Will BOM bomb its forecasts and could AI do better?

Our beach-side enclave received nothing that day as forecast but the next "clear day" miraculously produced 60 mm in the gauge.

If they can't get it right a day or a week in advance, how can they, the CSIRO, the Climate Council or Albo tell us what the climate might be like by the end of the century?

They can't. Sorry to disillusion all the hand-wringers and news presenters who keep claiming recent weather events are "unprecedented" because of climate change. History says otherwise and here are some real facts: There have always been cyclones and floods and they have not become more severe, as BOM's own records confirm.

We have had floods, wild weather, droughts and cyclones since early settlement, including the devastating 1974 Brisbane flood, Cyclone Althea which hit Townsville on Christmas Eve in 1971 with three deaths, Cyclone Tracy which flattened Darwin with 71 fatalities on Christmas Day in 1974, and Cyclone Justin in 1917 with 34 fatalities. But worse than any of those - "Cyclone Mahina was the deadliest cyclone in recorded Australian history, and also likely the most intense tropical cyclone ever recorded in the Southern Hemisphere. Mahina struck Bathurst Bay, Cape York Peninsula, Queensland, on 4 March 1899, and its winds and enormous storm surge combined to kill more than 300 people".

Brisbane experienced the Great Flood of 1893 - the Black February flood - when the Brisbane River burst its banks on three occasions with about 35 fatalities.The highest 24-hour rainfall total for anywhere in Australia was 907 mm at Crohamhurst, in the Brisbane catchment, on February 3, 1893.

Nothing in recent times comes close to those; there were no coal-fired power stations back then, and as former chief scientist Alan Finkel admitted to a Senate hearing several years ago, if Australia cut all its CO2 emissions immediately, its impact on world climate would be negligible.

But when it comes to forecasting, a pair of scientists critical of BOM think we could do better if we let Artificial Intelligence (AI) in on the act.

Well, I'm no fan, as I wrote in an article titled Beyond Mad Max with Dystopian AI: "... the charmless i-bot that answers almost all government and major business phone calls, the source behind our internet searches and algorithms, the brain behind the helpful tools that do a student's assignments or write a novel (badly) or create nude images of anybody from a normal photograph .. get the picture?...

But is there a useful place for AI in weather forecasting? Yes, according to the very active and productive Dr Jennifer Marohasy, an Australian biologist who has devoted many years to studying national climate records as well as corals on the Great Barrier Reef, and her husband Dr John Abbot, a highly qualified scientist and senior fellow at the Institute of Public Affairs.

In her December newsletter, Dr Marohasy stated, "If the Bureau was serious about weather and climate forecasting it would give up with its pretend General Circulation Models.

John Abbot writes:

Weather and climate forecasts using Artificial Intelligence (AI), particularly artificial neural networks (ANNs), are more reliable and skilful than traditional methods using General Circulation Models (GCMs). I have been publishing in the international peer-reviewed literature and running courses for Indonesian meteorologists, demonstrating the technique for over a decade. I first visited the Australian Bureau of Meteorology with Jennifer Marohasy in August 2011, to show the skill of my monthly forecasts using AI, relative to the Bureau's using General Circulation Models that attempt to mimic the physical process. These are the same models used by the IPCC and have proven unable to forecast more than 4 days in advance.

The Bureau has steadfastly refused to consider AI despite it showing a capacity to produce skilful location-specific rainfall forecasts from one day (models developed by Google) to 18 months in advance (models developed by Abbot and Marohasy)."

I'll leave that debate to the scientists, because I've found a seemingly reliable weather forecaster right here at home. Donnie is an interloper who set up camp under our back deck some years ago and who occasionally appears sunning his lithe, muscular body on the deck or on a flat granite rock beside the fish pond as if he owns the place. A week ago we had a Mexican stand-off when I almost tripped over him right at the back door and he refused to move. But I knew from experience that within a few days we could count on some prolonged rain.

Who needs BOM or AI when we have Donnie the Water Dragon in our own backyard?