Post on 18-Jan-2016
transcript
Broken Hockey Sticks, 1500 Year Cycles, and Ocean Cooling
Dr. Craig LoehleNCASI Principal Scientist
552 S Washington St., Ste. 224Naperville, Illinois 60540
The Broken Hockey Stick
• Fundamental assumptions of tree ring reconstruction are untested
• In fact, they are violated• Ms published online in Climatic Change,
Sept. 08• “A Mathematical Analysis of the Divergence
Problem in Dendroclimatology”
Climate Reconstruction(based largely on tree ring data)
How is it Done?
• Assume linear model ring vs. temperature• Fit to temperature histories, 20th Century• Inverse function for past temperature from
ring width / density
Signs of Trouble
• Widespread observations of divergence– good fit up to 1960– more warming than predicted by tree rings after
1960– Explanations are speculative
• Individual trees– may not respond to temperature– may grow worse when warmer
• (PPT limits)
IF Linear Response, Perfect Fit
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
0 200 400 600 800 1000
Years
Tem
pera
ture
0.00
0.10
0.20
0.30
0.40
0 200 400 600 800 1000
Years
Rin
g W
idth
IF Nonlinear Response, Get Truncation (inversion)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
-2 0 2 4
Temperature
Rin
g W
idth
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
0 200 400 600 800 1000
Years
Tem
pera
ture
Peaks become troughs!
IF Ramp Response (no PPT Limitation),Get Truncation
0.20
0.25
0.30
0.0 0.5 1.0 1.5 2.0
Temperature
Rin
g W
idth
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
0 200 400 600 800 1000
Years
Tem
pera
ture
Ramp results from species maximum growth rate
Conclusions
• Linear growth response invalid• Nonlinear growth explains divergence• Consequence
– can’t guarantee detection of temperatures warmer than calibration period
• Flattens out timeseries; reduces mean, range, and maximum• Can’t evaluate MWP or how unusual recent decades are from
linear reconstructions• Can’t use nonlinear model because gives nonunique inverse
model
The 1500-Year Climate Cycle
• Dansgaard-Oschger events in ice core data• About 1500 years apart• Cause: solar or ocean oscillator (multi-state
system)• Did they continue?
Greenland Data AnalysisUpdated GICC05 ice core chronology
Years BP (Yr 0 = 2000 AD)
δ18O
Greenland Data AnalysisBest-fit model for 30,000 to 37,000 BP
(R2 = 0.36, period = 1486 yrs)
Years BP (Yr 0 = 2000 AD)
δ18O
(val
ues
cent
ered
for b
etter
esti
mati
on o
f cyc
les)
Greenland Data AnalysisBest-fit model for 40,000 to 46,000 BP
(R2 = 0.30, period = 1598 yrs)
Years BP (Yr 0 = 2000 AD)
δ18O
(val
ues
cent
ered
for b
etter
esti
mati
on o
f cyc
les)
Greenland Data AnalysisBest-fit model for the Holocene to 8000 BP
(R2 = 0.45, period = 1434 yrs)
Years BP (Yr 0 = 2000 AD)
δ18O
(val
ues
cent
ered
for b
etter
esti
mati
on o
f cyc
les)
Best-fit Model 1681- and 1470-Year CyclesCompared to Loehle Reconstruction (R2 = 0.68, 0.65 respectively)
(confidence intervals not shown)
Years BP (Yr 0 = 2000 AD)
Anom
aly
(°C)
1470-yr cycle
1681-yrcycle
Best-fit Model to Moberg Low-Frequency Data (1152-yr cycle, R2 = 0.69)
and Fit to 1470-yr cycle (R2 = 0.54)(confidence intervals not shown)
Years BP (Yr 0 = 2000 AD)
Anom
aly
(°C)
1470-yr cycle1152-yr cycle
Best-fit Model to North Iceland Sea Surface Temperature(1552-yr cycle, R2 = 0.20)
Data shown after linear cooling trend removal
Years BP (Yr 0 = 2000 AD)
Anom
aly
(°C)
Oxygen Isotope Data from Central Alps, past 9000 yrsBest-fit models for 2004 AD to 3500 BP and 6500 BP to 8996 BP
(1479-yr cycle, R2 = 0.27, peak-to-trough amplitude 0.33°C)
Years BP (Yr 0 = 2000 AD)
-del
ta O
-18
(%)
Temperature inferred from Japanese CedarBest-free-form-fit (1089-yr cycle, R2 = 0.28); Best-fit when
forced with 1470-yr cycle looks similar (R2 = 0.22)
Years BP (Yr 0 = 2000 AD)
Anom
aly
(°C)
1089-yr cycle
1470-yr cycle
Conclusions
• Mean period 1524 yr vs. 1470 yr hypothetical (3.7% off)
• Models, data strongly agree on timing of MWP, LIA
• Supports role for this cycle in recent warming
Ocean Temperatures:Recent Trends
• ARGOS float data• To 700 m depth• 4.5 years data• Global coverage
Ocean heat content
Calendar Year
Oce
an H
eat C
onte
nt (x
1022
J)
Ocean heat content smoothed with a 1-2-1 filterand overlaid with a best-fit linear plus sinusoidal
(seasonal) model (R2 = 0.85)
Calendar Year
Oce
an H
eat C
onte
nt (x
1022
J)
Ocean heat content smoothed with a 1-2-1 filterand overlaid with linear trend portion of best-fit
model (slope = -0.35 x 1022 J/yr)
Calendar Year
Oce
an H
eat C
onte
nt (x
1022
J)
Recent Temperature Downturn Evident in Hadley Data
Conclusions
• Cooling trend over past 4.5 years matches satellite, surface data
• Rate of cooling similar to rate of warming before 2003