Anologues!

[summer blizzard]

I have opened this thread in order to see how much use anologues actually are (i.e. how far can we accurately predict with them), while this thread will only be related to the indicee values, because anologues are subjective (i.e. you can take La Nina to mean higher pressure in one are, but somebody may take it to mean lower pressure in that area...).

I will do this by posting both the duo-monthly anologues, and the predicted indicee values for the October to March period, at the end of the month, allowing a 0.3 margin of error, i will then deem the forecast to be a success or failure and post the next set of anologues and predicted indicees, however at the end of that month, i will rate both, because it could be that several anologues were present in both sets of data, over a longer time period, this would allow me to find a persistant anologue which would allow me to better my future forecasts for the coming period.

Here is the first set of data...

Oct to March

MEI

Sep

1999

-1.1 (Correct)

-1.2

-1.4

-1.3

-1.3

-1.1

http://www.cdc.noaa.gov/Correlation/mei.data

QBO

Sep

2005

1996

-2.1 (Correct)

-2.1

-1.5

-0.6

-0.6

0.1

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

AO

Sep

2003

2002

2000

1997

1985

1980

1973

1970

1967

1962

1954

0 (Wrong)

-0.6

-1.1

-1.4

-1.3

0.1

http://www.cpc.noaa.gov/products/precip/CW...ent.ascii.table

As you can see from this, all three forecasts were within 0.4 of the forecast value, however as my margin of error is 0.3, i did not predict the October AO value well enougth, thus on a A-F scale, i would rate this forecast a C.

I now have the updated anologues, with associated indicee values and also the median CET value for the anologues which are present in more than one indicee...

Oct to March

MEI

Sep

1999

-1.1 (Correct)

-1.2

-1.4

-1.3

-1.3

-1.1

Oct

1999

1970

1964

1954

-1

-1.1

-0.9

-0.7

-0.9

http://www.cdc.noaa.gov/Correlation/mei.data

QBO

Sep

2005

1996

-2.1 (Correct)

-2.1

-1.5

-0.6

-0.6

0.1

Oct

2005

-2.5

-2.1

-1.7

-1.1

-0.1

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

AO

Sep

2003

2002

2000

1997

1985

1980

1973

1970

1967

1962

1954

0 (Wrong)

-0.6

-1.1

-1.4

-1.3

0.1

Oct

2000

1975

1973

1970

1969

1955

1954

-0.3

-0.4

-0.9

0.2

-0.5

Temperature Outcome

1954

1970

November: 7.4C, 0.5C above average

December: 5.6C, 0.5C above average

January: 3.6C, 0.6C below average

February: 2.9C, 1.3C below average

March: 4.1C, 2.2C below average

Please feel free to give your own opinion on the use of anologues.

Edited November 8, 2007 by summer blizzard

[Stratos Ferric]

SB, it might help if you explain a bit more clearly what your various numbers are.

[BrickFielder]

I had 1970 down as an analgue based on the time of year and direction of change of the QBO and El Nino indices. As to how usefull they are then I am not so sure. The stratosphere is a lot colder, the ice melt during the summer was a lot more and these things are likely to throw the patterns in a different direction.

[summer blizzard]

Oct to March

MEI

Sep

1999

-1.1 (Correct)

-1.2 (Correct)

-1.4

-1.3

-1.3

-1.1

Oct

1999

1970

1964

1954

-1 (Correct)

-1.1

-0.9

-0.7

-0.9

http://www.cdc.noaa.gov/Correlation/mei.data

QBO

Sep

2005

1996

-2.1 (Correct)

-2.1 (Correct)

-1.5

-0.6

-0.6

0.1

Oct

2005

-2.5 (Correct)

-2.1

-1.7

-1.1

-0.1

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

AO

Sep

2003

2002

2000

1997

1985

1980

1973

1970

1967

1962

1954

0 (Wrong)

-0.6 (Correct)

-1.1

-1.4

-1.3

0.1

Oct

2000

1975

1973

1970

1969

1955

1954

-0.3 (Correct)

-0.4

-0.9

0.2

-0.5

Temperature Outcome

1954

1970

November: 7.4C, 0.5C above average (Correct, actual was 7.3C)

December: 5.6C, 0.5C above average

January: 3.6C, 0.6C below average

February: 2.9C, 1.3C below average

March: 4.1C, 2.2C below average

Please feel free to give your own opinion on the use of anologues.

