August gave us 25.2 mm and we called it close to normal. September didn't even try. The gauge collected 4.2 mm for the whole month, and the most of that came on one day, 2.2 mm on the 29th, which barely wets the top of a pot plant. What September did deliver is heat and pressure. The ridge that sat over us through winter got stronger, not weaker, and the barometer averaged the highest September figure our station has ever logged. We checked that against an independent dataset going back to 1995 and it holds up there too.

What actually happened, in numbers

Rain came on eight days but only six of them reached the 0.2 mm we count as a rain day, and the month total was 4.2 mm against a September median of 35.0 mm on our record. Mean daytime top 28.6°C, mean overnight low 9.8°C. The hottest day was 36.9°C on the 6th and the coldest morning was 5.9°C on the 30th, so no frost readings this month either.

Nine days made 30°C or better: the 3rd to the 6th, the 19th to the 22nd and then the 27th, which scraped in at exactly 30.0°C. That equals September 2024 for the most 30 degree days we have. Five afternoons dropped to 25% humidity or lower, with the lowest at 20% on the 19th. Mean dew point was 12.8°C, so the air wasn't short of moisture overall, it just never organised into rain.

Mean sea level pressure averaged 1025.9 hPa. The previous best September on our station was 2023 at 1022.6 hPa, so this one beat it by more than 3 hPa. That's a big margin for a monthly average.

How it compares with every other September on the record

Have a look at the right hand end of that chart. The two lowest bars on 16 Septembers of record are the last two. 2025 gave 3.3 mm and 2026 gave 4.2 mm, and the next driest after those is 5.4 mm in 2012. September here can swing wildly, 247.2 mm in 2010 and 175.0 mm in 2008 at one end, so two near empty ones back to back stands out.

2020 and 2021 came into the database from an older import where only the monthly rain total is reliable, spread evenly across the days, and there is no temperature data for those two Septembers at all. So those cells stay blank rather than show numbers we do not trust. No station data for 2007, 2013 or 2017 to 2019.

Warm, yes. A record for heat, no.

On paper 28.6°C is the highest September average top we have, 3.1°C above the long run figure of 25.5°C. We're not going to call it a record, and here is why. The station got a new temperature sensor in February 2025. When we lined the two sensors up, the old one read daytime tops about 1°C cooler then the new one and overnight lows about 0.6°C warmer. Correct September 2024 for that and its 27.8°C becomes roughly 28.8°C, which is warmer than this year. The only clean comparison is against September 2025 on the same sensor, and against that this September was 0.9°C warmer on top and 0.3°C cooler underneath.

So the honest summary is a warm September, not a record one. What it did have is big daily swings. On the 19th the morning sat at 7.8°C and the afternoon reached 32.9°C, a 25.1°C range inside one day. Dry air does that, it cools fast at night and heats fast once the sun is up, and it is the same reason the overnight average came in a touch below normal while the tops ran hot.

Is September getting warmer? The 137 year view

Our station only goes back to 2006, which is not long enough to say anything about climate. So we pulled the SILO record for Dayboro. SILO is the Queensland Government's daily climate dataset, built from Bureau observations and interpolated to any point, and for our spot it runs from 1889 and is still updated every day.

The answer is yes, warmer, but mostly in the afternoon. Across 1889 to 2025 the September daytime top has risen 0.13°C per decade, and since 1970 that sped up to 0.41°C per decade. The 1961–1990 September average top was 24.0°C, the 1991–2020 average 25.5°C. The nights tell a different story. The September low rose only 0.06°C per decade over the full record, and since 1970 the trend in lows is too small to separate from the year to year noise. The two 30 year averages for lows are 11.3°C and 11.4°C, near enough identical.

On SILO's numbers September 2026 was the 12th warmest of 138 Septembers for tops and only 36th for lows, and its 4.4 mm makes it equal 11th driest since 1889. Warm and dry, but not off the chart. Two caveats. SILO temperatures before 1957 are reconstructed from surrounding stations, not measured on the spot, so treat the early years as softer numbers. And SILO is a smoothed grid point, so its 26.3°C September average top is lower then the 28.6°C on our own station. The trend is the useful part, not the single month.

We're not going to hang that on one cause in a monthly review. Our reading is that warmer days with flat nights is also what drier air and clearer skies do, and that is pretty much what this spring has been. What we can say plainly: a September top of 26 degrees used to be a warm one. Now it's just a normal one.

The pressure record, checked twice

A record on our own barometer is one thing, but a single sensor can drift. So we went and control the barometer before calling anything, against ERA5, the European reanalysis that rebuilds the atmosphere from every observation available going back decades. For the grid box over Dayboro, September 2026 averaged 1024.5 hPa. That is the highest September in 32 years of that data, 5.7 hPa above the 1995 to 2025 average of 1018.8 hPa and 1.9 hPa clear of the previous top, 2023. Our station reads about 1.4 hPa higher than ERA5 month after month, which is normal for a point versus a grid average, and both agree on the ranking.

