Sydney weather forecast gets less accurate one day before hot day

On Tuesday 8 Oct 2013 the Sydney Morning Herald reported – NSW temperatures may hit 40 degrees on Thursday prompting extreme fire warnings – In the article it says – Sydney is forecast to reach 37 degrees – Although the photo at the top of the article carried the misleading caption – Hot in the city: Sydney is likely to swelter on Thursday with temperatures predicted to reach 40 degrees.

Then on 9 Oct 2013 SMH has a headline – Sydney weather forecast revised to 39 degrees – Much mention in the article of high wind speeds including 100k gusts at Huskisson – yet no gust was recorded at Nowra or Point Perpendicular exceeding 70k.

So what actually happened on Thursday the 10th –
Sydney temperatures were 2° below forecast –
Canberra, Goulburn, Wagga Wagga, Albury were all 2° below forecast.
Further west temperatures were closer to forecasts
No wind gust was recorded at Nowra or Point Perpendicular exceeding 70k – what happened to the 100k figure that was concerning fire people?
Overall an inaccurate scaremongering updated forecast on the 9th.
Here is yesterdays effort at the SMH posted early evening – Sydney weather: heat to continue before big drop
What would be a rational statement about Sydney weather on 10 Oct 2013 – a very hot day in October similar but not quite as hot as some previous October days over a century or more ? And of course all of the BoM and media comment makes no mention that the ever expanding Sydney urban heat island could have had any effect at all yesterday.

10 thoughts on “Sydney weather forecast gets less accurate one day before hot day”

  1. On the Channel 7 evening news, presenter Chris Bath reported that Sydney August temperatures had reached 37ºC four times in 150 years.

    Then a representative from the BoM came on and said that the temperature was UNPRECEDENTED for Sydney.

    Alarmism?

  2. The ABC joined in the general climate alarmism , continuing to advise listeners that the temperature would reach 39 degrees, long after the BOM had reduced its prediction to 36-37. There is undoubtedly a UHI effect in Sydney as parts of Wahroonga reached only 27 degrees.

  3. Warwick

    Wind gusts recorded to 89 km/h at Norah Head with the change.

    Are you seriously claiming that the bureau of meteorology forecasters are deliberately overforecasting extreme maximum temperatures? What is your evidence of this?

    George

  4. No – you didn’t say over-enthusiastic – you said scaremongering – from Wikipedia – “is the use of fear to influence the opinions and actions of others towards some specific end. The feared object or subject is sometimes exaggerated, and the pattern of fear mongering is usually one of repetition, in order to continuously reinforce the intended effects of this tactic ..” – that is a serious allegation to make against public officers!

    The forecast was in error by 1.7 degrees – not 2 degrees as you claimed – for a forecast that was 15.3 degrees hotter than normal.

    To be taken seriously you can’t just summon up one or two random examples of over-forecasts – do you have a set of statistics that demonstrate that forecast maximum temperatures are systematically biased high?

  5. A summary of BoM temperature forecast checks on my HDD

    At odd times over a decade I have noticed a divergence between BoM temperature forecasts and actual weather. Sometimes these instances have prompted me to start a spreadsheet recording forecasts and later checking aginst actual daily max temperature. Boredom would set in causing me to usually cease these efforts after a week or so.
    The first experiment in Perth October 2002 ran for 7 days – 5 to 11 October.
    Over the seven days I logged the forecasts 1 day (day before), 2, 3 & 4 days.
    The sum of errors per day were (1 day first): 0.4, 0.97, 1.54, 2.54.
    So, warmer errors outweighed cooler errors over that week and skill improved closer to the weather.

    The second experiment in Perth started 28 Dec 2006 and ran for 15 days until 12 Jan 2007.
    Over the seven days I logged the forecasts 1 day (day before), 2, 3 & 4 days.
    The sum of errors per day were (1 day first): 0.56, 1.36, 1.56, 1.56.
    So, warmer errors outweighed cooler errors over that fortnight and skill improved closer to the weather.

    The third experiment was done using forecasts in the Canberra Times and ran for 8 days from 27 Nov 2012 to 4 Dec 2012 for the following towns – the error per day is noted and averaged 0.88 forecasts too warm –
    Brisbane – 1.49° per day forecasts too warm –
    Sydney – 1.27° per day forecasts too warm –
    Canberra – 0.14° per day forecasts too cool –
    Goulburn – 0.61° per day forecasts too warm –
    Hay – 1.02° per day forecasts too warm –
    Wagga Wagga – 1.72° per day forecasts too warm –
    Broken Hill – 0.16° per day forecasts too cool –
    Albury – 0.96° per day forecasts too warm –
    So, warmer errors outweighed cooler errors over those eight days.

    The fourth experiment was done using forecasts in the Canberra Times and ran for 11 days from 30 Dec 2012 to 9 Jan 2013 for the following towns – the overall average error per day was 0.46° forecasts too warm. The breakdown for the towns was.
    Adelaide -0.09 – so Adelaide weather was slightly warmer than forecast
    Brisbane 0.55
    Canberra 0.90
    Darwin 0.18
    Hobart -1.00 – so Hobart weather was warmer than forecast
    Melbourne 0.91
    Perth 1.09
    Sydney 1.18

  6. Very interesting Warwick. A few points:

    1. All experiments you report are October-January. So the warming bias is pretty clear in late spring and summer. Might be an idea to try one next winter and see if you get similar results.

    2. In your two multi-day forecast experiments, the BoM’s overpredicting of daily temperatures declines as the day gets closer, i.e., the warming bias is larger over a longer timeframe.

    3. This chimes in with a tendency for the BoM’s 3-month forecasts to be also too warm, suggesting that models assuming some consistent and continuing warming are infecting medium-term forecasts, and that this overprediction is propagating back into multi-day forecasts. However, the warming error then decreases as the prediction period shortens and the data on actual temperatures in the approaching weather systems take over in determining the forecast.

    There could be many reasons for all this. One might be that global warming models with excessive climate sensitivity are somehow affecting medium-term forecasts, e.g. by being initialized with 1990 or 2000 temperatures and then assuming a large amount of warming since then. Another source of error could be use of datasets such as ACORN that have too many warming adjustments in them and are thus pushing up estimates of current “normal” conditions.

    But these are just guesses. Who knows how exactly all this is happening? One thing is for sure, the BoM should be doing serious work to find out. It is no good their just whingeing that global warming is making prediction harder for them when it is much more likely that they are just making it harder for themselves by believing in it.

  7. I have not seen where the BoM archives weather forecasts.
    Until we can access past forecasts it is not easy to run checks. You either have to use newspapers or remember to record the forecast every day.
    I have also never seen where the BoM report on comparing weather forecasts against actual weather data.

Leave a Reply

Your email address will not be published.