Weather Outlook and Evaluation- April 3 2013

How did I go for the first 3 months of this year?   On 27 November, I wrote:

“Outlook for 2013:

January: 3, 14, 18, 25, and the big one 30-31.

February: 7, 14, 20, and maybe 26th. (Rain will continue through the first three weeks of February, but with peaks near these days.)

March: 2, 7-8, maybe 13, maybe 17, 20, 31st .

April: 7, 13-14, 19.”  (Back then, I was looking at a range of +/- 5 days.)


On January 2, I refined this to:-

“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.”


On February 20 I tried a different method and predicted changes for:

“February 24-25-26-27 28

March 1-2-3-4; 6-7-8-9-10-11-1213-14-1516; 17-18-19-2021. ( I think there’ll be 2 major events, 6-11 March and 12-16 March with a follow up on 19 March.) and 27-28-29-30-31- 1 April.”  (Only 2 extra discrete events predicted- the others are within a day of previous predictions.)


Results:  (click on image for a closer look)qld march progress

Every predicted date matches events often exactly, mostly within 1 day, and (three times) within 2 days.

November 27 predictions:- 15 out of 15 correct.

January 2:-  4 out of 4 extra correct.

February 20:-  2 out of 2 extra correct.

Predictions of changes


No change

Unpredicted changes





91.3% accuracy.  From 27 November, using only the long range (160 days, without the 40 day refinements) the accuracy is still 65.2%.

I am satisfied that my method works.

160 days out, 100% of predicted changes happen.

40 days out, additional changes are predicted accurately.

These account for 91% of observed weather events.  (There have to be additional events, or weather patterns would never change.)

Learning as I go, I am going to refine predictions by trying Mean Sea level Pressure (MSLP) and Tmin  lagged 40 days instead of rain.  I am also now using the 2nd derivative of Tmin advanced 160 days (looking for de-celeration of temperature rise) to assist forecasting.  I am in the process of developing an objective method of testing the accuracy of my predictions.  More later.

Meanwhile, refined predictions for April to early May (>90% certain) and long range predictions to early September (about these dates, plus several others), for weather changes (cold change, pressure drop, or rain):-


3-4-5-6-7-8-9; 10-11-12; 13-14-15 (A succession of small changes)

17-18-19-20; 22-23-24



April 30-1-2-3-4-5

7-8-9-10-11-12-13-14 (unsettled for the first 2 weeks)




June: (more instability)



19- 20-21-22-23-24

27-28-29-30-1 July












28-29-30-31-Sept 1-2



An alternative method of predicting changes using second derivatives is shown here (up to 11 May), which may be easier to follow.  You’ll notice most dates are one day after those above.  We’ll see which method is better.  Allow +/- 1 day, but remember, by definition changes cannot be on consecutive days.April-may pred

And longer term (there will be extras, not discernible until 40 days out).May-Sep pred

I will now leave it at that while I concentrate on some other projects (including objective evaluation).


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: