Archive for January, 2013

Of heatwaves and flooding rains…

January 29, 2013

NatureBoy has just posted a comment to my blog, which I think is an excellent comment on the current state of affairs.

Here’s a little food for thought… How come there were alarmist predictions of the ‘unprecedented’ heatwave being whipped up into a media frenzy, with colours being added to maps and numbers being fudged to get the point across, yet the impending dilemma of torrential downpours causing massive flooding of towns and deaths up and down the eastern seaboard wasn’t given the same amount of “the sky is falling” prediction and response days out from the event? They can’t predict one disaster and ignore the other, even more destructive ones, surely? ;) If the same time and effort were injected into predicting and planning for ACTUAL weather events, then maybe, just maybe, people might not have died…


To give due credit to the Bureau’s forecasters, the system was difficult to analyse in the short term as it kept slowing down before moving on, however anyone could see a tropical rain depression just inland from the coast was going to bring very heavy rain.  It’s not as if it hasn’t happened before, indeed many times in the past.

And while I don’t necessarily think any amount of accurate predictions will prevent accidental death,  we could certainly be better prepared.   

Unfortunately, our masters in the Bureau, the CSIRO, and the Climate Commission are convinced (and would love to convince us) that global warming will impact Australia with heatwaves and drought.  Remember, we are supposed to be in the new “climate normal” where drought is the norm and dams will not be able to supply enough water for our cities….

Enough of your infatuation with global warming, Bureau bosses, it’s time to get back to your core business of weather forecasting.

Thanks, NatureBoy!

Outlook for 2013 – Update 21 January

January 21, 2013

Throughout the past 18 months or so that I have been studying weather cycles, I have found this study to be a continuing learning experience, and one that I have enjoyed very much.  This month is no different.

At the beginning of this month I repeated my predictions made back on 27 November, and extended out to the beginning of June.  For the year so far, I predicted weather enhancements or changes for:  January 3, possibly 8, 14, 18, possibly 21, 25, 30-31, all +/- 3 days.

Results so far: 

Here’s the plot of  the 3 day Tmin anomaly index and 3day rain anomaly index. Jan 2013 actual T & rain

Predicted dates to watch are in green.  Notice the temperature (red) peaks on January 5 (within 2 days of 3rd) , and 14 (spot on).   The rain peak on 16 relates to 14th, and 20th relates to 18th.  But what happened to the 8th?

I decided to check with the mean of the 9.00 a.m. Mean Sea Level Pressure (MSLP) at 8 of the 10 locations, and this is what it shows (the MSLP is inverted and scaled for comparison):Jan 2013 actual with mslp

Definitely a learning experience.  The black MSLP surprisingly comes 3 days before the Tmin peak but close to the predicted 3rd, however it matches 14th exactly and is 1 day off 18th.  But compare the predicted 8th with the massive (inverted) low pressure on 9-10th.  While perhaps I can be accused of moving the goal posts, and therefore I won’t claim the 8th as a success as it didn’t show in temperature change or rainfall, in future I will use MSLP as a useful third metric to identify weather change.

There was a weather disturbance here in Central Queensland this morning with cloud and showers, but I won’t claim it just yet.  Nevertheless, all of 3rd, 14th, and 18th I predicted on 27 November I will claim as successful.  That’s 3 out of 4 if I don’t use MSLP.  I won’t be so hard on myself in future.

Updated predictions for Sub-Tropical Queensland for January to June:

An enhancement (pressure change, temperature change, and/or rain) between these dates:

January:  23-27,  28-31.

February:  28 January-3, (I’m tipping the wet season to start around this time), 4-8, 12-16, 18-22, 23-27, 28-5 March (possibly 2 surges).

March:  5-10, 11-14, 15-22 (2 surges?), 25-29, 30- 2 April.

April:  3-6, 6-10, 10-12, 12-14, 15-21 (2-3 surges?), 24-29, 30-3 May.

