Page 18 of 18 < 1 2 ... 16 17 18
Topic Options
#1449899 - 23/01/2018 23:52 Re: Observations of climate variation [Re: Surly Bond]
Surly Bond Offline
Weatherzone Addict

Registered: 23/08/2003
Posts: 2104
Loc: Manilla, near Tamworth NSW
Australian climate Quasi-Biennial Oscillations.

[This material appeared originally as Post #810237 of 27 December 2009 (above). Since the graphs as posted are lost, due to the action of the "Photobucket" image-hosting web-site, I am re-posting the graphs from my records.]

The above graphs plot the time series of monthly data for the whole of Australia, smoothed with a gaussian window of half-width 6 months.
The two independent series of (a) mean maximum monthly temperature anomaly in degrees celsius and (b) total monthly rainfall anomaly in mm are plotted on the same graphs, but the scale for temperature is inverted for easy comparison.
One degree on the left axis corresponds to 10 mm on the right axis, but the zero lines may differ.
There is an obvious sine-wave cycle with a wavelength of between one and three years. (This is the "quasi-biennial cycle" that A.B. Pittock identified in 1971.)
Most peaks and troughs on these independent time series almost coincide, and their relative heights and depths tend to agree. In fact. the correlation between values of temperature and rainfall is poor, but the shapes of the sinusoidal curves match extremely well.
Peaks and troughs on the rainfall curve tend to lead those on the (inverted) temperature curve by one, two, or three months. In these graphs, I have lagged rainfall values one month, to show that many of the peaks and troughs are aligned.
Taking the whole of Australia in the last 60 years, it is fairly clear that:
* points of lowest maximum temperature have generally lagged about one month behind points of highest rainfall;
* points of highest maximum temperature have generally lagged about one month behind points of lowest rainfall.
Given this lag effect, times of lowest rainfall cannot be caused by times of highest temperature, but it is possible that times of highest temperature may be caused by times of lowest rainfall.
I find it plausible that temperature swings would closely follow rainfall swings (but in the opposite sense) due to lack of cooling by evapotranspiration in times of drought and effective cooling by evapotranspiration in times of deluge.
Data source.
The data was sourced at the following web-page.
http://reg.bom.gov.au/silo/products/cli_chg/
That web-page no longer exists.
Blog post.
This material is also posted (with a graph at higher resolution) in my blog.


Edited by Surly Bond (23/01/2018 23:56)
_________________________
Data are cheap; information is expensive!

Top
#1450425 - 27/01/2018 12:13 Re: Observations of climate variation [Re: Surly Bond]
Seira Offline
Meteorological Motor Mouth

Registered: 27/08/2003
Posts: 7449
Loc: Central Adelaide Hills
That is an awesome post Surly smile !!

Top
#1450453 - 27/01/2018 15:25 Re: Observations of climate variation [Re: Surly Bond]
Seira Offline
Meteorological Motor Mouth

Registered: 27/08/2003
Posts: 7449
Loc: Central Adelaide Hills
Originally Posted By: Surly Bond
graphs plot the time series of monthly data for the whole of Australia, smoothed with a gaussian window of half-width 6 months.

Bold - Are you able to elaborate on this in more detail smile ?

Top
#1450468 - 27/01/2018 16:58 Re: Observations of climate variation [Re: Surly Bond]
Surly Bond Offline
Weatherzone Addict