[summer blizzard]

As you can see above, the use of anologues at the one and two month timeframe was succesful, as such, we could use Novembers data to get a clear picture of the indices for the coming winter.

Here is the November data...

MEI

Sep

1999

Oct to March

-1.1 (Correct)

-1.2 (Correct)

-1.4

-1.3

-1.3

-1.1

Oct

1999

1970

1964

1954

Nov to March

-1 (Correct)

-1.1

-0.9

-0.7

-0.9

Nov

1999

1974

1964

Dec to March

-1.1

-0.8

-0.6

-0.5

http://www.cdc.noaa.gov/Correlation/mei.data

QBO

Sep

2005

1996

Oct to March

-2.1 (Correct)

-2.1 (Correct)

-1.5

-0.6

-0.6

0.1

Oct

2005

Nov to March

-2.5 (Correct)

-2.1

-1.7

-1.1

-0.1

Nov

N/A (Trend)

Dec to March

-2.1

2005

-1.4

-0.8

+0.2

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

AO

Sep

2003

2002

2000

1997

1985

1980

1973

1970

1967

1962

1954

Oct to March

0 (Wrong)

-0.6

-1.1

-1.4

-1.3

0.1

Oct

2000

1975

1973

1970

1969

1955

1954

Nov to March

-0.3 (Correct)

-0.4

-0.9

0.2

-0.5

Nov

1995

1964

1954

Dec to March

-1.1

-1.2

-1.4

-1.4

http://www.cpc.noaa.gov/products/precip/CW...ent.ascii.table

Temperature Outcome

1964

December: 3.5C, 1.6C below average

January: 3.2C, 1C below average

February: 3C, 1.2C below average

March: 5.1C, 1.2C below average

[summer blizzard]

MEI

Nov

1999

1974

1964

Dec to March

-1.1 (Correct)

-0.8 (Correct)

-0.6

-0.5

http://www.cdc.noaa.gov/Correlation/mei.data

QBO

Nov

N/A (Trend)

Dec to March

-2.1 (Wrong)

2005

-1.4 (Wrong)

-0.8

+0.2

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

AO

Nov

1995

1964

1954

Dec to March

-1.1 (Wrong)

-1.2 (Wrong)

-1.4

-1.4

http://www.cpc.noaa.gov/products/precip/CW...ent.ascii.table

Temperature Outcome

1964

December: 3.5C, 1.6C below average (Wrong)

January: 3.2C, 1C below average (Wrong)

February: 3C, 1.2C below average

March: 5.1C, 1.2C below average

The main points of analysis were that the MEI values were once again predicted within an acceptable margin of error, while the QBO (easterly trades) collapsed much quicker than expected, and the AO values were just wrong.

This in turn led to my temperature profile anologue (1964), also being wrong and as a result of the QBO weakening quicker than forecast, i would surmise that this has allowed the Polar Vortex to be much stronger than forecast, though it should be noted that with December and Janury averaging an AO value of +0.8, this would actually be considered near neutral.

[summer blizzard]

Looking ahead to summer now, i am now going to include the forecast data to August.

MEI

2000

1999

1989

1971

1957

1955

1951

1950

March to August

-0.7

-0.7

-0.4

-0.8

-0.6

-0.6

http://www.cdc.noaa.gov/Correlation/mei.data

[Stratos Ferric]

The main points of analysis were that the MEI values were once again predicted within an acceptable margin of error, while the QBO (easterly trades) collapsed much quicker than expected, and the AO values were just wrong.

This in turn led to my temperature profile anologue (1964), also being wrong and as a result of the QBO weakening quicker than forecast, i would surmise that this has allowed the Polar Vortex to be much stronger than forecast, though it should be noted that with December and Janury averaging an AO value of +0.8, this would actually be considered near neutral.

So are we concluding that the method is flawed yet? How might it be improved?

