North Atlantic Decadal Variability of Ocean Surface Fluxes

1
North Atlantic Decadal Variability of Ocean
Surface Fluxes

• Mark A. Bourassa1,2, Paul J. Hughes1,2, Jeremy
  Rolph1, and Shawn R. Smith1
• 1. Center for Ocean-Atmospheric Prediction
  Studies
• 2. Department of Meteorology
• The Florida State University

2
Objectives

• Examine Atlantic Ocean surface turbulent energy
  fluxes for multi-decadal variability.
• Latent heat flux
• Sensible heat flux

• What related variables are changing with the
  fluxes
• Sea surface temperature (SST)
• Near surface (10m) wind?
• Near surface (10m) air temperature
• Near surface (10m) humidity

• SST minus 10m air temperature?
• Surface humidity minus 10m humidity?
• Can these changes be linked to Atlantic Tropical
  Cyclone variability?

3
What are surface turbulent fluxes?

• Latent Heat Flux (E)
• Vertical transport of energy associated with the
  phase change of water
• Forced by wind speed and air/sea temperature
  differences
• Sensible Heat Flux (H)
• Vertical transport of energy associated with
  heating, but without a phase change
• Forced by wind speed and vertical moisture
  differences
• Stress (?)
• Vertical transport of horizontal momentum
• Forced by vertical momentum differences

Atmosphere
E
E-
H
H-
?-
?
Ocean
4
Relevance of Surface Turbulent Fluxes
Latent Heat flux Latent heat released via
condensation
Sensible Heat Flux Cold air outbreak over the
warmer Gulf Stream
Stress Forcing the upper ocean circulation,
upwelling, and downwelling
Top http//home.cc.umanitoba.ca/dunlopaj Middle
http//science.hq.nasa.gov Bottom
http//www-evasion.imag.fr
5
Relevance of Surface Turbulent Fluxes

• Sensitive indicators of changes in the climate
  system, integrating changes in the following
  variables.
• Wind Speed
• Air/Sea temperature difference
• Vertical moisture differences

• Modulations in the above induce changes in the
  latent and sensible heat fluxes OR modulations
  in heat fluxes change the above variables.

6
Past Studies of North Atlantic Flux Variability

• Zhao and McBean (1986)
• Cayan (1992)
• Alexander and Scott (1997)

• Examined the longer time scale basin wide
  variability of the turbulent heat fluxes over the
  North Pacific and Atlantic Oceans
• Concluded that the latent and sensible heat
  flux respond to changes in the low level
  atmospheric circulation patterns, e.g., the North
  Atlantic Oscillation (NAO)
• Showed that anomalous fluxes are organized over
  regions of atypical zonal and meridional flow

7
Forcing Product Inconstancies Zonal Averaged
Latent Heat Flux

• NWP Products
• NCEPr2
• JRA
• Satellite Product
• HOAPS
• NWP/Satellite Hybrid
• WHOI
• In Situ
• NOC (AKA SOC)
• FSU3

Latitude
40 60 80 100 120
140 160 180
Wm-2
8
Forcing Product Inconstancies Zonal Averaged
Sensible Heat Flux

• NWP Products
• NCEPr2
• JRA
• Satellite Product
• HOAPS
• NWP/Satellite Hybrid
• WHOI
• In Situ
• NOC (AKA SOC)
• FSU3

Latitude
0 10 20
30 40
Wm-2
9
Latent Heat Flux January 1989
Sensible Heat Flux January 1989
15 45 75 105 135 175 Wm-2
10 30 50 70 90 110 Wm-2
Latent heat flux
Sensible heat flux
10
Input Data for Our Flux Product(FSU3 Winds and
Fluxes)

• International Comprehensive Ocean-Atmosphere Data
  Set (ICOADS Woodruff et al. 1987 Worley et al.
  2005)
• Reynolds SSTs (Reynolds 1988)
• Bias corrections for ship based SSTs is difficult
  because it varies greatly on ship to ship basis

11
Average Number of Ship Observations
January
Average Number of Ship Observations
0 1 4 9 16 25
36 49 64 gt81
August
0 1 4 9 16
25 36 49 64 gt81
12
Creating the FSU3 Fluxes

