85
GEOG5426 Statistics in paleoclimatology

Class 9, Statistics in paleoclimatology

Embed Size (px)

Citation preview

Page 1: Class 9, Statistics in paleoclimatology

GEOG5426 Statistics in paleoclimatology

Page 2: Class 9, Statistics in paleoclimatology

November 24

Brief (15-minute) summaries of project topics.

(1) What are the most important features of the modern climate in your region?

(2) What proxies are available in your region, over the time interval specified? How are they related to climate? and

(3) How different were past climates from modern conditions? Why is that important?

Page 3: Class 9, Statistics in paleoclimatology

GEOG5426 Statistics in paleoclimatology

Page 4: Class 9, Statistics in paleoclimatology
Page 5: Class 9, Statistics in paleoclimatology
Page 6: Class 9, Statistics in paleoclimatology

A time series is a set of observations ordered in time.

Page 7: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000

Year (A.D.)

-10

-5

0

5

10

PDSI

Page 8: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000

Year (A.D.)

-10

-5

0

5

10

PDSI

resolutionannual

Page 9: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000

Year (A.D.)

-10

-5

0

5

10

PDSI

chronological uncertaintysub-annual

Page 10: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000

Year (A.D.)

-10

-5

0

5

10

PDSItime spanlast century

Page 11: Class 9, Statistics in paleoclimatology

Variance

samplesize

variance observation

sample mean

Page 12: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000

Year (A.D.)

-10

-5

0

5

10

PDSI

-3

-2

-1

0

1

2

3

Ring

wid

th

PDSI Ringwidth

drought-sensitive tree-ring records Eastern Canadian Rockies

St. George et al., (2009), Journal of Climate

Page 13: Class 9, Statistics in paleoclimatology

Correlation Pearson’s product-moment correlation

covariance

product of both standard deviations

Page 14: Class 9, Statistics in paleoclimatology
Page 15: Class 9, Statistics in paleoclimatology

Source: Wikipedia

r = 0.816

Page 16: Class 9, Statistics in paleoclimatology

statisticalsignificance

practicalsignificance

Page 17: Class 9, Statistics in paleoclimatology

nnumber of observations

Page 18: Class 9, Statistics in paleoclimatology

Source: Wikipedia

Page 19: Class 9, Statistics in paleoclimatology

T I M E S E R I E S T E R M I N O L O G Y

Page 20: Class 9, Statistics in paleoclimatology

Trends are progressive increases or decreases in the levels of a particular climate variable.

Bartlein 2006

Page 21: Class 9, Statistics in paleoclimatology

Trends Bartlein (2006)

Page 22: Class 9, Statistics in paleoclimatology

Steps are abrupt transitions from one level to the other, relative to the timescale of variations under investigations.

Bartlein 2006

Page 23: Class 9, Statistics in paleoclimatology

Steps Bartlein (2006)

Page 24: Class 9, Statistics in paleoclimatology

Oscillations are either periodic or quasi-periodic variations about a stationary or slowly changing level.

Bartlein 2006

Page 25: Class 9, Statistics in paleoclimatology

Oscillations Bartlein (2006)

Page 26: Class 9, Statistics in paleoclimatology

Fluctuations are aperiodic variations of climate that appear at all timescales (but tend to be more evident at shorter timescales).

Bartlein 2006

Page 27: Class 9, Statistics in paleoclimatology

Fluctuations Bartlein (2006)

Page 28: Class 9, Statistics in paleoclimatology

Events are variations that return rapidly to a previous state. Because events are reversible, they are distinct from steps.

Bartlein 2006

Page 29: Class 9, Statistics in paleoclimatology

Events Bartlein (2006)

Page 30: Class 9, Statistics in paleoclimatology

A U T O C O R R E L AT I O N

Page 31: Class 9, Statistics in paleoclimatology

Correlation Pearson’s product-moment correlation

covariance

product of both standard deviations

Page 32: Class 9, Statistics in paleoclimatology

Autocorrelation describes the correlation of a time series with its own past and future values.

Page 33: Class 9, Statistics in paleoclimatology

Autocorrelation

covariance

product of the standard deviation

Page 34: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000

Year (A.D.)

-3

-2

-1

0

1

2

3

Ring

wid

th

Ringwidthlag-0 autocorrelation

Page 35: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000

Year (A.D.)

-3

-2

-1

0

1

2

3

Ring

wid

th

Ringwidthlag-1 autocorrelation

Page 36: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000

Year (A.D.)

-3

-2

-1

0

1

2

3

Ring

wid

th

Ringwidthlag-2 autocorrelation

Page 37: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000

Year (A.D.)

-3

-2

-1

0

1

2

3

Ring

wid

th

Ringwidthlag-3 autocorrelation

Page 38: Class 9, Statistics in paleoclimatology
Page 39: Class 9, Statistics in paleoclimatology

1900 1920 1940 1960 1980 2000-3

-2

-1

0

1

2

3

-10

-5

0

5

10PDO index Mexican PDSI

Page 40: Class 9, Statistics in paleoclimatology

IF a time series (of length N) is significantly autocorrelated, then:

The series is not random in time

Each observation is not independent from other observations

The number of independant observations is fewer than N

Page 41: Class 9, Statistics in paleoclimatology
Page 42: Class 9, Statistics in paleoclimatology

The “effective sample size” is an estimate of the “real” number of observations a!er adjusting for the effects of autocorrelation.