Registered: 23/08/2003
Posts: 2104
Loc: Manilla, near Tamworth NSW
Hi Seira
Nearly every item of climate data that you care to name jumps up or down from one observation to the next. Daily rainfall is an extreme case, but even monthly rainfall totals do it (in Australia).
To reveal a pattern usually requires time-averaging, or smoothing, or applying a sampling window - these terms are related.
I demonstrated the problem and the solution in a post describing the extreme drought of 2002 at Manilla:
https://climatebysurly.com/2014/06/16/profile-of-an-extreme-drought/
The sequence of raw monthly rainfall totals and mean temperatures showed no pattern at all. The pattern was scarcely any clearer when I removed the seasonal effect by plotting monthly anomalies rather than raw values. Then I tried plotting six-month average values, which I did not even show because the result was so poor. I achieved a clear pattern that showed drought onset and retreat only by applying a Gaussian smoothing function.
(A Gaussian smoothing function and a Binomial smoothing function are very similar.) The effect is like a moving average except that the resulting graph is smoother for the same effective width of sampling window.
My favorite Gaussian window uses the data points from six units before a given data point to six units after it. The weights applied to these thirteen points are:
0.009; 0.022; 0.044; 0.078; 0.116; 0.149; 0.164; 0.149; 0.116; 0.078; 0.044; 0.022; 0.009 (total=1.000).
I speak of this sampling window as having a half-width of six units: that is the width of the function where the weights are more than half of the maximum weight (0.164).
Now that these weights are entered in my spread sheets, I use them over and over again.
_________________________
Data are cheap; information is expensive!

Top
#1450473 - 27/01/2018 17:49 Re: Observations of climate variation [Re: Surly Bond]
Seira Offline
Meteorological Motor Mouth

Registered: 27/08/2003
Posts: 7449
Loc: Central Adelaide Hills
Thanks smile ; might have to do some further research on smoothing functions in the forseeable future smile .

Top
#1451239 - 01/02/2018 23:40 Re: Observations of climate variation [Re: Surly Bond]
Seira Offline
Meteorological Motor Mouth

Registered: 27/08/2003
Posts: 7449
Loc: Central Adelaide Hills
Surly, I wonder if you've had the time/resources/opportunity to run your data -- any of it -- through a Fourier Transform program to check for particular frequency peaks (signals). It might give you more details (from the microclimatic to broader scale) about what's going on in your area smile .


Edited by Seira (01/02/2018 23:40)

Top
#1451247 - 02/02/2018 00:21 Re: Observations of climate variation [Re: Surly Bond]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 441
Loc: Pinjarra Hills, Qld
Originally Posted By: Surly Bond
Australian climate Quasi-Biennial Oscillations.

[This material appeared originally as Post #810237 of 27 December 2009 (above). Since the graphs as posted are lost, due to the action of the "Photobucket" image-hosting web-site, I am re-posting the graphs from my records.]

The above graphs plot the time series of monthly data for the whole of Australia, smoothed with a gaussian window of half-width 6 months.
The two independent series of (a) mean maximum monthly temperature anomaly in degrees celsius and (b) total monthly rainfall anomaly in mm are plotted on the same graphs, but the scale for temperature is inverted for easy comparison.
One degree on the left axis corresponds to 10 mm on the right axis, but the zero lines may differ.
There is an obvious sine-wave cycle with a wavelength of between one and three years. (This is the "quasi-biennial cycle" that A.B. Pittock identified in 1971.)
Most peaks and troughs on these independent time series almost coincide, and their relative heights and depths tend to agree. In fact. the correlation between values of temperature and rainfall is poor, but the shapes of the sinusoidal curves match extremely well.
Peaks and troughs on the rainfall curve tend to lead those on the (inverted) temperature curve by one, two, or three months. In these graphs, I have lagged rainfall values one month, to show that many of the peaks and troughs are aligned.
Taking the whole of Australia in the last 60 years, it is fairly clear that:
* points of lowest maximum temperature have generally lagged about one month behind points of highest rainfall;
* points of highest maximum temperature have generally lagged about one month behind points of lowest rainfall.
Given this lag effect, times of lowest rainfall cannot be caused by times of highest temperature, but it is possible that times of highest temperature may be caused by times of lowest rainfall.
I find it plausible that temperature swings would closely follow rainfall swings (but in the opposite sense) due to lack of cooling by evapotranspiration in times of drought and effective cooling by evapotranspiration in times of deluge.
Data source.
The data was sourced at the following web-page.
http://reg.bom.gov.au/silo/products/cli_chg/
That web-page no longer exists.
Blog post.
This material is also posted (with a graph at higher resolution) in my blog.

I suggest the multiple plots for different decades need to be done with the same y axis range for max T and y axis range monthly rain anomaly for each decade to avoid potential misleading visual interpretation. Does not mean that discussion is flawed, just that graphs should be equal.
_________________________
Models are for estimating and gauges are for knowledge.

