Forecasting Melbourne's grass pollen

Oct. 19, 2018


Here's an image from today's slide for Melbourne. Still some tree pollen about, a smidgen of grass, but not too many of those pesky plane tree trichomes I referred to in a previous blog.

The 9am reading on Tuesday October 16 said there'd been an average of 57 grass pollen grains per cubic meter of air in the preceding 24-hour period.

That's the first high grass pollen day of the 2018 counting season. A high day is one with more than 50 grass pollen grains per cubic meter of air.

For me, the first high day marks the start of the 2018 grass pollen season, a season I'm expecting will be light in comparison to past seasons.

Problem is that we’d forecast it to be a low. The feedback on social media was that people wanted to know more about how we forecast and why we’d got it wrong.

Of course, we'd love to be right all the time but nothing's harder to predict than the future.

And forecasting grass pollen is particularly tricky, as there are lots of factors at play.

To guide us, we use a quite complicated formula that integrates these factors and produces a best guess at the daily grass pollen count.

Sometimes our forecasts don’t come off quite as we expected, especially early in the season.

Believe me, there's nothing’s harder for a pollen counter to predict than that first high day.

Last week nicely illustrates the problem.

Our forecasts are in part based on the weather forecasts. But at this time of year two days with very similar weather forecasts can have very different grass pollen counts.

Take Sunday October 14. It was forecast to be mostly sunny with northeasterly winds 25 to 35 km/h and a maximum temperature of 27oC.

We forecast the next 24 hours would be moderate, as the model predicted 27 pollen grains per cubic meter of air.

A moderate day is one with between 20 and 50 grass pollen grains per cubic meter of air.

The 9am reading taken on Monday October 15, the end of the 24-hr period, was 19. So, a bit below what the model had predicted.

But, because it was less than 20 it was a low rather than a moderate day.

The weather forecast for Monday October 15 was for a day much like Sunday, mostly sunny with a maximum 28oC and northeasterly winds 25 to 35 km/h.

And the model forecast was for 15 grass pollen grains per cubic meter of air.

Same sort of day, same grass pollen forecast – right?

So, we forecast low but the 9am reading on Tuesday October 16 was high – ouch!

Like I say, very hard to pick at this time of year.

But that's why we bother counting pollen - forecast, then verify.

Forecast accuracy should improve from now but there's always room for improvement.

< Back