Ocean exchanges with the atmosphere

1
Ocean exchanges with the atmosphere
.did we learn anything during WOCE?

• Peter K. Taylor
• Southampton Oceanography Centre UK

2
Overview

• What surface fluxes were needed for WOCE

• How the flux estimates are obtained

• How far have we progressed during the WOCE period

Separately for

• Future Flux Observing System

3
The Goals of WOCE
4
The important Air-Sea fluxes for WOCE
Net heat flux is sum of
5
The important Air-Sea fluxes for WOCE
but little in this talk on precipitation since
accuracy still poor
6
The important Air-Sea fluxes for WOCE
7
Goal 2 of WOCE
8
Air-Sea Flux aims for WOCE

• Produce estimates of the global air-seafluxes of
  heat, freshwater and momentumon a range of time
  and space scales

• Produce climatological fields for these fluxes

• Work toward definition of an on-going observing
  system for the surface fluxes

9
How surface fluxes are determined

• Budget methods give total heat flux

• divergence of ocean heat transport (e.g.
  Ganachaud Wunsch, 2000)

• atmospheric flux divergence with top of
  atmosphere radiative balance the residual
  method(e.g. Trenberth et al. 2001 )

10
Determining the individual flux components

• SW and LW Radiative fluxes can be obtained from
  Satellite data and from NWP models

• Turbulent fluxes from in situ data, models, and
  satellites, are based on meteorological variables
  (temperature, wind, etc.) and the bulk formulae

11
Example of Bulk FormulaLatent heat flux (W/m2)
Flux
Transfer x Wind x humidityCoefficient
speed difference
12
Determining the Transfer Coefficient
SeaSat coincided with JASIN
GARP Air-Sea Interaction experiments BOMEX,
AMTEX, IFYGL
13
The Legacy of GARP

• Budget experiments are difficult!

• Experimental data on transfer coefficients was
  available

The Legacy of SeaSat

• Satellite scatterometers could define wind
  forcing

• We must continue to maintain (and improve) the in
  situ observing systems

14
Developing the Voluntary Observing Ship (VOS)
system

• Due to research during the WOCE period (partly
  funded by TOGA and WOCE)

• The random and systematic errors in VOS data are
  much better known

we can now plot a map of error values like
this one

• Now greater emphasis on meta-data .how the
  observations are obtained

Mean random errors in ship SST obs ( C )1970 -
1997 (Kent, 2002)
15
Wind Stress

• The choice of Drag Coefficient, CD10n

• Effect of using other CD10n values

• Climatic variations in mean wind stress

• Effect of poor sampling in the SO

16
The variation of the Drag Coefficient with wind
speed
but some models are still using these higher
values
Before WOCE Smith (1980)
Smith (1988) used for TOGA and
scatterometer Data on WOCE DVD
WOCE Southern Ocean Cruises confirmed Smith
(1980)
17
Comparison of the zonal mean wind stress
( Josey et al. 2002, J.Phys.Oceanogr.
32,1993 - 2019)
18
Comparison of the zonal mean wind stress
( Josey et al. 2002, J.Phys.Oceanogr.
32,1993 - 2019)
19
Change in wind stress with NAO
SOC
HR
SOC HR wind stress fields look similar
20
Change in wind stress with NAO
SOC
HR
Scaling HR by Cd ratio gives values similar to
NCEP
but NCEP has lower stress for
period representing most of HR data
remaining differences between NCEP, HR and
SOC may be due to bad sampling

NCEP
Apparent agreement between HR SOC was due to
NAO variations
1980-93
1949-79
21
Comparison of the zonal mean wind stress
( Josey et al. 2002, J.Phys.Oceanogr.
32,1993 - 2019)
22
Zonal Wind stress in the Southern Ocean July
mean values
Where data is lacking values are extrapolated
from other regions
SOC Climatology
ECMWF ERA and scatterometer winds show
extensive belt of high winds in SO
ECMWF
ERS-1 AMI
23
Summary Wind stress

• The WOCE cruises have helped confirm the Smith
  (1980) CD10n to U10n relationship

• HR (and Oberhuber) over-estimate the wind stress
  over much of the world ocean by around 30

.but we knew that (e.g. Harrison, 1989)so why
do models still use these stress fields?