Top
#1451248 - 02/02/2018 00:40 Re: Observations of climate variation [Re: Surly Bond]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 441
Loc: Pinjarra Hills, Qld
Originally Posted By: Surly Bond
Hi Seira
Nearly every item of climate data that you care to name jumps up or down from one observation to the next. Daily rainfall is an extreme case, but even monthly rainfall totals do it (in Australia).
To reveal a pattern usually requires time-averaging, or smoothing, or applying a sampling window - these terms are related

Agree data has variations that impede identifying trends or correlations.
Smoothing is a method that helps, but not always as sometimes it masks important influences at different time scales.
Other methods, applicable for some applications such wet, dry extremes frequency can consider sampling these, such periodic maxima for wet interests (annual or other time periods months, decades), periodic minima for dry interests (again for a set timescale).
In some forms of data analysis for hydrology, even periodic max/min sampling can be a problem. For example the two big events within a period (say a year) would be recognised as one event. So partial series analysis which samples above a threshold (for example wets periods such as days > 20mm, or months > 100mm) can be helpful. Best approach depends on questions / hypothesis/ or simply exploration of what is being pursued.
_________________________
Models are for estimating and gauges are for knowledge.

Top
#1451327 - 02/02/2018 17:56 Re: Observations of climate variation [Re: Surly Bond]
Surly Bond Offline
Weatherzone Addict

Registered: 23/08/2003
Posts: 2104
Loc: Manilla, near Tamworth NSW
Flowin

I drew these smoothed time-series graphs to show how temperature and rainfall in these published data sets had patterns that were closely matched.
To do that, I had to smooth the data and choose suitable scales for plotting.
A suitable ratio of scales was one degree of temperature anomaly to 10 mm of monthly rainfall anomaly.
In the first drafts of all three graphs, I matched the two zero values and I used a grid of half-degree (and 5 mm) intervals.
In the resulting graphs a lot of space was wasted where one variable or the other went to extremes. For example, the Australian rainfall anomaly in the miraculous month of January 1974 was nearly +50 mm, which is 20 mm higher than any other month in this data set. On the other hand, the temperature anomalies of +1.5 degrees in 2002 and 2005 are nearly one degree higher than at any date before 1972.

I do not think the graphs are misleading.
You could easily sketch them out with common scales if that helps your understanding.
_________________________
Data are cheap; information is expensive!

Top
#1451364 - 02/02/2018 22:43 Re: Observations of climate variation [Re: Surly Bond]
Surly Bond Offline
Weatherzone Addict

Registered: 23/08/2003
Posts: 2104
Loc: Manilla, near Tamworth NSW
Seira

I have done Fourier transforms occasionally.
They are not very helpful, particularly in the case of "quasi-biennial oscillations". The period is about two years, or two-and-a-half years, but it tends to lock onto one year now and then (as you can see).
Then again, this record is far too short to resolve the periods of interest.
_________________________
Data are cheap; information is expensive!

Top
#1451412 - 03/02/2018 14:51 Re: Observations of climate variation [Re: Surly Bond]
Seira Offline
Meteorological Motor Mouth

Registered: 27/08/2003
Posts: 7449
Loc: Central Adelaide Hills
Originally Posted By: Surly Bond
Seira

I have done Fourier transforms occasionally.
They are not very helpful, particularly in the case of "quasi-biennial oscillations". The period is about two years, or two-and-a-half years, but it tends to lock onto one year now and then (as you can see).
Then again, this record is far too short to resolve the periods of interest.

There are two things I’d like to convey smile ; both are intended to be constructive.
First, regarding this statement:

Originally Posted By: Surly Bond
There is an obvious sine-wave cycle with a wavelength of between one and three years. (This is the "quasi-biennial cycle" that A.B. Pittock identified in 1971.)

Bold – How do you know this? It is possible you can quote A.B. Pittock (1971) and draw a replicable comparison with what they identified?