May:  3-5??, 7-10, 11-15, 17-21, 25-27, 25-2 June (possibly 2 surges)

June: May 28-2, 3-8 (2 surges?), 9-12, 13-15, 16-19, 20-24, 27-30+.

SOI 30 day mean (to 21 January) is +2.0 (neutral).

Weekly NINO 3.4 Index (to 13 January) was -0.27 (neutral).

The Indian Ocean Dipole is currently +0.37 (neutral).

What record?

January 11, 2013

The Bureau of Meteorology (BOM) is at it again.

“On Monday the average maximum daily temperature record for Australia was broken at 40.33°C. The previous record, 40.17°C on 21 December 1972, was held for 40 years. The daily average maximum temperature yesterday (8 January 2013) is a close third at 40.11°C.”

This statement was contained in a press release on Wednesday 9 January 2013, quoting Assistant Director of Climate Information Services, Neil Plummer, and repeated with breathless enthusiasm by the media.

The average maximum daily temperature is a new term, one we haven’t seen before, to my knowledge.  However, if BOM can calculate an average annual or monthly temperature, then they probably can calculate an average daily temperature.

This announcement is crying out for a reality check.

Maxima are recorded at 9.00 a.m. on the day following, so Monday’s maxima would not be recorded until Tuesday morning, 8 January.  It normally takes a couple of months for data to be “quality assured”, checked for obvious mistakes, and then analysed for climate statements.  Yet Monday’s data, recorded on Tuesday, were accepted as correct and analysed one day later, and Tuesday’s data, recorded on Wednesday morning were processed and analysed in much less than 12 hours, and released to a gullible media and public.  Why the haste?

Australia’s climate analyses until last year were based on the High Quality (HQ) Network of weather stations, which had enormous problems and has now been superseded by the Australian Climate Observation Reference Network- Surface Air Temperatures (ACORN-SAT, or Acorn) which comprises daily data from 112 sites supposed to be the best available for length of record and quality siting.  One can only assume that these Acorn sites were the ones used to calculate the record average maximum daily temperature.  It didn’t take long to check daily maxima on Climate Data Online for 7 January from these sites.

Of the 112 sites, 3 have ceased recording and 1 had no data for the 7th January.

The remaining 108 sites had an average daily maximum temperature of:

35.91 degrees Celsius.

That’s 4.42 degrees short of 40.33, and 4.26 short of the record.

Pretty warm, but no cigar.

The median was 35.2C.  34 of 108 sites had maxima of 40.3 or more.  74 had less.

Perhaps they used a weighted average?  Here’s the temperature map for Monday 7 70113

See the dark brown area representing 45C and above?  Six (6) Acorn thermometers are in that area, recording 45C or more.

Perhaps BOM used stations other than Acorn?  If so, transparency demands they are revealed.  A map of their locations would help as they would have to be over represented in hotter parts of the country (in the dark red, grey, and brown areas on the map above).

Further, BOM needs to explain why they were used and not the official Acorn sites.  If analysis of Australia wide temperature is to be based on these sites, BOM needs to explain why they went to the trouble and expense of creating Acorn, if they are not going to use it.

Neil Plummer, you have some explaining to do.  Half truths, exaggerations, and misrepresentations are not acceptable.

Desperate for a Record

January 10, 2013

During the current heat wave, the Bureau of Meteorology (BOM) have been speculating about the likelihood of maximum temperature records being broken, including Queensland’s record of 49.5 degrees Celsius set in Birdsville on 24/12/1972.

Birdsville is a tiny but well known outpost in the far south west of Queensland.  The nearest other site is Bedourie 170 km away, but it has no data from before 1998.  There are no records for comparison on CDO within 290 km.

