monthly estimate; more if a dry spell comes early.

What I read into it

Brisbane changes little in the table: modelled irrigation is 54 mm in September, 53 mm in October and 52 mm in November, while mean rain climbs from 32.2 to 89.3 mm. Those monthly means don't describe how rain was distributed within a month.

Across spring, Brisbane's estimate stays near 52–54 mm — about two 25 mm soaks a month, while the rule's comparison is about four.

Sydney steps up from 30 mm in September to 55 mm in October and 54 mm in November. Mean rain is 58.1, 55.8 and 71.7 mm respectively.

Sydney starts near one 25 mm soak in September and tips to about two in October and November, while the rule's comparison is about four.

Melbourne rises from 17 mm of modelled irrigation in September to 53 mm in November, while mean rain rises from 46.1 to 60.1 mm. More rain does not necessarily mean a smaller modelled irrigation estimate.

In September, Melbourne's 17 mm is about one 25 mm soak against the rule's roughly 108–110 mm — about four soaks; by November, 53 mm is about two.

Perth rises from 8 mm modelled in September to 117 mm by November, while mean rain falls from 78.9 to 28.7 mm. Perth also has watering-day rules; local restrictions take priority over a modelled estimate.

By November Perth turns the comparison around: the model's 117 mm is above the rule's roughly 108–110 mm, or about five 25 mm soaks, and watering-day rules decide when they can happen.

Adelaide follows the same drying shape, 23 mm to 111 mm, with mean rain falling from 47.4 to 24.4 mm.

Adelaide also flips above the comparison in November: 111 mm is just over the rule's roughly 108–110 mm, or about four 25 mm soaks.

Canberra climbs from 15 to 56 mm of modelled irrigation between September and November.

Canberra moves from about one 25 mm soak in September to about two in October and November, while the rule's comparison stays around four.

Hobart rises from 11 to 53 mm over those months, with mean rain falling from 55.6 to 49.4 mm.

Hobart moves from less than one 25 mm soak in September to about two by November, while the rule's comparison is about four.

Darwin runs the table in reverse: September has 14.3 mm of mean rain against 143 mm of modelled irrigation. By November mean rain rises to 143.5 mm and the modelled figure drops to 35 mm. The scenario keeps Darwin's Kc at 0.70 throughout, without imposing a southern winter slowdown. That is a modelling choice for the tropical comparison, not a validated local lawn calendar.

Darwin runs from roughly six 25 mm soaks in September down to about one by November, while the rule's comparison stays around four.

The limits I'd keep taped to the shed wall

• These are 30-year monthly means from named reference stations, predominantly airport locations; Hobart uses Ellerslie Road. They are not city-wide averages or a forecast for this spring.
• SILO's reference ET0 uses an assumed wind speed; local wind can change the result.
• Rainfall timing, runoff and soil storage aren't simulated. A monthly balance can hide a dry spell.
• The warm-season Kc scenario is a declared modelling choice, not a validated species calendar, and cool-season lawns sit outside it.
• Local watering restrictions override any modelled estimate.

I built LawnSuite around this kind of water balance — weather in, modelled deficit out. The watering explainer describes the approach. I read these city figures as context for the model, with its limits alongside the numbers.

Related guide

Read Post 1: the FAO-56 versus 25 mm-a-week city comparison for the full method, annual context and source pack.

About Macca: I built LawnSuite and its FAO-56 water-balance engine, and I grow Sir Walter on heavy clay, including the mistakes.