• The magnitude and patterns of wind stress varies
  significantly between different periods WOCE
  will not be typical of other decades

• WOCE helped implement satellite scatterometer
  missions which are now coming to fruition

24
Heat Fluxes

• Global Heat balance for in situ climatologies

• Adjustment using WOCE hydrography

• Comparison with other estimates Reanalyses,
  Residual Method

• Comparison of the implied latent heat flux
  distributions

25
Annual heat input to Ocean (W/m2)(SOC
Climatology, Josey et al. 1999)
This annual mean is deceptive with regard to
regions of heating and cooling
75 60 45 30 15 0 -15 -30 -45
-60 -75
30 90 150 -150
-90 -30 30
-100 -50 0 50 100 W/m2
26
Monthly heat input to Ocean (W/m2)(SOC
Climatology, Josey et al. 1999)
Heating occurs over most of summer Hemisphere
we will use January fields in following
comparisons
27
Before and after WOCE
OSU (Esbensen Kushnir 1981)
-500 -250 0
250 W/m2
28
Comparison of SOC OSU climatologies

• SOC has
• Correct flux averaging method

• Higher resolution

( Fluxes calculated from individual observations
and then averagedi.e. sampling rather than
classical )

• More information revised version of COADS with
  observations corrected on a ship by ship basis

• Larger Global Heat Budget imbalance

29
Comparison of ClimatologiesNet Heat Flux for
January and Mean Annual imbalance (W/m2)
there is obviously more summer heating in SOC
fields
-500 -250 0
250 W/m2
30
The Heat Budget problem

• Unless adjusted, climatologies show too much
  heat flux into the ocean(e.g. Bunker et al.
  1982, Isemer et al. 1989, DaSilva et
  al. 1984, Josey et al. 1999)

• This heat imbalance varies little year to year (
  few W/m2 )

• Adjusting the heat fluxes degrades the
  comparisons with buoy data(Josey et al. 1999)

31
Can WOCE help?
Grist Josey (see poster) have adjusted SOC
climatology using WOCE section data
0.002 Aagard Greisman (1975) 0.1
-0.09 ( R McC. 89)
1.22 (Hall Bryden 82)
0.76 (Bryden et al. 91)
1.22
(Klein et al. 95)
1.18
0.70 (Wijffels et al. 96)
0.60 (Speer et al. 96)
0.29 (Holfort Siedler 01)
0.46 (McDonagh 02)
0.90 (Wijffels et al. 2001)
Heat Transports in PW(adapted from Grist
Josey, 2002)
32
Effect of Constraining Heat Budget
33
Comparison of Constrained SOC UWM heat fluxes
Fields look similar but DaSilva (UWM) has e.g.
stronger cooling over Gulf Stream, greater
heating in summer hemisphere this causes small
differences in implied ocean heat transport

(adapted from Grist Josey, 2002)
34
Comparison of other Flux fields
-2
-4
some differences are obvious, for example the
el Nino region
1
6
35
Atlantic Zonal Mean Values
in Residual method, more cooling over the Gulf
Stream implies greater ocean heat transport
northward
in contrast NCEP cooling occurs in the Trade
Wind zone rather than higher latitudes
lack of net heat input in ERA implies too
large ocean heat transport in Southern Ocean
compared to SOC, there is slightly more
heating in the UWM climatology, hence less
ocean transport
(adapted from Grist Josey, 2002)
36
Atlantic OceanMean area heat fluxClimatology -
WOCE( W/m2 )
area mean air-sea flux can be calculated from
difference in ocean heat transport between
hydrographic lines
Bar plot shows difference from this mean for
the flux fields listed
compared to hydrography, rest have too little
cooling at high latitudes, too little heat input
in low latitudes
original SOC climatology has too much heat
input everywhere
(adapted fromGrist Josey, 2002)
37
Pacific Indian Oceans Mean area heat
fluxClimatology - WOCE ( W/m2 )
any such pattern in the Pacific is less clear
(Grist Josey, 2002)
38

• The Residual Method gives the best agreement with
  Hydrography

• The in situ climatologies can be adjusted to
  give agreement with Hydrography

• But have the individual heat flux components been
  properly adjusted?

