The Meeting Place

Weather forecasts may soon be accurate for a month

Scientists working on the next frontier of weather forecasting are hoping that weather conditions three-to-four weeks out will soon be as readily available as seven-day forecasts.

Having this type of weather information -- called subseasonal forecasts -- in the hands of the public and emergency managers can provide the critical lead time necessary to prepare for natural hazards like heat waves.

Scientists are leading the way to close this critical gap in the weather forecast system through the SubX project.

SubX is filling the gap between the prediction of weather and the prediction of seasonal conditions, which is guided by slowly evolving ocean conditions like sea surface temperatures and soil moisture and variability in the climate system that work on time scales of weeks.

"The SubX public database makes three-four week forecasts available right now and provides researchers the data infrastructure to investigate how to make them even better in the future," said Assistant Professor Kathleen Pegion.

SubX has already shown great promise forecasting weather conditions.

Would you trust weather forecasts a month in advance?


The Bureau of Meteorology has yet to perfect its maximum seven-day forecast, let alone a month in advance. 

The seven-day maximum daily temperature forecast for Brisbane is invariably adjusted several times over that short period. Even then, there can be a 1-3 degree difference in the actual maximum temperature for that day.

Climate is complex.

Even the weekly forcasts get changed half way thru the week.  They can't even get that right


Unfortunately, we will never (ever) see accurate forecasts in weather, for much longer than we have today. They might be closer, but it will never be "accurate". 

There are strong mathematical reasons (chaos/complexity/etc) why, and any detailed forecast can be thrown out by something as simple as a jet aircraft, a fire in a forest, a small volcanic eruption, or, as the saying goes, by a butterfly in Bolivia. One of the most interesting effects in recent times was the 911 attacks, which changed weather strongly in the US.

However, there are strong trend analyses that can reasonably predict general conditions for a few months, eg a wet winter. I see no reason to criticise the SubX work, but it can only predict with a modifier of "probably".

You are wrong Janus.  As computer models get more and more complex, requiring enormous amounts of computing power, then forecasts will become more and more accurate as they update at shorter and shorter intervals.

I agree with Ben but I suspect we are talking about quantum computers which as still in the labs at invention stage.

For those who doubt that weather predicting will imporve I direct you back to 50 years ago.  The weather forcast was never correct and long term was a joke.  These days its pretty good but sometimes the forcast hits an iceberg.  Give it time!


I woud be a bit wary about weather forecasts a month in advance these days. The changes we are seeing on a day to day basis make the whole concept rather tenuous.

Despite the great advancement in computer technology and the knowledge of weather patterns, the ability to give accurate forecasts is a distant dream. Our weather is in a state of flux and although we can get a better idea, surety is not a given. Only my opinion.


Forget it, they cannot even get it right for the next day! Often one has to look outside the window to be sure!

As a farmer I believe noted - this is the only profession where they can get it wrong over 90% of the time and yet they retain their jobs! It is amazing the number of people (mainly women) employed by news channels to read out rubbish forecasts - maybe they should just get the main newsreader to read out the Met Office forecast, and save huge recurring costs!