A New Technique for the Estimation of Sea Surface Salinity in the Tropical Oceans from OLR

1
A New Technique for the Estimation of Sea Surface
Salinity in the Tropical Oceans from OLR
Bulusu Subrahmanyam Marine Science Program
Department of Geology University of South
Carolina
2
Infrared Images
cold upwelling water south of Cape Cod.
Land and clouds have been added from the near IR.
westward moving warm eddy
Gulf Stream
West Atlantic 15th June 1996 (NOAA/AVHRR)
3
Salinity from space

• Two satellite missions the US Aquarius and
  the European Space Agency Soil Moisture and
  Ocean Salinity (SMOS) are planned for launch in
  2008.
• Provide the first global observations of SSS,
  covering Earth's surface once every 8 days
• Deliver monthly 150-kilometer resolution SSS maps
  over a 3-year mission lifetime ?
• Achieve SSS accuracy of 0.2 (psu) this is about
  a "pinch" (i.e., 1/6 of a teaspoon) of salt in 1
  gallon of water

4
Aquarius

• Aquarius satellite is targeted for launch in
  2009 for a three-year mission
• The Aquarius mission is being developed in an
  international partnership with Argentinas space
  agency, Comision Nacional de Actividades
  Espaciales (CONAE)
• Passive Salinity Sensor L-Band Radiometer
  operating at 1.4 GHz (NASA)
• Active Surface Roughness Sensor L-Band
  Scatterometer operating at 1.2 GHz (NASA)
• Microwave Radiometer (MWR) (CONAE Instrument)

5
Mean Annual Distribution of Surface Salinity in
PSU
Average Salinities
Atlantic 34.9
Pacific 34.6
Note in general the surface salinities are
higher in Atlantic than Pacific
6
http//aquarius.gsfc.nasa.gov/science.html
7
North Indian Ocean

• There are geographical similarities in Arabian
  Sea and Bay of Bengal.
• semi enclosed basins
• open to the equatorial Indian Ocean in the south
• both forced by changing monsoon winds
• receive similar amount of solar radiation

• There are also striking dissimilarities between
  the two basins.
• evaporation exceeds precipitation in the Arabian
  Sea
• precipitation exceeds evaporation in the Bay
• Annual rainfall in the Arabian Sea is 1m in
  the Bay varies between 1-3 m
• Run off from rivers into Arabian Sea is meager,
  Bay receives annual runoff of 1.5x1012 m3
• Arabian Sea is much saltier than Bay.
• Strong stratification near the surface in the
  Bay.

8
ARABIAN SEA
BAY OF BENGAL
weak convective activity
strong convective activity
P-E gt 0
P-E lt 0
weak near-surfacestratification
strong near-surfacestratification
warm SST
coolSST
strong mixing and upwelling
weak mixing and upwelling
strong winds (Findlater Jet)
weak winds
9
Technique for Retrieval of Ocean Surface Salinity

• A new technique developed for retrieval of sea
  surface salinity (SSS) from space-borne satellite
  measurements of OLR through the Effective Oceanic
  Layer (EOL).
• OLR is used to study the convection over the land
  and ocean. This is based on the idea is that
  intense convection over the ocean is associated
  with warmer surface temperatures maintained by
  low surface salinities formed due to large
  riverine input and compounded by
  convection-induced precipitation.
• SSS estimated in this way may be useful in
  improving the existing climatologies at least for
  those parts of the worlds ocean where intense
  convection is the regular feature, such as the
  tropical Indian and Pacific Oceans.

10
Effective Oceanic Layer

• The EOL is defined as the geopotential thickness
  (m2/s2) of the stratified layer and is computed
  from
• where ? is the mean specific volume anomaly
  (m3/kg) obtained from the temperature and
  salinity profile data over the pressure interval
  (dp). The integration is over the depth from
  surface to the depth (D) of stability maximum.
• The integration is carried out over a constant
  depth of 30m uniformly in the Indian Ocean, as
  the depth of the stability maximum varies between
  20 and 40m.

