The Bureau of Meteorology (BOM) has reported several recent advances in improving the accuracy of weather forecasts — most notably for temperature predictions. Over the last several decades, BOM’s temperature forecasts have improved at a rate of approximately one day per decade since the 1970s. This advancement is a testament to our growing capacity to more accurately model and forecast weather patterns. This innovative development is key to the resiliency and economic development of all sectors — including agriculture, tourism and urban communities.
In its latest verification report BOM just disclosed the hottest temperatures for the 2024-25 financial year. They discovered that their forecasts were correct 91 percent of the time, within 2 degrees Celsius of the observed temperatures. There is new, exciting technology being developed every day to advance forecasting techniques. Consequently, the BOM is building public confidence in meteorological information through the delivery of accurate weather forecasts.
Temperature Forecast Improvements
The BOM’s analysis shows the difference an improved weather model has made in temperature forecast accuracy. For instance, a detailed review of Melbourne’s forecasts for April 2025 revealed that the mean error associated with persistence forecasting—where past weather patterns inform future predictions—stood at 3.1 degrees Celsius when looking just one day ahead. Long-term forecast Sono, however, has suffered forecasting errors of up to 2.5 degrees Celsius for one-day forecasts in general Sydney. For the seven-day outlook, errors grew to 3.1 C.
It’s important to remember the longer the forecast period, the less accurate forecasts become. As an example, even at a nine-day forecast horizon, BOM reported that its predictions for temperature are still better than climatological averages. This unfortunate trend highlights the critical need for ongoing investments in new and innovative meteorological modeling techniques to improve predictive capabilities over even longer time horizons.
The Challenge of Rainfall Forecasts
Forecasting rainfall is much more complicated than forecasting changes in temperature. For example, the BOM describes precipitation chances with terms like “likely to shower.” This expression means that there’s a 65 to 84 percent chance of rain on the same day. Using this terminology is important because it enables the public to better understand predictability of a forecast.
Checking the accuracy of rainfall forecasts can be as simple. You can take a glance at their accuracy by comparing their predicted rainy day ratio with what has really happened. Impressively, BOM found that its rainfall forecasts maintained an accuracy within three to four percentage points, even amid the unpredictable nature of precipitation. The group noted that its predictions were able to forecast precipitation with timeliness and accuracy as much as eight days out, outperforming strictly climatological averages.
Recent Trends in Forecasting
Temperature forecasts have greatly increased in skill in recent years. Over the last 10 years, we’ve increased the reliability of these forecasts by an additional two days. The BOM has pointed out that its current four-day forecasts are as reliable as the two-day forecasts it published back in 2015. This progress is indicative of continued advancements in meteorological technology and methodologies.
Interestingly, residents of Darwin with limited meteorological knowledge could potentially match forecast accuracy by simply assuming that the maximum temperature will align with long-term averages. This observation emphasizes that while sophisticated models yield reliable data, basic assumptions can still provide satisfactory outcomes for general weather predictions.