Second, when you mentioned “they are not very helpful” in relation the Fourier Transform analysis. With any data series, provided it is a factor N of some number, they can be very useful. I do not understand how it would not be, unless you have a very different approach or way of looking at it.

I'd be curious to understand on both counts smile .

Top
#1451437 - 03/02/2018 18:10 Re: Observations of climate variation [Re: Surly Bond]
Surly Bond Offline
Weatherzone Addict

Registered: 23/08/2003
Posts: 2104
Loc: Manilla, near Tamworth NSW
Hi Seira

I never did find Barrie Pittock's 1971 article in which he referred (I believe) to a quasi-biennial oscillation in Australian surface climate.
However, here is a detailed discussion of that particular topic:

"Historical El Nino/Southern Oscillation variability in the Australasian region" by Neville Nicholls, Chapter 7 (p151-173) in "El Nino: Historical and Paleoclimatic Aspects of the Southern Oscillation", Henry F.Diaz and Vera Markgraf (eds.), Cambridge U P, 1992, 476pp.

On p.158, Nicholls has a section headed "Biennial cycle" that refers to papers written in the 1970's, 1980's, and 1990's. He says:

"The biennial cycle is observed over the equatorial Pacific and Indian Oceans and is tightly phase-locked with the annual cycle. It varies in amplitude from cycle to cycle and sometimes changes phase. It is not strictly a 2-year cycle so it may be characterised better as a quasi-biennial cycle.....Rainfall over much of Australia displays a quasi-biennial cycle (e.g.Kidson 1925)."

Just like my graphs, in fact.

About Fourier transforms.
My experience is slight and a long time ago. Perhaps they can resolve much better than they used to, but I doubt it.
In this case, with only 60 years of records, the resolution at periods near one year can hardly be precise. But the "quasi-biennial" part is the killer. The period of oscillation is already known to be ill defined, so that is what the transform can be expected to show.
_________________________
Data are cheap; information is expensive!

Top
#1451531 - 04/02/2018 16:12 Re: Observations of climate variation [Re: Surly Bond]
Seira Offline
Meteorological Motor Mouth

Registered: 27/08/2003
Posts: 7449
Loc: Central Adelaide Hills
Originally Posted By: Surly Bond
Hi Seira

I never did find Barrie Pittock's 1971 article in which he referred (I believe) to a quasi-biennial oscillation in Australian surface climate.
However, here is a detailed discussion of that particular topic:

"Historical El Nino/Southern Oscillation variability in the Australasian region" by Neville Nicholls, Chapter 7 (p151-173) in "El Nino: Historical and Paleoclimatic Aspects of the Southern Oscillation", Henry F.Diaz and Vera Markgraf (eds.), Cambridge U P, 1992, 476pp.

On p.158, Nicholls has a section headed "Biennial cycle" that refers to papers written in the 1970's, 1980's, and 1990's. He says:

"The biennial cycle is observed over the equatorial Pacific and Indian Oceans and is tightly phase-locked with the annual cycle. It varies in amplitude from cycle to cycle and sometimes changes phase. It is not strictly a 2-year cycle so it may be characterised better as a quasi-biennial cycle.....Rainfall over much of Australia displays a quasi-biennial cycle (e.g.Kidson 1925)."

Just like my graphs, in fact.

I acknowledge the graphs, however I think there is much more to it.


Edited by Seira (04/02/2018 16:21)

Top
#1451703 - 05/02/2018 20:57 Re: Observations of climate variation [Re: Surly Bond]
Seira Offline
Meteorological Motor Mouth

Registered: 27/08/2003
Posts: 7449
Loc: Central Adelaide Hills
Hi Surly Bond smile ,

From what I recall, there are an array of scientific axioms/principles that are key to any scientific endeavour/idea that have to be covered (e.g. review of one's work by others, independently) before the work can be consider to have any reasonable traction. I was going to go through your last post point-by-point to address any concerns - in interpretation - I may have had about the connectivity of what was presented to actual published work...however I have subsequently decided not to because I am not on this forum to unnecessarily scrutinise ideas beyond the scope of contributors expressing their views/insights smile .


