How BOM seasonal outlooks are made: probability, accuracy, and how to use them

The Australian Bureau of Meteorology (BOM) publishes seasonal climate outlooks 3 to 4 times per year, covering 3-month rainfall and temperature probabilities for regions across Australia. These products are among the most sophisticated publicly available seasonal climate tools in the world. They are also frequently misunderstood.

What a BOM seasonal outlook actually says

A BOM seasonal outlook for rainfall expresses probability. A typical statement reads: "There is a 65% probability of above-median rainfall over the next three months in southern NSW."

This means:

• If you were to observe many seasons with the same climate indicators, 65% of those seasons would produce above-median rainfall
• There is still a 35% probability of at or below-median rainfall
• "Above median" means above the historical 50th percentile, which is not the same as "good" or "adequate" for cropping

What it does not say:

• The amount of rainfall that will occur
• When in the 3-month period the rainfall will fall
• Whether individual rainfall events will be well-distributed or concentrated
• Whether the outlook will verify in this specific year

How BOM generates its outlooks

BOM uses a statistical model (the POAMA model, now succeeded by the ACCESS-S model) that combines:

• Current sea surface temperatures in the Pacific, Indian Ocean, and Coral Sea
• The current SOI and IOD values
• The Madden-Julian Oscillation (MJO) phase
• Historical rainfall observations for each grid cell

The model outputs probability distributions for rainfall and temperature at each location, compared to the historical climatology. Where the model outputs a probability significantly above 50%, the outlook shows a "favourable" signal; where significantly below, an "unfavourable" signal.

Skill and accuracy

BOM publishes verification data for its seasonal outlooks going back decades. The skill statistics are publicly available and show:

• Forecasts of above or below-median seasonal rainfall verify correctly approximately 60 to 70% of the time in regions with strong ENSO influence
• Forecasts in regions with weaker ENSO or IOD influence (particularly northern and western Australia) have lower skill
• Temperature forecasts generally have higher skill than rainfall forecasts at seasonal timescales
• Skill is highest in autumn and winter forecasts; lower in spring forecasts when climate system predictability is lowest

Combining BOM outlooks with other indicators

The BOM seasonal outlook is best used in combination with other indicators rather than in isolation:

SOI and IOD simultaneously: A strongly positive SOI combined with a negative IOD (cool Indian Ocean anomaly, associated with WA drought) creates a mixed signal for WA and SA. Neither indicator dominates; the forecast skill is reduced. In these situations, wider probability distributions are appropriate.

Medium-term weather: 10 to 14 day outlooks from BOM and commercial weather services give much higher confidence for specific rain events. Use seasonal outlooks for strategic planning (input rates, marketing) and medium-term forecasts for operational decisions (spray timing, fertiliser application, harvest timing).

Practical planning applications

Reducing input rates in unfavourable outlooks: If the May outlook shows 30% probability of above-median June-August rainfall for your region, consider reducing nitrogen top-dress rates to levels appropriate for a below-average growing season. Do not eliminate nitrogen on this basis alone, but do not assume full yield potential.

Increasing marketing pre-harvest: An unfavourable seasonal outlook for spring rainfall increases the probability of below-average yield, which is bearish for Australian prices (tighter supply locally) but could be bearish or bullish for global prices depending on the wider production picture. Monitor WASDE alongside BOM outlooks.

On-farm risk management: An unfavourable spring outlook increases the priority of:

• Reviewing MPCI coverage and application deadlines
• Checking crop insurance covers frost if relevant to your region
• Reviewing forward contract position relative to expected lower yield

GRDC seasonal climate tools research consistently shows that farmers who systematically incorporate BOM seasonal outlooks into production and marketing decisions achieve 5 to 10% better financial outcomes over a decade than those who plan from climatological averages alone. The improvement comes not from correctly predicting each season but from consistently adjusting decisions in the direction the evidence suggests is more probable.