The weather workshop hosted by the Northern Tablelands Local Land Services and Bureau of Meteorology was an eye-opener. There was no word on when the drought will break, but there was a honest conversation on the value of weather predictions and warnings on how they should be interpreted.
Subscribe now for unlimited access.
$0/
(min cost $0)
or signup to continue reading
Luke Shelley from the BOM's new Agriculture Group is currently doing the rounds of the Northern Tablelands, sharing weather insights but also keen to develop his role as a bridge between BOM scientists and users of its weather information.
READ ALSO:
Seeking feedback on how the information can be made more accessible, he took on board a suggestion that users be able to create their own dashboard profile on the BOM website, to automatically curate the location-based information that is relevant to their operations.
As he said there's masses of information available, but much of it is in science-speak and easily misinterpreted by the layperson. Mr Shelley put up an example of a district forecast that predicted possible rainfall of 15-30mm that day, coupled with a 100 per cent chance of rain.
Most people seeing that would anticipate at least 15 mm of rain. He's been waylaid in the past by farmers who said they'd been on the tractor all night planting in anticipation of the predicted rain, only to have 'bugger all' fall.
He said the two pieces of information are independent. The 100 per cent chance of rain actually means the bureau would expect a weather station in the district to register rain, but only 0.2 mm needs to fall to trigger that.
In this example there's a 50 per cent chance of receiving 15mm (or not), and a 25 per cent chance of receiving 30mm.
Mr Shelley also said concerns that the nearest weather forecast may be shown for a centre some distance away are unfounded. He explained that while the computer algorithm draws in data from weather stations it uses a grid basis to produce a forecast for ever six kilometres. Of course there are variations within the grid cells so he advised users to monitor the discrepancy between the forecast and their onsite readings, to relate forecasts to their position.
And when it comes to forecasts, he said the accuracy of predictions more than three or four days out drops off sharply, as ultimately there are weather elements in the atmosphere that can't be predicted a number of days out.
"Three days out is a forecast, seven days out is a guide, three months out there's little reliability."
Although many weather apps offer 28-day rainfall predictions the BOM does not, as it feels it can't provide that information with an acceptable degree of accuracy.
Mr Shelley advised app subscribers to 'check under the hood' to see where their app's weather data and interpretation are coming from. Some use BOM information, some put their own interpretation on the data, while others share predictions that are entirely computer-generated with no human intervention.
He suggested the easiest way to access local weather forecasts is via www.bom.gov.au/meteye/. It allows users to enter their location (even by latitude and longitude) and then drill down to three-hourly forecasts, handy for planning activities throughout the day.
The three-month climate outlooks are also often misinterpreted. They map out the chance of exceeding median rainfall. In Tenterfield's case the BOM says the district has a 25 per cent chance of receiving its average rainfall over winter (Mr Shelley said the term 'average' is interchangeable with 'median' in this context), not that it will receive 25 per cent of its average winter rainfall.
Arguably a more useful graph he produced shows Tenterfield with a 95 per cent chance of receiving 25mm over the period, 75 per cent chance of 50mm, 30 per cent chance of receiving 100mm etc, which may be a more useful tool on which to base farming decisions.
Mr Shelley said don't look at the seasonal outlook map and say 'it's going to be dry', but if you're planning to buy livestock on the prospect of decent rain, in this case you're going against the odds.
Furthermore if you need 200mm over the the next three months for sufficient soil moisture for a spring planting, he said "you're probably kidding yourself."
A new BOM tool modeling soil moisture is in the pipeline, due by the beginning of 2020.
Furthermore stronger weather patterns lead to better forecasts, so the models perform better at different times of the year and this is termed 'skill'.
Models have to be better than flipping a coin.
- Luke Shelley
"Models have to be better than flipping a coin," Mr Shelley said.
How much better they are is gauged by the accuracy rating. At this time of year, predictions have a 55-65 per cent accuracy, so 5-10 per cent better than flipping a coin.
"It's not the best thing in the world, but it's the best we've got," he said.
As a bridge between users and the bureau, Mr Shelley said his Agriculture Group may need to temper enthusiasm for big, data-intense websites that users with poor internet connections may have trouble accessing, although he liked the dashboard idea.
He said we'd see more changes over coming years in how the bureau presents information, making it more location-based than data-based. As for the forecasts themselves, however, he doesn't anticipate any advances for decades.
"Everyone wants a perfect forecast, but then everyone has it," he said.
This would eliminate weather risk across an industry, aligning market forces to a level playing field.
He instead advised informed risk takers operating in a climate of uncertainty to use that imperfection to their advantage.
He also said with considerations like stock on hand, market prices, available soil moisture, etc, if the seasonal outlook is the first thing someone is looking at to make a farming decision, that's about 10 steps too high.