The Acorn record starts from 10/5/54, which according to the metadata is when observations commenced. On Climate Data Online Birdsville Police Station record starts on 1/1/57.  Where is the missing data?  It’s no excuse to say it hasn’t been digitised from the paper records because it obviously has been digitised into Acorn.  It should be fairly straightforward to copy it from Acorn into CDO.  By contrast, Brisbane has CDO data from the old Regional Office from 1887 yet Acorn’s creators have chosen to exclude this early data before 1948.

Acorn uses Police Station data exactly until the Airport site opens on 29 June 2000 when it adopts the Airport data exactly, with no adjustment despite good overlap- with occasional exceptions.  (Airport data is slightly less than Police Station data by on average 0.1 C.)

Before 1 September 1972 Police Station data was recorded in whole degrees Fahrenheit, therefore the uncertainty is +/- 0.3 C.  In the months following 1 September 1972, the date of metrication, temperatures are recorded in whole and half degrees Celsius, so the uncertainty is +/- 0.25 C.  Queensland’s “record” maximum temperature was recorded at Birdsville on 24 December 1972, so the actual temperature on this day could have been between 49.2 and 49.8C.

When the Airport site opened and its data used by Acorn, it was recording temperatures in whole degrees Celsius only, and the uncertainty then was +/- 0.5C.

Second highest temperature was 49C on 6/12/1981.

The next highest temperatures were 48.8 on 22/12/1990, and 48.7 on 17/11/90.

Interestingly, there are 5 other recordings of the next highest temperature, 48.5C, on 23/12/1973, 5/12/1981, 5 and 6 January 2004, and 30/11/2006.  Despite having an Airport recording of 47.9C on 5 January 2004, Acorn has chosen to use the higher Police Station recording instead of the Airport.birdsville max jan04

So the more extreme temperature was chosen!

Despite all expectations of the record being broken, the highest BOM could get out of Birdsville this summer has been 47.3C on 4 January.  That’s 19th, and not even close to the record.

To put this in context, taking 35 C as BOM’s latest standard for heatwave conditions, 34.4% of Birdsville’s maxima are 35C or more.  Birdsville’s January mean is 40.2 degrees.

Why has the BOM been so concerned and excited about the possibility of breaking maximum records?  This graph of the 365 day running mean of Birdsville maxima might give a clue:birdsville max acorn

They’re hoping temperatures will get back up to where they’re supposed to be, otherwise that trend line of nearly 0.25C per decade will change a bit.

Conclusion? They’re desperate and dateless.

How hot was 2012?

January 8, 2013

Yes it has been very hot in most parts of the country in recent weeks.  In light of the special Climate Statement released by the Bureau I thought I would show some context and some different data.

Here’s a graph of the daily Minimum temperature anomalies for the 10 sub-tropical Queensland locations I have been monitoring.  First, the 30 day means:30d tmin 2012

Next, 90 days means:90d tmin 2012

Finally, 365 day means:365d tmin 2012

For sub-tropical Queensland, the 10 year cooling trend is plainly obvious, and 2012 was the coolest year of the last 10 years.  The 90 day plot shows that every season of 2012 was below the long term average, and the 365 day mean has been below the long term average for 18 months.  It’s only in the past few weeks that the 30 day mean has risen above the long term average.

BOM supports this:min map 2012

Large areas of northern and eastern Australia had cooler  than normal minima in 2012.  In fact,

For the year as a whole, minimum temperatures were 0.28 °C below average. Minima were below average across much of northern and central Australia as well as New South Wales west of the Great Divide. A large part of this area recorded minima in the lowest decile. … Winter ranked as the third-coolest on record nationally for minimum temperatures (0.91 °C below average), and coolest for the Northern Territory, while autumn (0.93 °C below average) was the fourth-coolest nationally.

The large contrast between warm maxima and cool minima resulted in the mean diurnal temperature range being the third-highest on record. The more extreme years of 1994 and 2002 also saw severe drought over most of Australia.

As for maxima, above average temperatures were recorded in the south and south west.max map 2012

So, despite the scary stories about the heatwaves and how hot it has been for the past couple of months, not so bad after all.  Climate is regional and local, and variable, with naturally occurring great extremes.