[summer blizzard]

So are we concluding that the method is flawed yet? How might it be improved?

I will update later on today hopefully, if not soon...

However, from this the conclusions thus far are that some teleconnections have a much better anologue response than others (MEI as opposed to the AO for example), however given that we saw a record low QBO value, there are no comperable anologues for that which is why the anologues could not forecast that.

[summer blizzard]

QBO

2006

2000

1997

1990

1988

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

MEI

1976

1956

1950

June to November

[Gavin P]

So, combining all of those years, this summer will either be amazing or terrible?

[summer blizzard]

Sorry, i did'nt have time to finish, after applying a linier ajjustment, here are the results.

QBO

2006

1997

1990

1988

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

MEI

1976

1956

1950

http://www.cdc.noaa.gov/Correlation/mei.data

June to November

QBO

0.6

0.8

0.7

0.7

0.6

0.6

MEI

-0.4

-0.2

0

0.1

0

-0.2

Temperature profile

June: 15.1C, 1C above average

July: 17.2C, 0.7C above average

August: 16.2C, average

September: 14.9C, 1.2C above average

October: 11.2C, 0.8C above average

November: 6.8C, 0.1C below average

[summer blizzard]

Here are the anologues which matched the April-May period as duo-monthly anologues...

MEI

1999

1989

1976

1967

1963

http://www.cdc.noaa.gov/Correlation/mei.data

QBO

2006

1997

1990

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

After applying a linier ajustment, here are the expected monthly readings for the QBO and MEI...

MEI

0.1

0.1

0.2

0.2

0.2

0.1

QBO

1.6

1.8

1.6

1.5

1.3

1.1

Based on the data above, here would be my forecast for the next six months...

Forecast...

June: Above average temperature, Above average rainfall

July: Above average temperature, Average rainfall

August: Below average temperature, Above average rainfall

September: Below average temperature, Average rainfall

October: Below average temperature, Average rainfall

November: Below average temperature, Above average rainfall

As you can see, the QBO anologues have not changed at all, and it is interesting to see 1976 and 2006 both as anologues for the current tri-monthly period.

[summer blizzard]

Here are the anologues which matched the May-June period as duo-monthly anologues...

MEI

1996

1985

1976

1963

1961

1951

http://www.cdc.noaa.gov/Correlation/mei.data

QBO

2006

2004

2002

1999

1997

1995

1990

1985

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

After applying a linier ajustment, here are the expected monthly readings for the QBO and MEI...

MEI

+0.3

+0.4

+0.3

+0.4

+0.5

+0.3

QBO

Will complete at later date..

Interestingly, it looks like we will border on a weak El Nino for the next six months, i believe the chance of a weak El Nino being declared is now 40%, with neutral conditions favoured.

[johnholmes]

from IRI 16 July

It indicates a 75% probability for ENSO-neutral conditions in the coming season.

[Gavin P]

I still think we'll see a neutral autumn and early wimter and then probably trending towards weak or moderate La Nina late winter and spring 2009.

[Weather-history]

When you say "anologue" do you actually mean "analogue"?

[summer blizzard]

When you say "anologue" do you actually mean "analogue"?

Yes!

[summer blizzard]

Here are the anologues which matched the May-June period as duo-monthly anologues...

MEI

1996

1985

1976

1963

1961

1951

http://www.cdc.noaa.gov/Correlation/mei.data

QBO

2006

2004

1999

1997

1990

http://www.cpc.ncep.noaa.gov/data/indices/qbo.u30.index

After applying a linier ajustment, here are the expected monthly readings for the QBO and MEI...

MEI

+0.3

+0.4

+0.3

+0.4

+0.5

+0.3

QBO

+1.5

+1.4

+1.3

+1.1

+0.7

+0.2

Interestingly, it looks like we will border on a weak El Nino for the next six months, i believe the chance of a weak El Nino being declared is now 40%, with neutral conditions favoured.

August: Above average rainfall, close to average temperature

September: Above average rainfall, close to average temperature

October: Close to average rainfall, below average temperature

November: Below average rainfall, close to average temperature

December: Below average rainfall, close to average temperature

January: Above average rainfall, close to average temperature