Challenges and Opportunities for new ocean observations with the SWOT Mission�Surface Water and Ocean Topography (SWOT)

Rosemary MORROW, LEGOS/CNES (FR)

SWOT:

​

1st global wide-swath altimetry mission using SAR-Interferometry

​

3 science communities : Hydrology, oceanography & coastal objectives.

​

Orbit to 78° latitude designed for tides & internal tides

​

Launch : Dec 2022

• Opportunities in observing the ocean SSH from 10-100 km wavelength & validating ocean models at these scales

​

• New altimetric technology and small-scale dynamics

• Dynamical contributions from tides, internal tides,

​

• Techniques for inversion/data assimilation, including using stochastic processes

​

​

​

​

Link with STUOD discussions

« Increased resolution in ocean models needs to be accompanied by higher resolution observations. »

Carl Wunsch, Lisbon 2010

Satellite altimetry : Ocean sea surface height observations

Sea Ice height & SSH in leads

Temporal Mean SSH : ocean geoid & Bathymetry (100 m)

Barotropic tides & Internal Tides High-frequency < f (1-10 m)

Ocean SSH is a sum of many signals with different space-time characteristics

Depth-integrated surface height & derived geostrophic currrents “Balanced motions” (cm)

Small Mesoscale processes Horizontal Transport & mixing

(Sub) mesoscale Fronts & eddies

50 % vertical transport

Predominant in deep winter mixed layers

Satellite Altimetry monitoring « dynamic height  »

5

Today’s altimetry maps monitor :

• Global & regional sea level rise & geostrophic circulation
• Deep, large mesoscale eddies

> 100 km diameter

​

Missing smaller, rapid dynamics :

• Open ocean
• Marginal seas,
• High-latitudes
• Coastal (0-10 km)

(Heat, carbon, nutrients)

Mapped altimetric geostrophic currents > 100 km & horizontal advection

Day 0

Day 10

Day 15

Chlorophyll Day 15

Credits : F. d’Ovidio

• horizontal stirring structuring the tracer fields (SST, SSS, ChlA)
• Lagrangian particle tracking & statistics, eg Lyapunov Exposants (FSLE, FTLE, …)
• Improved strain/deformation driving vertical transfers

Zhang et al., 2019, Nat Comm.

Altimetry derived FSLEs and Chlorophyll pattern on 7 Jul 99 (Seawifs, NE Atlantic spring bloom)

Today’s mapped ocean 2D geostrophic currents : �30 years of multi-mission nadir altimetry mapping

Corrected alongtrack sea-surface height data are mapped onto a regular grid :

• editing, filtering,
• Correction for long wavelength errors
• optimal interpolation scales > 10d, > 150 km

2D SSH & currents – Jason-2 only

2D SSH & currents –

4 nadir satellites

2D currents & SST

SWOT : Surface Water Ocean Topography

Innovation in 2D altimetry : wide-swath SAR/interferometry

Image Credit : K. Wiedman

50 km wide swaths; high precision for 1-2 cm SSH

SWOT (2022)

Guanlan, S3-NGT?

SAR interferometry

SWOT -2D data products

Image Credit : K. Wiedman

Over Land : Full resolution within land mask

Resolve Rivers > 100 m, Lakes > 250m²

Higher noise

​

Over Oceans : Onboard Processor performs initial SAR-Interferometry processing reduces data download volume

​

Basic SSH product

(2 km resolution/posting)

(0.6 GB/day) σ = 1.37 cm

​

High resolution product: SSH & SAR images (250m posting / 500m res)

(37 GB/day) σ = 5.48 cm

Comparison SWOT/S3 spatial coverage

S3-A & S3B (27 days)

SWOT (21 days)

• Global coverage only possible with wide swath altimetry -> many new coastal and hydrology zones will be seen
• Nadir altimetry miss many smaller coastal regions & water bodies, no river slopes or coastal currents < 100 km

SWOT Ocean Opportunities

SWOT Ocean Opportunities

Opportunities :

• Improved observation of ocean currents & energy cascade down to finer scales
• Finer horizontal currents to derive vertical velocities : study the ocean’s capacity to stock heat, carbon and nutrients
• Extending 2D SSH/currents into coastal/estuary regions
• Improved 2D coverage of coastal and high-latitude tides and marine gravity fields
• Observe the dynamical interaction between ocean eddies / jets and internal tides => important for dissipation & mixing