It isn't a one month thing either. ERA5 has June at 1024.0, July at 1023.7, August at 1024.2 and now September at 1024.5 hPa. Four months in a row with a high parked over southern Queensland. That is the reason July, August and September together gave us only 37.2 mm.

Day by day: two hot runs and a lot of nothing

The month opened mild, 24.8°C on the 1st, and then the heat climbed fast. 28.9°C on the 2nd, 32.9°C on the 3rd, 34.2°C on the 4th with the humidity down at 24%. The 5th was the most humid day of the month with a mean dew point of 16.0°C, and the 6th was the hottest, 36.9°C at 13:36 off a westerly (we wrote that one up the next morning). Pressure on the 6th averaged 1016.9 hPa, the lowest daily average of the month, which is the classic setup ahead of a change. The change came on the 7th, knocked the top back to 27.0°C and brought exactly nothing in the gauge.

The 10th to the 12th was the cool patch, 23.0°C and 23.2°C tops with pressure averaging between 1028.8 and 1030.7 hPa, and three days of drizzle that added up to 0.8 mm. Then dry again. The 14th gave a 6.1°C morning, and tops worked their way back up through the high twenties.

The 19th to the 22nd was the second hot run and in some ways the nastier one. 32.9°C, 32.3°C, 34.7°C and 31.1°C, with afternoon humidity of 20%, 24% and 21% on the first three days. Three afternoons in a row at or below 25% humidity, on cured grass, is the kind of stretch the fire people worry abotu, and we wrote about that in our spring bushfire risk post.

The 22nd and 23rd turned muggier, dew points over 15°C both days, and the 23rd never dropped below 13.1°C. After that pressure climbed to 1031.1 hPa on the 25th, the highest daily figure of the month, and the 24th to the 26th managed 1.0 mm between them. The 27th touched 30.0°C for the ninth 30 degree day, the 28th had a mild night that bottomed out at 14.3°C with the dew point up at 15.7°C, and the 29th finally gave us the wettest day of the month: 2.2 mm with the top held to 21.7°C. Behind it the 30th opened at 5.9°C, the coldest morning of September, under clear skies and high pressure again.

Why the ridge would not budge

Indeed, the drivers all point the same direction this month, which makes them easier to read then usual.

First, El Niño close to record strength. The BOM's relative Niño3.4 index was +2.72°C for the week ending 27 September. The El Niño threshold on that index is +0.80°C, so it sits more than three times over it, and the BOM describes the event as near record. The 30 day Southern Oscillation Index was −20.5. A strongly negative SOI goes with higher pressure and weaker trade winds on our side of the Pacific, which is exactly the ridge we have been living under since June.

Second, the Indian Ocean Dipole leaning positive. The weekly IOD index was +0.76°C, the third week running above the +0.4°C threshold. Officially it's still called neutral until it holds there long enough, but it's clearly heading positive. A positive IOD cuts off one of the moisture sources for inland and eastern Australia in spring, and stacked on top of a strong El Niño it is the classic dry spring combination.

Third, the Southern Annular Mode. SAM ran mostly negative through September before easing back to neutral around the 26th. In spring a negative SAM lets the westerlies sit further north, and dry westerly air is what came over the D'Aguilar Range on the 6th.

None of that is our forecast, it is the state of the drivers as measured at the end of the month. Our own 7 day and 30 day outlooks are in the member area, see below.

So is a wet summer coming after the El Niño?

Plenty of people around town, us included, have it pegged that a big El Niño gets followed by a wet summer. We wanted that to be true, so we checked it properly against the SILO rain record for Dayboro back to 1889.

The honest answer is sometimes, not reliably. The median Dayboro summer, December to February, is 438 mm. After the 1997 El Niño the next summer gave 888 mm, a proper wet one. After 1982 it gave only 222 mm. 2015 was dry in the El Niño summer itself (163 mm) and dry again the summer after (200 mm), while 2023 went the other way with 694 mm in the El Niño summer. Across all the El Niño years the result flips depending on which list of events you use, and none of it is strong enough to call a pattern. Several of the wet summers that came after an El Niño were La Niña summers, and that's what really brought the rain.

Our own long range outlook, built on the Inigo Jones cycle method, leans a touch wetter for December to February, about 107% of normal rain with a 38 to 39% chance of the wettest third against the usual one in three. That lean comes from the slow Pacific decadal cycle, not from the El Niño ending, and the outlook labels every summer month "no clear lean". It is also still experimental, on our own scoreboard it doesn't yet beat plain long term averages, and we would rather tell you that then sell you a wet summer. So keep watering carefully and treat a wet summer as a bonus if it comes.