• We currently have research quality fields for
  1978 through 2004
• Atlantic Basin Indian Ocean

• The bias corrected observations of winds,
  temperatures, and humidities are averaged
  monthly, and in 1x1? bins.
• Bias adjustments
• Air temperature for heat of the superstructure
  (Berry et al. 2004)
• SST adjusted from bulk to skin temperature
  (Donlon and Robinson 1997)
• Winds

• Beaufort speeds converted to 10 m values (Lindau
  1995)
• Buoys height adjusted to 10m winds (Bourassa et
  al. 1999)
• Ship anemometers treated as 20m winds, and height
  adjusted.
• These data are input into a variational technique
  (Bourassa et al. 2005)

13
Is Our Gridding Technique Effective?Validation
of Wind Fields

• Monthly mean Winds from August 1999 through Dec.
  2004 are compared to similarly averaged fields
  from SeaWinds on QSCAT.

• Biases (not shown) are very small.
• Random errors small over most of be basin.
• Larger in areas of relatively poor sampling
• Larger in areas with more natural variability

0.3 0.6 0.9 1.2 1.5 1.8
2.1 2.4 2.7 ms-1
14
Atlantic Mulitdecadal Oscillation (AMO)

• Thought to be forced by fluctuations in the
  thermohaline circulation (Schlesinger and
  Ramankutty 1994 Kerr 2000 Delworth and Mann
  2000)
• Period of 65-70 years
• Linked to anomalous precipitation patterns and
  North Atlantic hurricane activity (Enfield et al.
  2001 Sutton and Hodson 2005 Goldenberg et al.
  2001)

Enfield et al. 2001
15
Tropical North Atlantic

• Recall that we have a research quality time
  series for the period 1978 through 2004.
• This is slightly longer than the satellite period
  for which has arguably been called good for NCEP
  reanalyses.
• A longer time series would be better!

16
Stretching Our Time Series

• The density of in situ (Volunteer Observing Ship)
  data from the Atlantic Ocean peaks in the 1980s.
• A data set based on in situ data could be
  extended much further back in time.
• We used ICOADS data from Jan. 1956 through Dec.
  1977 to extend our data set.
• All our automated procedures were used however,
  we did not apply the visual quality control step.
• Skipping this step is analogous to adding noise.
• We reduce this noise by applying spatial averages
  in Hovmueller diagrams.

17
Regions Examined

• The regions that I will discuss are the
• Gulf of Mexico (97 to 83?W, 21 to 29?N),
• Caribbean Sea (84 to 62?W 10 to 19?N)
• Northern Atlantic tropics (60 to 18?W 0 to 20?N)
• Southern Atlantic tropics (35?W to 8?E 0 to 20?S)

18
Latent Heat Flux Gulf of Mexico

• There is a large increase in LHF in the
  southeastern Gulf starting in 1998.
• This increase is preceded by a period of weak LHF.

19
Gulf of Mexico Wind Speed

• The winds decrease at about the same time the LHF
  increases.
• The percentage change in wind speed is much less
  than the change in LHF

20
Gulf Air/Sea Humidity Differences qsfc - qair

• A very good match to the change in LHF!

21
Latent Heat Flux Caribbean Sea

• The peak is in the early 1960s.
• There appears to be a 10 to 13 year cycle
  superimposed on a multi-decadal cycle.
• There is a period of increased LHF after 1995,
  the same period as increased Atlantic hurricane
  activity.

22
Caribbean Wind Speeds

• Another poor match to variability in LHF.
• Slight contribution to increase in LHF in post
  1995 period.

23
Caribbean Air/Sea Humidity Differences qsfc -
qair

• A good match to the LHF variability

24
Latent Heat Flux Northern Tropical Atlantic

• Again there is a 11 to 13 year cycle superimposed
  on a longer term trend or cycle.
• Hints of ENSO-related variability
• The large LHFs extend further East during the
  periods of enhanced hurricane activity.