SSH of Internal Tides

Gulf Stream vorticity

Today altimetric 2D sea surface height maps resolve ocean scales of 150-200 km

=> SWOT aims to observe 2D ocean dynamics at 15-40 km effective resolution

​

Geostrophic currents & vorticity – �why horizontal SSH resolution is important

Chelton et al, PO, 2019

Mapping of SWOT-like observations of SSH, currents, vorticity

​

• More anisotropic
• Sharper gradients in currents & vorticity

Today’s Mapping of AVISO-like observations of SSH, currents, vorticity

SSH

Geostrophic Currents

Geostrophic Vorticity

WINTER

SUMMER

6

ζ/f

ζ/f

w

w

0

500

Winter submesoscales have energetic vertical velocities

​

Small scales important in horizontal transport of energy, & heat, freshwater, nutrients

​

Frontal regions / eddy interactions = > small horizontal scales drive strong ocean vertical velocities

​

Small scales induce 50% of global vertical heat and CO2 pumping and vertical transfers of nutrients, biomass.

​

Not included in today’s climate models

​

​

Sasaki et al, 2012

Importance of sub-mesoscale heat fluxes at depth

​

Modelled & in-situ submesoscales show a net upward

heat flux at depth

​

These processes are not included in CMIP climate models,

not well parametrised

​

Surface 2D strain from SWOT, combined with sQG dynamics … reconstruct the deeper submesoscales

Siegelman et al., 2020

SAR-Interferometry : Colocated all-weather heights and SAR imagery for ocean front detection

credits Ifremer

Upper panel : modelled surface currents, SST & vorticity across the Gulf Stream

Lower panel : simulated SWOT swath of SSH, geostrophic currents and SAR surface roughness

SST and surface current (18 km res.)

Vorticity & Surface current (18 km res.)

Altimetric SSH can act as a global network of tide gauges

Tides and internal tides

SWOT orbit chosen to resolve tides – 3 years of SWOT data will provide finer-scale 2D tides

–> non sun-synchronous : designed to resolve tides

-> Barotropic tides improved in coastal and high-latitude regions

-> 2D signature of Internal tides – important for ocean mixing & energy dissipation

Model : HYCOM 1/12° Arbic et al., 2012

~20 m

~2 cm

St. Dev of 7 global tide models, M2

Stammer et al., 2014

Errors in Barotropic Tide models

​

Non-linear interactions between Internal tides and mesoscale turbulence

-> instabilities, mixing, dissipation

Ponte & Klein, 2015

Interaction of internal tides and eddies: Amazon coast

Internal tide energy is deviated and dissipated by ocean eddies

• 2 seasons
• Different spring and neap tides
• 2 contrasted mesoscale eddy conditions

Are Model results true?

Need small-scale SWOT observations to find out!

​

Baroclinic tidal flux

Vorticity around eddies

ASOND Spring Tide

ASOND Neap Tide

MAMJJ Spring Tide

Tchilibou et al., 2022

​

Altimeter obs : AL & J2 (- BT tide)

Model spectra :

No Tide : 1/36° - 5d

No Tide : 1/36° - 1h

​

W Tide (- BT tide) :

Full internal Tide : 1/36° - 1h

Coherent BC modes (M2 only)

​

​

SSH Wavenumber Spectra modified by internal tides

Tchilibou et al., 2019

SSH wavenumber spectra altimetry & models

SSH variability ; 1/36° model :

LF

HF

Altimetry

Model with NO Tides

Model with Tides

Challenges of SWOT’s new SSH observations 15-200 km

SWOT orbits

First 6 months : 1-day orbit :

1st 3 months – instrument checkout

2^nd 3 months – Mar-May 2023 – Validation

• Ideal for ocean studies of rapidly evolving small mesoscales and submesoscales

3-years in 21-day repeat orbit

Nominally : 2023 to 2025

Full global coverage

1-day and 10-day sub-cycles for better mesoscale coverage

​

Orbit kmz files available on AVISO+ SWOT orbits

​

Nominal Launch date : Dec 2022

​

​

After 3 days

After 21 days

​

New solutions needed to maintain fine-scales in observation gaps :

• Combine SWOT & nadir altimetry
• Dynamical interpolation with simple models (MIOST); Simple models & stochastic processes; AI & full assimilation techniques

Day 17 mapping ?