To be fair, here’s BOM’s plot of annual and decadal means from mean graph

A word of warning: all BOM plots are derived from ACORN-SAT.  Past data should be viewed with caution, as it has been massively homogenised, and does not meet international standards for accuracy and station proximity in remote areas. ( )

In other words, we don’t have a clue.

Evaluation: August- December 2012 Predictions

January 2, 2013

Mid to late December:

While the predicted timings lined up very well, the “heavy rain” I expected did not eventuate in southern Queensland, but did fall at the right times on the Central Coast of North Queensland.  Rockhampton got a little of this on January 1.  Fronts did come through and there were widespread storms, but not heavy or general enough for me.

Starting from 1 August, I announced I would try to predict weather enhancements up to 5 months ahead.  By “enhancement”  I meant “an upper level disturbance or upper trough, which can lead to anything from extra clouds, isolated showers and storms, through to widespread heavy rain”. (I would now include a heatwave or bushfire conditions, followed by a change.)  My initial forecast was for enhancements in “early to mid-August (which may last a week or two), another late September to mid-October, and maybe late October to early November. Indications are that there will be heavier rain events starting in the period early to mid-November and early to mid-December” including specifically “a small rain event around 5-8 August and another probably larger one in the period 15-25 August, (I’m tipping close to 18-20 August).”

Over time I improved my methods, to the point I am confident in predicting specific dates +/-  5 days, and made more precise predictions.

What have I learnt?

That this method doesn’t necessarily predict rain (I admit I got carried away thinking it could) and I will leave predictions of amount of rain to the experts- the Bureau.

That I can accurately predict a change in the weather (indicated by a temperature change or rain) up to 160 days ahead.

That there are regularly repeating cycles of weather.

That this method may be used in southern parts of Australia as well.

Here is a plot showing predicted weather enhancements vs rainfall events (the average of 3 day sums as a percentage of the median across 10 locations) vs minimum temperature (the average of 3 day means of percentage anomalies from the long term mean in Kelvin).  The observed changes are colour coded according to the date that I made the predictions.Pred v actual 2012

16 out of 17 rain events are in the predicted time range.  (The one I missed –October 11-13- was my own stupid fault for not believing what the method indicated, because I had only just realised that a 3 day average of temperatures gave a far more accurate prediction than a 30 day average!)  That’s 94%.

Another 2 changes I did predict did not bring rain, and there were 2 changes that I didn’t predict because I was focussed on rain rather than weather change.

Therefore, I make the claim: I can predict when there will be a weather enhancement or change, indicated by a change in minimum temperature and/or a rainfall event, 160 days in advance, with an accuracy of +/- 5 days.  This can be refined 40 days out by analysing actual rainfall.  The weather change will affect sub-tropical Queensland, but also can include Central Coastal Queensland (Bowen to St Lawrence) and Northern New South Wales.

I again mention that rainfall is measured at 9.00 a.m. on the day after it falls.  As well, my method captures the average of 3 day rainfall and temperature  anomalies across 10 sites in subtropical Queensland.  I do not predict rain for specific locations.

Outlook for 2013 in sub-tropical Queensland

An enhancement (weather change, temperature change, or rain) on these dates +/- 5 days:

January:  3, possibly 8, 14, 18, possibly 21, 25, 30-31.

February:  Possibly 4, 7, 14, 20, and 26.

March:  2, possibly 5, 7-9, maybe 13, 17, 20, 31.

April:  7, 9, 11, 13, 15, 19, 27.

May: 4, 9, 13, 19, 29.

June: 6 (so far).

SOI 30 day mean (to 31 December) is -7.9 (neutral, but borderline El Nino if it stays below -80).

Weekly NINO 3.4 Index (to 30 December) was + 0.05 (neutral).

The Indian Ocean Dipole is currently +0.38 (neutral).