Edited by Seira (05/02/2018 21:00)

Top
#1451925 - 08/02/2018 18:12 Re: Observations of climate variation [Re: Surly Bond]
Surly Bond Offline
Weatherzone Addict

Registered: 23/08/2003
Posts: 2104
Loc: Manilla, near Tamworth NSW
Hi Seira

I am glad you are not planning a deeper analysis of my last post.
Forums such as this have their value in starting people thinking about things that they have not thought about before.
(That is why the prohibition on discussion of a particular topic here was a far-reaching disaster.)

For 25 years I played the peer-reviewed scientific publication game. I both wrote and reviewed as I was required to do. It is a great system, except that the publishing houses use the scientists, reviewers and editors as slave labour.
It did depend on trust, however. Now, peer-review seems not as trustworthy as it was.

Putting out scientific ideas in blogs and forums is an alternative way to advance science. If ideas are wrong, they will be corrected in time.
"Wikipedia" works splendidly on that basis, with errors usually corrected almost immediately.
_________________________
Data are cheap; information is expensive!

Top
#1451929 - 08/02/2018 18:53 Re: Observations of climate variation [Re: Seira]
Petros Offline
Meteorological Motor Mouth

Registered: 30/12/2002
Posts: 7005
Loc: Maffra, Central Gippsland, Vi...
Originally Posted By: Seira
That is an awesome post Surly smile !!


Yep, great presentation SB.

Top
#1452065 - 09/02/2018 22:09 Re: Observations of climate variation [Re: Surly Bond]
Seira Offline
Meteorological Motor Mouth

Registered: 27/08/2003
Posts: 7449
Loc: Central Adelaide Hills
When one is wanting to understand the truth of matters, be it the truth of current understanding, honesty, naivety, humility, data analysis, speculative reasoning, critical reasoning, insight, thought-provoking ideas, connotations (positive, negative or anything in-between, direct or indirect, implied or inferred), or wisdom...and so on...

...And one is not finding what is being presented clarifying of that truth, perhaps it is understandable that one would question both the veracity and usefulness of information presented.


Edited by Seira (09/02/2018 22:13)

Top
#1457179 - 14/03/2018 19:14 Re: Observations of climate variation [Re: Surly Bond]
Surly Bond Offline
Weatherzone Addict

Registered: 23/08/2003
Posts: 2104
Loc: Manilla, near Tamworth NSW
The evaluation of times when extremes in climate were more or less common could do with some attention.



Source:
https://climatebysurly.com/2018/03/14/annual-rainfall-extremes-at-manilla-nsw-v/
_________________________
Data are cheap; information is expensive!

Top
#1463921 - 16/05/2018 11:57 Re: Observations of climate variation [Re: Surly Bond]
Surly Bond Offline
Weatherzone Addict

Registered: 23/08/2003
Posts: 2104
Loc: Manilla, near Tamworth NSW


I have re-done this analysis that I posted here on 16 November 2017. By using 12-monthly rainfall totals, rather than annual rainfall totals, I have achieved twice the resolution and six times the stability of moment estimates.

https://climatebysurly.com/2018/05/15/moments-of-manillas-12-monthly-rainfall/
_________________________
Data are cheap; information is expensive!

Top
#1464626 - 30/05/2018 18:43 Re: Observations of climate variation [Re: Surly Bond]
Surly Bond Offline
Weatherzone Addict

Registered: 23/08/2003
Posts: 2104
Loc: Manilla, near Tamworth NSW


In which year of the 134 would you say that the rainfall at Manilla NSW was "Normal"?

https://climatebysurly.com/2018/05/29/relations-among-rainfall-moments/
_________________________
Data are cheap; information is expensive!

Top
Page 18 of 18 < 1 2 ... 16 17 18


Who's Online
4 registered (Warmfront, Sidney, Mathew, Stormy3), 339 Guests and 3 Spiders online.
Key: Admin, Global Mod, Mod
Today's Birthdays
Christopher M.J.T, hiho, mscactus, Yangvews
Forum Stats
29589 Members
32 Forums
23924 Topics
1493194 Posts

Max Online: 2925 @ 02/02/2011 22:23
Satellite Image