11
Objectives

• We developed algorithms for retrieval of sea
  surface salinity (SSS) from space-borne satellite
  measurements of Outgoing Longwave Radiation
  (OLR). It is an inverse method to derive the SSS
  from OLR.
• A preliminary assessment of the potential SSS
  observations from satellite observations of OLR
  through the Effective Oceanic Layer (EOL) is
  presented for the Indian Ocean.
• Given the sparse distribution of in situ SSS
  observations, it follows that even a threshold
  requirement of 1 PSU (Practical Scale Unit) over
  a distance of 1 degree for a time period of
  weekly or monthly will be useful in improving
  existing climatologies.
• It is envisaged that errors in the SSS obtained
  from satellite derived OLR may come close to the
  threshold requirements. This means that satellite
  derived OLR seems capable of providing SSS
  values, at least for those parts of the ocean
  that are covered by thick clouds, such as the
  tropical Indian Ocean.

12
Objectives

• To develop algorithms for the estimation of sea
  surface salinity (SSS) from space-OLR over the
  tropical oceans.
• To validate the estimated SSS using in situ
  surface salinity observations along WOCE and
  Joint Global Ocean Flux Studies (JGOFS) sections,
  along ships of opportunity tracks where salinity
  is collected, and available time-series data sets
  and cruises data
• To refine the estimated SSS for advection and
  mixing through HYCOM (Hybrid Coordinate Ocean
  Model) simulations
• To utilize the estimated SSS data to study the
  seasonal and interannual variability of SSS and
  the associated upper ocean processes
• To further utilize the estimated daily SSS to
  study the air-sea coupling during tropical
  cyclones and possibly help improving prediction
  of tropical cyclone tracks.

13
Data and Models

• Daily OLR at 2.5x2.5 grids from 1975 to 2003.
• Monthly Southampton climatology E-P from
  1980-2003
• Monthly CMAP precipitation from 1979-2003
• World Ocean Atlas 2001 Temperature and Salinity
  data
• HYCOM model- surface currents are used to
  estimate advection and Mixing Processes
• SSS physical model - to identify the lag between
  the OLR and SSS, between P-E and SSS, and between
  P and SSS, and its effect on the estimated SSS in
  the tropical Oceans

14
Computational Methodology
15
Based on WOA 98 and INSAT OLR during July in the
Bay of Bengal
16
Scatter plots in Arabian Sea
17
Scatter Plots in Arabian Sea (OLR vs EOL EOL vs
SSS)
18
Scatter Plots in Bay of Bengal (OLR vs EOL EOL
vs SSS)
19
Scatter Plots in the Equatorial Indian Ocean
(OLR vs EOL EOL vs SSS)
20
Scatter plots- Estimated Sea Surface Salinity Vs
Rainfall
21
Estimated Sea Surface Salinity from OLR during
1995
22
Estimated Sea Surface Salinity from OLR during
1996
23
Estimated Sea Surface Salinity from OLR during
1997
24
Estimated Sea Surface Salinity from OLR during
1998
25
Deviation of Estimate Sea Surface Salinity
Anomalies from WOA98 During 1995
26
Deviation of Estimate Sea Surface Salinity
Anomalies from WOA98 During 1996
27
Deviation of Estimate Sea Surface Salinity
Anomalies from WOA98 During 1997
28
Deviation of Estimate Sea Surface Salinity
Anomalies from WOA98 During 1998
29
Comparison with Levitus Surface Salinity
30
Comparison with Levitus Surface Salinity
31
Comparison with Levitus Surface Salinity
32
The seasonal variation of estimated SSS based on
these relationships agreed closely with that of
WOA98 SSS with a deviation of 0.5 PSS (Practical
Salinity Scale) over the vast part of tropical
Indian Ocean.
33
Comparison with WOCE sections
34
Comparison with observations
35
Physical Model

