Mistake jokes about the BOM (Bureau of Meteorology) are starting to fill the gap created by the absence of Irish, Scottish and other race-based jokes.
Australia’s BOM mistakes are developing an international reputation, and those mistakes caused the biggest single “non interest rate” force reducing inflation in the latest data. This caught my attention.
Accordingly, with the help of overseas experts it’s time to look more seriously at what is going wrong at the BOM, including the lessons learned in the 1974 Cyclone Tracy disaster in Darwin and then forgotten in the 1999 Sydney hailstone debacle plus many subsequent disasters including the flood disasters of 2023. No single storm event in 2023 dramatically depressed Australia’s inflation.
But on September 21 last year, the BOM issued a detailed and extensive long-term forecast that included maps and other data.
It had a similar qualification to those used by the Reserve Bank in forecasting that interest rates would stay low until 2024, but was much more detailed.
In clear terms, the BOM forecast:
“October to December rainfall is likely (60 to 80 per cent chance) to be below median for much of Australia, excluding most of northern and central WA and southern and central coastal NSW, where the chances of above or below median rainfall is roughly equal.
“Unusually low rainfall is at least twice as likely for parts of southern and northeastern Australia, with the chance of unusually low rainfall increasing to 3 times as likely for southeastern WA and parts of southern Victoria.”
Then came 19 chilling words for farmers: “Unusually low rainfall equates to the driest 20 per cent of October to December periods from 1981 to 2018.”
Memories of the droughts between 1981 and 2018 are enshrined in the DNA of many farmers, particularly lamb farmers. Accordingly, there was no time to lose. Farmers began selling their lambs at any price they could get. Prices collapsed.
In the Weekend Australian, the analysis showed that the biggest single price fall in the latest inflation figures was lamb (and goat) which fell 15 per cent – more than twice the next biggest fall, child care, which was slashed by government subsidy increases.
As we now know, it was a very wet October to December.
The BOM got it totally wrong.
The anger in farming communities – not just lamb – is white-hot. And places like Rochester in Victoria were flooded because the managers of the nearby dam that was full, refused to release water prior to the rain because they believed the BOM’s “it will be dry” forecasts.
The BOM has carefully prepared reasons for the September 21 mistakes, but my overseas weather friends tell me in weather forecasting you must go back to the basics and understand that artificial intelligence and computer models don’t always work and when you make a mistake, update the community quickly.
And that’s a lesson that wide areas of the corporate community will learn if they are enticed into believing artificial intelligence data is always correct.
According to the Insurance Council of Australia, our two greatest property damage storm events are Darwin’s Cyclone Tracy ($7.4bn in current dollars) and the 1999 Sydney hailstorm ($8.8bn).
Back in 1974, Darwin, the couple of BOM forecasters in the city had no radar but gave 36 hours warning that the cyclone would hit Darwin and not pass down to the west as almost all other cyclones had done in past decades. They underestimated the ferocity of the cyclone, but the early warning saved many lives.
The only radar in the town was at the airport, so the airport people faxed to the BOM Darwin people the images on their radar. The Darwin officials were experts on northern cyclones and were able to predict the Tracy hit using faxes and local knowledge. No computers or artificial intelligence.
In 1999, the BOM forecast that the looming hailstorm would avoid central Sydney. Separate to the BOM was a storm analyst who claims to have phoned the BOM to alert them that they were wrong but was ignored.
No one can be sure that the BOM was alerted, but the analyst is still living on the story and loves to show his videos of the 1999 storm.
In everyday forecasting, the BOM, like similar bodies around the world, have a series of models which they bring together in a process I would call a sophisticated form of averaging.
That’s accurate enough for most people, but increasingly industries in Australia that depend on weather are reducing their dependence on the BOM.
We now know that in Australian storm events, local knowledge is needed to interpret the data. Overseas experts say that models that warn of extreme events need to be studied by those experts rather than simply averaged, but that is a much higher cost operation.
The BOM will have different views, but they are simply too many mistakes taking place in a nation where a lot of our industries are related to weather.
One of the latest errors came via cyclone Jasper. The BOM expected Jasper to pass over Cairns and nearby areas on a track west.
But it hovered over the coastline, causing large, unexpected floods in Cairns and other areas. In national terms, it was not a major insurance event, but the local communities were unprepared and suffered. And because the damage took place outside the so called 48 hour “cyclone window” it did not come under the Commonwealth’s cyclone disaster insurance scheme and was classified as a flood event for insurance purposes.
Reinsurers around the world have now turned upwards their danger dial north of the Tropic of Capricorn and are now looking much more closely at areas of Brisbane.