One of the most powerful supercomputers in the Southern Hemisphere earns its keep producing accurate, up-to-date and tailored weather forecasts to help farmers make better decisions.
The $13 million, IBM-built Fitzroy sits at the National Institute of Water and Atmospheric Research's Greta Point campus. From there it connects to a nationwide network of 300 weather stations to provide highly-localised forecasting for Niwa's Farm Met service.
Niwa chief scientist Murray Poulter says the service can provide separate, tailored forecasts for areas as little as 12km apart. The supercomputer is needed to break the country into smaller cells, Poulter says. Doubling the resolution - the distance between separate forecast areas - means a 16-fold increase in the number of calculations made on the computer.
He says reports are taken from the weather stations every two hours, so Farm Met provides immediate forecasts as well as reports on what is likely to happen over the next two days, the next six days and the next fortnight. There's also what he calls a heads-up report on the likely weather patterns over the next three months.
"Because we've tailored the forecasts to specific areas, the information provided is more relevant to the individual farmer.
"For example, with the short term forecasts we tell them what the rain is likely to do and what that means for the moisture in the soil. That may mean they need to irrigate or they can use the information to make decisions about using fertiliser. If flooding or high winds are likely farmers can decide whether they need to move stock. The data can also tell crop farmers the best time to harvest."
The longer-term forecasts are more in the nature of whether temperature and rainfall will be above average, below average or just average. This takes into account conditions out in the Pacific Ocean, such as the El Nino and La Nina patterns. Poulter says this information is useful for planning purposes and optimising farm operations. It provides insights such as how the soil conditions compare to previous years.
Poulter says Niwa developed Farm Met after studying ways it could provide weather forecasts in ways people could use them. There are products for other industries but the primary sector is probably the most important customer.
Farm Met is run as a subscription service. Farmers pay $370 a year per farm for the information, roughly a dollar a day. When signing up, they are asked to provide location details so the information can be tailored.
Information from Farm Met is provided in a series of easy to read graphs that farmers read using a browser.
Poulter says the service was designed from the start to be easy to use.
"We set it up so that it scales if farmers want to view the data on smartphones or tablets. It works with a wide range of devices".