25
Tropical North Atlantic Air/Sea Humidity
Differences

• Considerable matching variability on the 11 to 13
  year scale.
• Insufficient longer term trend.

26
Tropical North Atlantic Wind Speed

• Longer term trend of increasing wind speed.
• Or part of a longer cycle?

27
Latent Heat Flux Southern Tropical Atlantic
28
Southern Tropical Atlantic Air/Sea Humidity
Differences

• Another stunningly good match!

29
Southern Tropical Atlantic Winds

• Winds also contribute to the changes in LHF

30
Summary

• Changes in tropical Atlantic latent heat fluxes
  are closely linked to changes in air/sea moisture
  difference.

• Changes in LHF due to changes in wind speed are
  mixed
• Gulf of Mexico and Caribbean Sea changes in wind
  speed often counter the influence of changes in
  air/sea moisture difference
• Atlantic Ocean changes in wind speed often
  increase the changes associated with air/sea
  moisture differences
• There is a substantial trend in Atlantic Ocean
  LHF and wind speed, or perhaps a longer period
  (40 years) cycle.
• An 11 to 13 year cycle is found in the latent
  heat fluxes and in the air/sea moisture
  differences.
• This 11 to 13 year variability is not nearly as
  apparent in time series of qair or qsfc.

31
North Atlantic Decadal Variability of Ocean
Surface Fluxes

• Mark A. Bourassa1,2, Paul J. Hughes1,2, Jeremy
  Rolph1, and Shawn R. Smith1
• 1. Center for Ocean-Atmospheric Prediction
  Studies
• 2. Department of Meteorology
• The Florida State University

32
Air/Sea Temperature Difference SST - Tair
33
SHF???
34
Many Air/Sea Interaction Processes- Most are
modified by surface waves -
Graphic adapted from CBLAST
Wind waves
Swell waves
35
Results of Taylor and Yellands Parameterization
on SWS2 data

• Taylor and Yelland (2003) is considered an
  excellent parameterization for stress related to
  wind waves.
• It has two tuning parameters, the same number as
  used in Bourassa (2006).

36
Results of Bourassa (2006) Compared to SWS2
Observations

• This variation has a non-zero Newtonian frame of
  references and displacement height.
• Displacement height is a fraction of the
  significant wave height.

• Charnocks constant is actually constant.

37
iv. Quality control

• Comparison to climatology
• Applied to individual observations
• Excessive trimming not a problem
• Auto-flag procedure
• Applied to monthly mean gridded ship observations
  
• Flags and removes grid points that differ too
  much from adjacent points
• FSU3 fluxes are the first version of FSU winds to
  employ technique
• Flux editor
• Analyst visually inspects the in situ fields and
  subjectively removes suspect data not eliminated
  by the preceding quality control procedures
• Very few data removed

38
i. Cost function

• A cost function based on weighted constraints is
  minimized via a conjugate-gradient minimization
  scheme
• Three constraints for vector variables
• Misfit to observations
• Laplacian smoothing term
• Misfit of the curl
• Constraints help maximize the similarity of the
  solution fields to the observations and minimize
  unrealistic spatial feature
• Each constraint multiplied by a weight that is
  determined using cross validation (Wahba and
  Wendelberger 1980 Pegion et al. 2000)

39
i. Cost function

• A cost function based on weighted constraints is
  minimized via a conjugate-gradient minimization
  scheme
• Three constraints for vector variables (Two for
  Scalar Variables)
• Misfit to observations
• Laplacian smoothing term
• Misfit of the curl
• Constraints help maximize the similarity of the
  solution fields to the observations and minimize
  unrealistic spatial feature
• Each constraint multiplied by a weight that is
  determined using cross validation (Wahba and
  Wendelberger 1980 Pegion et al. 2000)

40
ii. Background fields

• Observations from only the month being examined
  are used to create background fields
• More effective than using a long-term climatology
  (Bourassa et al. 2005)
• Gaussian weighted spatial average applied to the
  in situ monthly fields to determine values at
  each grid point
• The weight is a function of distance away from
  the grid point of interest
• A line of sight constraint implemented
  (Bourassa et al. 2005)