Recently the BoM started including monthly Outlook maps – not sure why but there you go.
The December rain Outlook did not turn out so flash either – a few earlier months were better.
But January has little hope already at day 10 and still raining over vast areas. For larger maps –
With al that flash computing gear & all those meteorologists & climate science( symposium) wannbbes, you have to wonder how much closer Piers Corbyn could get with his laptop?
The concept of forecasting percentage of mean rainfall is a useless statistic. I have said before rainfall has a statistical distribution close to a Poisson distribution where the standard deviation is close to the average. One can not have less rain than zero but the maximum of any month is (at least over 120 years of data) in the range 6 to 8 standard deviations.
I live in SEQld and here are my recordings for the last 4 months of 2014 in terms of average for each month Sep 328%, Oct 123%, Nov 93% & Dec 93%. Jan is part of the wet season (av 240mm) -it has rained (lightly- mainly at night) every day -the total to date is 28% of the average for the month. I forecast the monthly total will be in the range 90-100% of average
Sorry, too quick to post -in the first line I meant to put forecast the CHANCE of mean rainfall as a percentage is useless. They should try and forecast actual rainfall expressed as a percentage of the average. I understand that is what Abbot and Marohasy are attempting.
Average rainfall is one of those things that exposes most people’s ignorance of statistics.
Because rainfall is almost everywhere is a skewed distribution with a tail of higher rainfall amounts, on average (most of the time) you will have less than average rainfall.
Particularly true in the Perth summer, where most months it doesn’t rain at all.
Just noticed that the graphic is labelled ‘Chances of exceeding the median rainfall’.
In a skewed distribution (tail to the right), the median value is less than the mean (average).
TBF, median is a better indicator of whether rainfall will be above/below what you would get most years.
Phillip Bradley re your comment “exposes most people’s ignorance of statistics” -just for you en.wikipedia.org/wiki/Poisson_distribution
I think the BOM should stop messing about with statistics. Their attitude to faulty rain gauges is that any rain not recorded by a faulty gauge isn’t entered, which sounds sort of logical, but then the statistic is wrong. So local amateur gauges readings should be used. But no that would not be an official reading . My take? Use the unofficial readings and mark them as “unoficical” We have had 30 mills for two times in two years “not recorded” due to faulty guages. Other places I have read about have the same problems But of course that fits their picture of distorting climate wouldn’t it? I bet there are many more examples.
I wonder how Jennifer Marohasy and John Abbott are progressing with the research into artificial neural networks (ANNs), a form of Artificial Intelligence (AI), with potential as a new tool in particular their application to rainfall forecasting.
jennifermarohasy.com/input-selection-and-optimisation-for-monthly-rainfall-forecasting/
cementafriend,
Rainfall amount isn’t a Poisson Distribution.
Rainfall amount is usually modelled as a Gamma Distribution.
en.wikipedia.org/wiki/Gamma_distribution