Page 43: Class 9, Statistics in paleoclimatology

Effective sample size

samplesize

effectivesample

sizefirst-order

autocorrelation

Page 44: Class 9, Statistics in paleoclimatology

T H E S P E C T R U M O F C L I M AT E

Page 45: Class 9, Statistics in paleoclimatology

The spectrum of a time series is the distribution of variance of the series as a function of frequency.

Page 46: Class 9, Statistics in paleoclimatology

redorangegreenbluevioletyellow

Page 47: Class 9, Statistics in paleoclimatology

redorangegreenbluevioletyellow

shortwavelengths

longwavelengths

Page 48: Class 9, Statistics in paleoclimatology

redorangegreenbluevioletyellow

Fastchanges

slowchanges

Page 49: Class 9, Statistics in paleoclimatology

200 250 300 350 400 450 500

-3

-2

-1

0

1

2

3

4

example of a ‘white’ time series

Page 50: Class 9, Statistics in paleoclimatology

200 250 300 350 400 450 500

-3

-2

-1

0

1

2

3

4

example of a ‘red’ time series

Page 51: Class 9, Statistics in paleoclimatology

200 250 300 350 400 450 500

-3

-2

-1

0

1

2

3

4

example of a ‘blue’ time series

Page 52: Class 9, Statistics in paleoclimatology

“White”

Page 53: Class 9, Statistics in paleoclimatology

“Red”

Page 54: Class 9, Statistics in paleoclimatology

“Blue”

Page 55: Class 9, Statistics in paleoclimatology

Schematic variance spectrum of climate variations Bartlein 2006

Page 56: Class 9, Statistics in paleoclimatology

Schematic variance spectrum of climate variations Bartlein 2006

BIG changes

LITTLE changes

Page 57: Class 9, Statistics in paleoclimatology

Schematic variance spectrum of climate variations Bartlein 2006

FASTSLOW

Page 58: Class 9, Statistics in paleoclimatology

S P E C T R A L A N A LY S I S

Page 59: Class 9, Statistics in paleoclimatology

Source: Burroughs, Weather Cycles: Real or Imaginary?

Page 60: Class 9, Statistics in paleoclimatology

Source: Burroughs, Weather Cycles: Real or Imaginary?

Page 61: Class 9, Statistics in paleoclimatology
Page 62: Class 9, Statistics in paleoclimatology

Source: Burroughs, Weather Cycles: Real or Imaginary?

Page 63: Class 9, Statistics in paleoclimatology

Source: Burroughs, Weather Cycles: Real or Imaginary?

frequency (cpy)

Page 64: Class 9, Statistics in paleoclimatology
Page 65: Class 9, Statistics in paleoclimatology

Source: Burroughs, Weather Cycles: Real or Imaginary?

Page 66: Class 9, Statistics in paleoclimatology

Source: Burroughs, Weather Cycles: Real or Imaginary?

Page 67: Class 9, Statistics in paleoclimatology

Source: Burroughs, Weather Cycles: Real or Imaginary?

Page 68: Class 9, Statistics in paleoclimatology

Source: Burroughs, Weather Cycles: Real or Imaginary?

Page 69: Class 9, Statistics in paleoclimatology

Source: Burroughs, Weather Cycles: Real or Imaginary?

Page 70: Class 9, Statistics in paleoclimatology

evolutive spectra

Page 71: Class 9, Statistics in paleoclimatology

200 250 300 350 400 450 500

-3

-2

-1

0

1

2

3

4

200 250 300 350 400 450 500

-3

-2

-1

0

1

2

3

4 Imagine a ‘window’ passing through your data

Page 72: Class 9, Statistics in paleoclimatology

Evolutionary spectra for the global 018 record Bartlein 2006

Page 73: Class 9, Statistics in paleoclimatology

Evolutionary spectra for the global 018 record Bartlein 2006

Page 74: Class 9, Statistics in paleoclimatology

Evolutionary spectra for the global 018 record Bartlein 2006

Page 75: Class 9, Statistics in paleoclimatology

Evolutionary spectra for the global 018 record Bartlein 2006

Page 76: Class 9, Statistics in paleoclimatology

Evolutionary spectra for the global 018 record Bartlein 2006

cone ofinfluence

cone ofinfluence

Page 77: Class 9, Statistics in paleoclimatology

what?compared to

Page 78: Class 9, Statistics in paleoclimatology

“Red”

Page 79: Class 9, Statistics in paleoclimatology

Source: Shanahan et al., 2009, Science

Page 80: Class 9, Statistics in paleoclimatology

200 250 300 350 400 450 500

-3

-2

-1

0

1

2

3

4

example of a ‘white’ time series

X

Page 81: Class 9, Statistics in paleoclimatology

200 250 300 350 400 450 500

-3

-2

-1

0

1

2

3

4

example of a ‘red’ time series

Page 82: Class 9, Statistics in paleoclimatology

Source: Shanahan et al., 2009, Science

Page 83: Class 9, Statistics in paleoclimatology

statisticalsignificance

physicalsignificance

Page 84: Class 9, Statistics in paleoclimatology

The history of meteorology is li"ered with whitened bones of claims to have demonstrated the existence of reliable cycles in the weather.

“”

William James Burroughs

Page 85: Class 9, Statistics in paleoclimatology