coming attractions. a gedalof, mantua, peterson production cig / jisao presents

COMING ATTRACTIONS

COMING ATTRACTIONS

A GEDALOF, MANTUA, PETERSON PRODUCTION

CIG / JISAO PRESENTSCIG / JISAO PRESENTS

A multi-century perspective of variability in A multi-century perspective of variability in the Pacific Decadal Oscillation: new insights the Pacific Decadal Oscillation: new insights

from tree rings and coralfrom tree rings and coral

Reconstructed PDO IndexReconstructed PDO IndexReconstructed PDO IndexReconstructed PDO Index

1825 1850 1875 1900 1925 1950 1975 2000Year

-2.0

-1.0

0.0

1.0

2.0

PD

OI

R = 0.64R = 0.64R = 0.64R = 0.64

• Based on leading principal component of five published paleoproxy reconstructions.

• Collective skill better than individual skill

Mean Intercorrelation...Mean Intercorrelation...

0.00

0.25

0.50

0.75

1.00

Me

an

Inte

rorr

ela

tio

n

G edalofB iondiEvansUrban

Linsley

1600 1650 1700 1750 1800 1850 1900 1950 2000Year

Interval of Reconstruction

Note Interval of Poor Intercorrelation

1825 1850 1875 1900 1925 1950 1975 2000

Year

-0.50

-0.25

0.00

0.25

0.50

-0.50

-0.25

0.00

0.25

0.50

-0.50

-0.25

0.00

0.25

0.50

period ~ 80 yrs. 21.4 % var. exp .

period ~ 20 yrs. 20.4 % var. exp .

period ~ 23 yrs. 10.8 % var. exp.

Period of Poor Intercorrelation...

Evans

Linsley

Gedalof

APPEARING SOON IN APPEARING SOON IN

GEOPHYSICAL RESEARCH GEOPHYSICAL RESEARCH

LETTERSLETTERS

APPEARING SOON IN APPEARING SOON IN

GEOPHYSICAL RESEARCH GEOPHYSICAL RESEARCH

LETTERSLETTERS

COLUMBIA RIVER FLOW SINCE COLUMBIA RIVER FLOW SINCE A.D. 1750 RECONSTRUCTED A.D. 1750 RECONSTRUCTED

FROM TREE RINGSFROM TREE RINGS

A Gedalof / Peterson / Mantua JointA Gedalof / Peterson / Mantua Joint

Based on 32 tree-ring sitesBased on 32 tree-ring sites

1750 1775 1800 1825 1850 1875 1900 1925 1950 1975 2000Year

5.0

5.1

5.2

5.3

5.4

5.5

Lo

g1

0 m

ea

n f

low

, Th

e D

alle

s, O

R (

cfs

)

O bserved F low

R econstructed F low R = 0.59

• Residuals exhibit positive trend over time (ca. +1.2 percent per century)

• Validates model results of Matheussen et al. (2000).

1930 1940 1950 1960 1970 1980 1990Year

-0 .2

-0.1

0

0.1

0.2R

eg

res

sio

n R

es

idu

al

1750 1800 1850 1900 1950 2000Single Year low-flow events

1

6

11

16

21

26

31

36

Flo

w R

an

k

1750 1800 1850 1900 1950 20005-year m oving average

1

6

11

16

21

26

31

36

Flo

w R

an

k

1750 1800 1850 1900 1950 200011-year m oving average

1

6

11

16

21

26

31

36

Flo

w R

an

k

1750 1800 1850 1900 1950 200025-year m oving average

1

6

11

16

21

26

31

36

Flo

w R

an

k

Persistent Droughts:

• The 1930s were not an anomaly...

MANUSCRIPT IN INTERNAL MANUSCRIPT IN INTERNAL

REVIEW...REVIEW...

FEATURE FEATURE PRESENTATIONPRESENTATION

FEATURE FEATURE PRESENTATIONPRESENTATION

FIRE & CLIMATE IN THE FIRE & CLIMATE IN THE AMERICAN NORTHWESTAMERICAN NORTHWEST

Douglas-Fir

CO-CONSPIRATORSCO-CONSPIRATORS

Ze’evZe’ev

Lolita (and Dave)Lolita (and Dave)

NateNate

Q. What causes wildfire?Q. What causes wildfire?A. Fuels Accumulation

"As with other areas of the country, we have experienced the unintended consequences of our very effective wildfire fighting program: The wildfires of today are getting bigger, more dangerous, harder to control, and are adversely affecting the safety of the public and our fire fighters.”