39
Transfer Coefficient for water vapour
40
Errors in estimating Latent Heat flux

• Air-Sea interaction experiments suggest that
  CE10n is known to 10 or better

possibly but we need independent verification

• Inverse analyses suggest that the Flux is
  underestimated by nearly 20

• Do the errors in the observations explain this
  difference?

41
Independent sources for Evaluating Bias Errorsin
Latent Heat Flux Estimates

• Freshwater Budget - but precipitation???

• Flux fields from models

• Satellite based estimates of Latent Heat Flux

• Reference data sets - buoys and ships

42
Annual mean zonal Latent Heat Flux
Constrained UWM/COADS would be similar to
ERA model fluxes bridge original and
constrained values
The SeaFlux group have performed flux field
comparisons these are original UWM/COADS
values
( from Curry et al. 2002 and Kubota et al.
2002 )
43
Example of a Satellite Flux field Product
Climatological mean (1988 - 1996) Latent Heat
Fluxin January from the HOAPS (Grassl et al.
2000) Atlas
44
Annual mean zonal Latent Heat Flux
satellite derived flux fields also show a range
of values
( from Curry et al. 2002 and Kubota et al.
2002 )
45
Comparison of SOC Climatology and WHOI Buoy
deployments
Only for FASINEX is the constrained field
(solid colour) closer to the buoy values
46
Comparison of SOC Climatology and WHOI Buoy
deployments
but for short wave heating it is TOGA that is
brought into better agreement
47
Summary

• Increasing the Latent Heat flux gives similar
  evaporation to the reanalysis results

• BUTcomparison with reference data suggests the
  models over-estimate evaporation

• Satellite data doesnt help!

• We need more in situ reference data

48
Beyond WOCEthe Observing System
49
Future Surface Flux estimation

• Move toward global fields from NWP models and/or
  Remote sensing (wind stress, shortwave, sst,
  latent heat? longwave? )

• Role of in situ data is increasingly for
  verification
• Flux reference Buoys
• Improved ship data (the VOS Climate project,
  VOSClim)

50
Using Flux Reference Data
51
The J-COMM VOS Climate Project VOSClim

• VOSClim initially aims to improve the meta-data
  available from the ships

PO Nedlloyd Southampton - a VOSClim ship
Observations during 2001 from ships recruited to
the VOSClim Project
52
Have We Learnt Anything?

• During WOCE we have learnt much about the error
  characteristics of our flux estimates

• Mean net heat flux fields can be brought into
  agreement with ocean heat transport values

• But we still dont have distributions of the
  component heat fluxes which give a balanced
  budget

53
However
For Surface Flux studies, the full dividend of
WOCE is still to come

• Continuing analysis of WOCE data

• New NWP reanalysis experiments

• The full exploitation of satellite data

• The Global Ocean Observing System

54
Acknowledgements
The content of this talk was based on the
conclusions of the joint WCRP/SCOR Working Group
on Air-Sea Fluxes1. However the specific
examples shown were obtained from the SOC
Meteorology team2, in particular Simon Josey and
Jeremy Grist the IRI/LDEO Climate Data
Library3 and the SeaFlux group4. Bob Marsh
supplied the title page graphic.
1 http//www.soc.soton.ac.uk/JRD/MET/WGASF/
2 http//www.soc.soton.ac.uk/JRD/MET/
3 http//ingrid.ldeo.columbia.edu/
4 http//paos.colorado.edu/curryja/ocean/
55
THE END