• Consider a volume of water in contact with the
  atmosphere, with no mixing or exchange. The
  average salinity of the volume (in terms of a
  mass ratio) is give as

Total mass of the enclosed volume of fluid
Mass of just the salt molecules in the volume
Time rate of change in volumes Salinity
We can write
36
Physical Model (Cont)

• where P is the precipitation rate, E is the
  evaporation rate, ? is the density of the water,
  and A is the area of the volume in contact with
  the atmosphere.  Using the definition of S and
  that

H is thickness of our volume
Assuming small variations in salinity and that H
is large with respect to the net amount of water
crossing the air-sea interface, we can write the
above as
where So and Ho are the mean salinity and layer
thickness respectively. For ease in writing we
will define RP-E.
37
Physical Model (Cont)

• Taking the Fourier transform of both sides of
  above equation

? is frequency, and the circumflexes denote
Fourier transformed variables. The model
co-spectrum between S and (P-E) is given by
Here 's denote the conjugate Cospectrum is
totally imaginary 
Phase between S and R is 90 (they are in
quadrature)
Model shows that for a mean annual cycle, a 90
phase lag between S anomalies and P-E anomalies
as show up in a lag correlation as maximum at a
lag of 3 months
38
Early detection of Tropical Cyclones/Hurricanes
39
Influence of tropical cyclones on Chlorophyll-a
Chlorophyll-a concentrations (mg m-3) from IRS-P4
(OCEANSAT-1) Ocean Color Monitor (OCM)
40
MODAS Salinity
Salinity estimated from OLR
(NRL Modular Ocean Data Analysis System)
41
Heat and Salt variability from Altimetry
WOA 98 T/S Diagram
Real-time T/P SSA
T, S Variability
Steric Height Anomaly profile
Real-time Steric Height change
WOA98 T,S
Reynolds SST
Adjusted Real-time T,S profile
Preliminary Real-time SST
Streic Height Anomaly Profile
Heat and Salt storage
42
Heat and Salt variability from Altimetry
43
Temperature and Salinity profiles from altimetry
WOCE I1 Section Observations
Synthetic Temperature Salinity derived from
altimetry
Difference between WOCE bservations and altimetry
temperature and salinity
TEMPERATURE
SALINITY
44
Near real-time Temperature Salinity profiles
T
WOCE I3 section 81.27E, 20S
S
45
Conclusions

• EOL parameter represents the thin, warmer and
  less saline near-surface stratified layer and
  contains the information of both the SST and SSS.
  
• The relatively higher heat content within the
  higher EOL region would help in the
  deep-convection and affects the OLR.
• The precipitation associated with the convection
  (OLR) lowers the SSS and affects the EOL.
• Thus, the close linkage between EOL, OLR and SSS
  is useful for the estimation of SSS from the OLR
  through the EOL parameter in the oceanic regions
  of intense convection such as the tropical Indian
  Ocean and Equatorial Pacific Ocean.

46
Conclusions

• This study provides algorithms through
  statistical relationships for obtaining the
  surface salinity in the tropical Oceans using
  OLR.
• We expect the oceanographers community will
  benefit from the sea surface salinity derived
  from satellite observations of OLR, for the
  obvious importance of salinity in many studies
  related to upper ocean processes.
• Whenever ocean models assimilate data and use to
  produce forecasts, it is important to investigate
  the impact of surface/subsurface salinity data on
  the forecast.
• In some oceanic areas like northern Indian Ocean,
  particularly Bay of Bengal, salinity contributes
  greatly, besides temperature, for a relationship
  with density, and investigators identified the
  consequences that arise ignoring salinity
  information, especially on the upper ocean
  thermodynamics.
• Salinity can have important consequences,
  especially on El Niño phenomena formation.
• The growing tendency for obtaining the sea
  surface of the global oceans would, if measured
  accurately, boost the coupled model studies, ENSO
  forecast models as well as ocean global
  circulation models (OGCM) or regional scale
  circulation models and predicting the tropical
  cyclones.

47
Thanks for coming !