National Fire Plan Strategy For the Pacific Northwest (2002)

Q. What causes wildfire?Q. What causes wildfire?

B. Weather

"…forest fire behavior is determined primarily by weather variation among years rather than fuel variation associated with stand age."

Bessie and Johnson (1995)

Q. What causes wildfire?Q. What causes wildfire?C. You

Evidence for fuels...Evidence for fuels...

Area burned by wildfire in 11 Western States

Source: National Interagency Fire Center

On national forest lands in the Pacific Northwest wildfires are more frequent and more extensive during the warm phase of the PDO.

1900 1920 1940 1960 1980 2000Y e a r

0

50

100

150

200

250

Bu

rne

d A

rea

Ind

ex

Cool PDO

Warm PDO

……for climate...for climate...

1900 1920 1940 1960 1980 2000Y e a r

0

50

100

150

200

250

Bu

rne

d A

rea

Ind

ex

……and you.and you.

Study OverviewStudy Overview

• To characterize patterns in annual area burned

• To relate those patterns to climatic features and ecological context

• To determine the extent to which climatic factors can be used to predict seasonal wildfire

Literature Literature ReviewReview

• Lots of work in the Canadian boreal forest.

• Very little work in the Pacific Northwest

Previous StudiesPrevious Studies

• Have generally treated area west of the Rocky Mountains as a single coherent unit– No allowance for spatial variability– No recognition of underlying ecology

• Emphasis has been on weather (not climate)

New Ideas:New Ideas:

(1) I do not treat the area west of the Rocky Mountains as a single coherent unit

(2) I address large fire seasons, rather than individual large fires

(3) I identify several key atmospheric structures that can potentially be used to forecast fire-season severity

?

EOF AnalysisEOF Analysis

• Empirical Orthogonal Function (EOF) analysis identifies underlying patterns in large data sets

– The EOFs describe the spatial variability in the data set

– Associated principal components (PCs) describe the temporal variability

Spatial RegressionsSpatial Regressions

• Can “regress” fields of climate data onto time series

• Produces characteristic response of climate field to 1 perturbation in time series

Superposed Epoch AnalysisSuperposed Epoch Analysis

• Develop map composites for selected years (i.e. epochs) based on quantitative criteria

• Derive descriptive statistics for subsets

– Can focus on extreme events– More powerful than correlations /

regressions– Does not assume linear relationship

EOF 1 - 17%

M ay

Septem ber

August

June

PC1 / 500 hPaPC1 / 500 hPaRegressionRegression

Shaded areas indicate

significant correlation

Pattern exhibits strong

blocking

• Five largest fire years minus five smallest fire years

• Patterns consistent, but magnitude greater

PC1 / 500 hPaPC1 / 500 hPaCompositeComposite

Sm all Years:1971 1974 1975 1957 1978Large Years:1987 1988 1994 1992 1959

M ay

Septem ber

August

July

June

Area burned is correlated to drought in winter and spring preceding the fire season

Correlations:

-0.59 -0.55 -0.61

June, July, Aug.

PC1 / PDSI PC1 / PDSI CorrelationsCorrelations

EOF 2 - 13%

PC2 / 500 hPaPC2 / 500 hPaRegressionRegression

Resembles “Summer

PNA”

Matches results across

border

August

July

June

Area burned is weakly correlated to drought in winter preceding the fire season

Correlations:

-0.06 -0.09 -0.11

June, July, Aug.

PC2 / PDSI PC2 / PDSI CorrelationsCorrelations

EOF 3 - 12%

• 3 only large fire years represented by PC-3

• Characterized by very strong, highly persistent blocking

PC3 / 500 hPaPC3 / 500 hPaCompositeCompositeM ay

Septem ber

August

July

June

CorrelationsCorrelations

CompositeComposite

EOF 4 - 10%

• Large fire years correspond to fire season cyclone activity

PC4 / 500 hPaPC4 / 500 hPaCompositeComposite

Sm all Years:1951 1990 1994 1970 1987Large Years:1992 1988 1977 1981 1955

M ay

Septem ber

July

June

Fire season is wet on the west side, dry on the east side

Preceding season is drier than normal

SummarySummary

1. Climate Matters

• Region wide increases in area burned are characterized by antecedent drought accompanied by persistent blocking events

SummarySummary

2. Ecology Matters

• Underlying ecology appears to modulate the response to drought and circulation– more mesic forests require persistent

drought, blocking events, and a source of ignition and spread

– drier forests are more responsive to shorter-scale (i.e. synoptic) processes

SummarySummary

3. These relationship are non-linear

• Implies that eigenvector techniques may not be the most appropriate method of investigation

• Small changes in mean climate may lead to dramatic changes in wildfire activity

Discussion ?Discussion ?