If you want the back story: in April we looked at what this El Niño could mean for Dayboro, in June we went through the super El Niño hype, in July we checked every number in the strongest in 75 years headline, and our June review is a good reminder that an El Niño year can still throw a wet month. And if you wonder how far to trust Inigo Jones at all, we tested him.

What this means on the ground

• Storage. Year to date the station has 905.9 mm against 1467.6 mm to the same point in 2025, so we are about 38% behind last year. July, August and September together gave 37.2 mm. Anything that runs off a roof into a tank this spring is going to have to come from storms, and the drivers are not friendly to storms.
• Soil. Our 20 cm soil probe (how the two probes work) sat between 2% and 5% moisture for the whole month. That sounds like a faulty probe, but it isn't, the ground really is that dry. The deeper 35 cm probe slid from 51% to 46%, so there is still some reserve down low but it is being drawn on. Soil temperature at 35 cm averaged 19.5°C against 18.4°C in September 2025.
• Fire fuel. Five afternoons at or below 25% humidity, nine days over 30°C and grass that has dried right off. If you haven't slashed around the house and sheds yet, this is the time, before the first proper storm season winds.
• Paddocks and lawns. Kikuyu, couch and paspalum get going once the soil holds above about 14°C, and kikuyu grows best between 15 and 25°C. Our probe sits deeper than the 10 cm those figures are usually measured at, but at 19.5°C warmth is not what is holding the grass back, water is. If you can irrigate, this is when it pays. Hold the fertiliser until the ground is moist and the grass is moving, nitrogen on drought stressed pasture is mostly money down the drain. The 20°C figure you hear around is for sowing kikuyu seed, not for waking up a paddock that is already there.
• In the garden. Beans, cucumbers and pumpkins want the seed row above about 16°C to come up, pumpkins are happiest from 21°C, and tomatoes and sweet corn manage from about 10°C. All of that is ticked now. The catch is the dry, so soak the row the day before you sow. GardenBuddy's Plant Now list shows what suits our zone this month, and their Water Smart guide covers making a dry spring's water go further.
• Watering. 4.2 mm over a month doesn't do anything for a garden bed. Water deep and less often so the roots follow it down, mulch thick, and water in the morning so the hot dry afternoons do not take it straight back.
• Frost is done. Coldest morning was 5.9°C on the 30th. We reckon the frosts are finished for this year, the bigger risk for seedlings is the heat.

Get the numbers before the month is over

These monthly write ups are the rear view mirror. If you want the same data while it is still useful, that is what the Weather Reports membership on dayboro.au is for. It runs $3.95 a month, or $39 for the year, and it gets you the member weather reports, our 7 day and 30 day outlooks, the forecast accuracy scoreboard where we publish where we got it wrong, the weather history reports and the rest of the hub. Our station sits right here in Dayboro because the nearest official gauge is in another valley and it does not describe what happens here. And if you just want to see what is happening this minute, the live weather display is free.

Two more local sites worth a look

A lot of what we get asked after a dry month is the same two questions: what can still go in the garden, and is anyone else seeing the same. These two answer both.

For the garden GardenBuddy.au A garden planner built for Australian conditions. Lay out your beds, get sowing dates for your own climate zone instead of a northern hemisphere calendar, see live vegetable prices so you know what is worth growing, and track what you harvest. Handy in a dry spring when every bed has to earn its water. Visit gardenbuddy.au For the neighbourhood LocalBuddy.au A community space for the town itself, with local events, news, a marketplace and messaging between neighbours, and no ads sold over the top of it. Dayboro has its own space there, a good place to swap notes on which tanks are running low and who has a slasher free on the weekend. Visit Dayboro on LocalBuddy

We will run the same numbers for October at the start of November. With El Niño near record and the IOD leaning positive, October storms are the main hope for the tanks, and we'll see if they turn up.

Method note: Dayboro numbers come from our own station archive. Rainfall totals are full calendar month and a rain day is any day with 0.2 mm or more. Temperature, pressure, dew point and humidity averages use all 30 days. The record covers 2006 to 2026 with no data for 2007, 2013 and 2017 to 2019, and the 2020 and 2021 Septembers hold a monthly rain total only. The temperature sensor changed in February 2025, the older sensor reading daytime tops about 0.98°C cool and overnight lows about 0.61°C warm, so temperature comparisons across that date are indicative only. Pressure in this review is mean sea level pressure, the figure weather maps use. The August review quoted the raw station pressure (1017.0 hPa), which runs about 8.7 hPa lower because of the station's height, so the two months are not directly comparable on that figure. The independent pressure check uses ERA5 reanalysis (Copernicus, via the Open Meteo archive) for the grid point at 27.19°S 152.82°E, averaging daily mean sea level pressure over each September from 1995 to 2026. Climate driver values are from the BOM ENSO and climate driver pages for the week ending 27 September 2026. Where something is our interpretation rather than a measurement, we have said so in the text.