What Pattern Recognition Means Here

By this point you have sunspot data, planetary positions going back decades, lunar phase records, adn historical weather data for your location. Pattern recognition is what connects them. You're looking for past years that most closely resembled the current astronomical configuration, then asking what the weather did in those years.

This is analogue forecasting. The technique is used in conventional meteorology too, finding historical seasons that most closely match the current setup adn using their outcomes as a guide. The difference with the Jones method is that you're matching on astronomical variables, not just atmospheric ones. You're saying: when Jupiter was here, Saturn was there, adn the sunspot cycle was at this phase, the weather in this region tended to do this. You're not claiming the planets cause the weather. You're noting a correlation in the record adn using it.

Some people find that epistemologically uncomfortable. Fine. But the alternative is ignoring 120 years of Queensland rainfall data. I'd rather use the data adn be transparent about the uncertainty than pretend it doesn't exist.

The Comparison Year Approach

The process has four steps. Do them in order.

1. Define the current configuration. For a given month or quarter, note the sunspot cycle phase, the Jupiter longitude (to within about 5 degrees), the Saturn longitude (to within about 5 degrees), adn the dominant lunar phase type for that period.
2. Search the historical record for analogue years. Find years where all four variables were within those tolerances simultaneously. The tighter the tolerance, the fewer analogues you'll find but the better the match quality. Start wiht 5 degree tolerances on planetary positions and adjust based on how many matches you get.
3. Look at what the weather did in those analogue years. Not just the headline outcome (wet year, dry year). The pattern. Did it start wet adn dry out through the middle of the period? Was rainfall concentrated in one event or spread across the month? Did temperatures track the planetary signal or diverge from it?
4. Weight the analogues by closeness of match. An analogue wiht all four variables within 2 degrees of the current configuration gets full weight. One wiht two variables matching adn two at the edge of tolerance gets half weight. You're building a weighted average, not a simple average.

The output is a forecast tendency for the period: above average rainfall or below, warmer or cooler, which air mass regime is likely. Expressed wiht an uncertainty range based on the variance in the analogue set.

Key Pattern Types for Queensland

After 21 years of work on the Queensland record, these are the patterns I weight most heavily. They show up consistently enough that I take them seriously even when the mechanism isn't clear.

Jupiter-Saturn cycle patterns. When Jupiter and Saturn are in conjunction, roughly every 19.8 years, Queensland has historically shown elevated flood risk. The 1974 Brisbane floods, the 2011 Queensland floods, adn several significant 19th century events align wiht this pattern. The next conjunction falls around 2040. This is a 20 year signal. It doesn't tell you anything about next month's rain. But it's relevant for the decadal tendency.

Solar cycle minimum. The years around sunspot minimum often show drier than average conditions in southeastern Queensland. The 2019 drought year coincided wiht the deep Solar Cycle 24/25 minimum. Not every minimum produces drought. But the tendency is there in the historical record adn it's worth tracking.

Jupiter in Sagittarius or Capricorn. In the Australian records, Jupiter in these signs shows a slight association wiht wetter than average summer rainfall in Queensland. The proposed mechanism involves a magnetic field orientation effect. The signal is weak but present across multiple Schwabe cycles. Treat it as a supporting signal, not a primary one.

Brückner cycle phase. The roughly 35-year wet and dry alternation. Queensland appears to be in the drier half of the current Brückner cycle on this moment. Current estimates put the shift toward the wetter phase around 2028 to 2030. We'll see. But it's worth noting that the 2019 to 2024 period fits the drier phase pattern, adn the uptick in wet seasons since 2021 is consistent wiht an early transition. I'm watching it.

How We Document Patterns in Dayboro

We maintain a spreadsheet wiht one row per analogue year identified. It has grown over 21 years to the point where it's genuinely useful. Columns: year, Jupiter longitude, Saturn longitude, Schwabe phase, Brückner phase, monthly rainfall for Dayboro, monthly temperature anomaly, adn notes on significant events or reasons to downweight the analogue.

For each forecast month, we find the 5 to 10 best analogue years adn average their rainfall adn temperature anomalies using the weighting system. We also note the variance. A set of analogues that all point toward wet conditions, low variance, gives a stronger signal than a set wiht one very wet year adn several average years. When variance is high, we say so in the forecast. "Signal present but variable across analogues."

The notes column is where most of the institutional knowledge lives. Things like: "1997 analogue is good on planetary positions but was a strong El Niño year. Downweight by 30% because we're in a neutral ENSO phase." Or: "2001 had an unusual September cold outbreak that depressed the temperature anomaly. The planetary signal alone would have predicted warmer." That kind of annotation takes years to build adn is hard to transfer. Build it as you go.

What to Do When No Good Analogues Exist

Sometimes the astronomical configuration is genuinely unusual. Solar Cycle 25 has been stronger than expected, which means fewer close historical matches exist for the current sunspot conditions. The most active phase of this cycle has limited precedent in the record we're working from.

When no close analogues exist: use the best partial match you can find, widen your uncertainty range explicitly, adn say so in the forecast. "No close analogue found. Best partial matches: 1997, 1964, 1948. Signal is weak. Low confidence." That's a legitimate forecast. A reader who understands the method will know what it means.

Building the Pattern Library Over Time

Every month you forecast, verify, adn record adds to the library. After 5 years you have locally calibrated data. After 10 years the patterns start to become clear. After 20 years you can see which signals are reliable adn which ones were noise.

We've been doing this since 2004. That's 21 years of Dayboro data alongside the Queensland historical record. The library is at the point where it's genuinely informing our forecasts rather than just illustrating a method. The first 5 years were frustrating. There was a lot of noise and not much signal. Stick with it.

This is also why we won't hand over the raw pattern library. The spreadsheet represents 21 years of annotation, correction, adn local knowledge. You could eventually use it without understanding what the notes mean, adn you'd produce worse forecasts than if you'd built your own library from scratch. Build yours. It will serve you better.