Effects of Sea Level Decadal Variability on Acceleration and Trend Difference*
James Houston1 and Robert Dean2
¹ Army Engineer Research and Development Center, Corps of Engineers
2 Department of Civil and Coastal Engineering, University of Florida
* Based on Houston and Dean, 2012, Journal of Coastal Research (Accepted)
Definition of Acceleration
Time (Years)
Sea L
evel
Car going a steady 120 miles/hr
is moving fast, but not accelerating
Velocity = Trend = Rate of change of sea level
(mm/yr)
Acceleration = Rate of change of trend (mm/yr2)
Sea Level Rise not Constant in Time - Has Decadal
Variability Likely Caused by Weather Variability
Holgate (2007)
1.7 mm/yr
About 20-yr
oscillation
Sea Level Rise not Constant in Time - Has Decadal
Variability Likely Caused by Weather Variability
Holgate (2007)
Calculated Acceleration
Depends on Record Length
40 years
X
X
Sea Level Rise not Constant in Time - Has Decadal
Variability Likely Caused by Weather Variability
Holgate (2007)
Calculated Acceleration
Depends on Record Length
40 years
X
X
Sea Level Rise is Not Spatially Uniform Like a
Bathtub Filling with Water
Sea Level Change Is Not Spatially Uniform
Satellite altimeter measurements over 19 + years
3-4 times world
average
Sea level falling
Douglas (1992)
All tide gauge records in the
world with length ≥ 10 years
Sea level rise in 100
yrs if acceleration
persists
+ 5 m
- 5 m
+ 1 m
- 1 m
+10m
- 10m
+5m
- 5m
All tide gages in world (1123)
with lengths ≥ 10 yr
Houston and Dean (2012)
75 yrs
Douglas (1992) used only
records ≥ 75 years
Sea level rise in 100
yrs if acceleration
persists
Violating the Prime Directive
• PRIME DIRECTIVE - Use record lengths > ~ 75 yrs to
determine sea level acceleration and don’t trust results
where this criterion is violated
• Sallenger et al (2012) consider
40 - 60 yr records, leading
to invalid conclusions
• 900 world media outlets picked
up the Sallenger story of a “hot
spot” of sea level rise along the
east coast
• Rahmstorf and Vermeer (2011) use record lengths as short
as 30 yrs to erroneously claim their model and projections
superior to those of the Intergovernmental Panel on Climate
Change
Sallenger et al Use Trend Difference = T2-T1
20-Year Record
Trend Difference ~ Acceleration * Record Length
20-Year Record
Trend Difference Versus Record Length
Significant scatter even for 100-yr records
All tide gages in world (1123)
with lengths ≥ 10 yr
Sallenger et al (2012) Trend Difference for 40-Year Period Centered in 1990
New York “Hotspot”
Sallenger et al (2012)
Sallenger et al (2012) obtained
trend difference of 3 mm/yr for
40-yr period centered in 1990
They
projected
This trend
difference
to 2100
They
projected
This trend
difference
to 2100
Sallenger et al (2012) Trend Difference for 40-Year Period Centered in 1990
New York
Fernandinaa
Neutral
Sallenger et al (2012)
Galveston
“Cold”
Sallenger et al (2012)
San Francisco
“Cold”
Sallenger et al (2012)
Bromirski et al
(2011)
Sallenger et al (2012)
• Violate “prime directive” that requires record lengths > ~ 75
yrs to determine the underlying acceleration
• By taking short records, they just see
decadal variations, not underlying
trends
• It is not valid to project trend difference
based on one 40-yr period because
trend difference varies every year
• They say, “Trend differences for
windows longer than 72 yr were not
significantly different than zero.”
• OF COURSE! When they analyze records of proper length,
the east coast is not even a “lukewarm” spot of sea level
acceleration
Based on Church and
White (2006) worldwide data
Rahmstorf and Vermeer (2011)
Based on Rahmstorf
and Vermeer model
They calculate acceleration from beginning year to 2001 and
plot value at beginning year
~ 60 to 30-yr record lengths
>75 yr record lengths
Dots are worldwide data of
Church and White (2006)
1. Two Linear Trends
2. 20-yr decadal oscillation with 5-mm amplitude
3. 50-yr decadal oscillations with 4-mm amplitude
Trend 0.8 mm/yr
Trend 1.9 mm/yr
Synthetic Record
Trend change
~ 1926
Synthetic Record
Data of
Church
and White
Synthetic Record Versus Data
Trend Change
Decadal Variations
Data of
Church
and White
Acceleration due to
trend change
Decadal variations set to zero Dip from ~ 1895-1930 due primarily
to trend change in 1926
Rahmstorf and Vermeer claim the dip from ~ 1895 to 1930 is
due to temperature changing little from “1940 to 1980”
Acceleration Due to Trend Change
Zero after ~ 1926
Acceleration from
1870 – 2001 matches data
Data of Church
and White
Acceleration from 1940 due to
decadal variations
Trend change set to zero
Negative contribution
1917 - 1933
Acceleration Due to Decadal Variations
Acceleration from decadal variations
Decadal oscillations
and linear trends
Data of
Church
and White
Synthetic Record Versus Data
Linear trends and decadal oscillations explain dip
from ~ 1895-1930 and acceleration after 1940
Rahmstorf and Vermeer, 2011
• Violate the “prime directive” by considering records as
short as 30 – 60 yrs, which are dominated by decadal
variations
• The trend change + decadal oscillations from 1917 to 1933
produce the dip in acceleration from
~ 1895 to 1930
• Decadal variations cause the
acceleration beginning in ~ 1940
• Their model does not include the trend
change or decadal variations that
determine the curve they claim shows
their model to be superior to that of the
Intergovernmental Panel on Climate Change
Conclusions
• Tide records > ~ 75 yrs must be used to determine
acceleration because decadal variations
dominate shorter records and obscure
the underlying acceleration
• It is not valid to project sea level rise
based on short records
• Journal papers are not always correct
- Just require 2 of 3 reviewers’ approval
- The Intergovernmental Panel on
Climate Change (IPCC) has > 40
eminent authors who agree on sea
level projections and several hundred
reviewers conduct two detailed reviews
- Use sea level projections of IPCC (Sept, 2013)