2D reconstruction using simple dynamical QG models:

SWOT has excellent local spatial coverage of SSH & geostrophic velocity, but weaker temporal coverage and small ocean dynamics evolve rapidly!

‘static’ OI’

‘dynamic’ Interpolation

Truth at day 17

From all SWOT obs between days 2 and 32

From all SWOT obs between days 2 and 32

Day 17 mapping

2D reconstruction using simple dynamical QG models:

Open ocean example : Ubelmann et al., 2014

S. Keating, UNSW, 2021

Fig. 2.

3D Reconstruction using Effective sQG

Surface vorticity

Vertical velocity

Full model

eSQG reconstruction

(Qiu et al,JPO, 2016)

Ex : Using simple models to solve for balanced & wave dynamics

27

Solving the shallow water momentum equations (Remove the BT tide)

​

QG dynamics for the balanced flow

​

Linear SW wave model for the internal tides

​

Data assimilation with reduced bases

​

Le Guillou et al., 2022

Freq-λ Truth

Freq-λ Reconstruct

SWOT In-situ Validation « Adopt a Crossover »

https://www.swot-adac.org/

• CLIVAR endorsed project for an international multi-site in-situ deployment under SWOT swaths and crossovers in 2023
• Development of SWOT-supported in situ campaigns for fine-scale dynamics in different regions and seasons
• In-situ data sets for validation of ocean models & inversion/assimilation

​

• Partners : US, France, UK, Canada, Australia, S Africa, Norway, Turkey, China, Argentina…

US SWOT Project’s Californian Crossover campaign

Validation of spectral requirements

GPS buoys & bottom pressure gauges

CTD Moorings & Wirewalkers 0-500m

Lidar for 2D SSH & SWH

Gliders

Lidar & gliders/drifters for 2D coverage

SWOT Ocean Final remarks

​

• SWOT will give us improved sea surface height observations of 2D structure of small mesoscales, and derived balanced currents, vorticity

​

• SWOT observes 2D SSH and SAR images … we need innovative ways to combine this information with multi-satellite & in-situ data (drifters, gliders..)

​

• SWOT will provide unique 2D observations of the interactions between balanced flow and internal tides

​

• Mapping the fine-scale SWOT SSH swath data onto regular 2D /3D fields presents many challenges, explored using interpolation techniques, different vertical projection schemes, full assimilation techniques, and including stochastic processes.

​

• Unique data set for validation of high-resolution ocean models and simple dynamical models including stochastic processes

TimeLine

​

Early Dec 2022 : SWOT Launch into 1-day orbit for 6 months

Dec 22 – Feb 23 : Instrument Checkout

Mar 23-May 23 : Science CalVal : 1-day

​

Jun 23 : SWOT moved to 21-day nominal orbit for 3 years

Sep 23 : 1^st processed swath data available for science validation

2024 : Validated swath data available

mid 2024 : first gridded DUACS maps with SWOT

​

​

Info : SWOT Mission, orbits, data products, Science Team : http://swot.jpl.nasa.gov & www.aviso.altimetry.fr => swot

​

SWOT Ocean Simulator (orbits & errors) :

https://github.com/SWOTsimulator

​

Thank you

Practicalities

• Orbits
• Mask for High-resolution data over land
• Ocean SSH and SAR imagery
• Ocean data Products
• SWOT Ocean simulator

LR data available globally to 78°; HR on Mask

34

Possibility of adjusting the HR mask up to 4 times per year (ex. for seasonal variations)

+ 4 patches of120x120 km² over ocean / sea-ice

Journées Nationales SWOT – Toulouse 22-24 Novembre 2016

35

• Basic SSH product (2 km resolution/posting) (0.6 GB/day) σ = 1.37 cm
• Expanded products (2 km):

Full corrections (2.4 GB/day)

Wind/waves/s₀ (0.6 GB/day)

• Expert high resolution product: SSH & SAR images (250m posting / 500m res)

(37 GB/day) σ = 5.48 cm

Global ocean data products – at 250m x 250m and 2x2 km

Tools : Preparing for SWOT – ocean simulator

SWOT 2D high-resolution SSH – capturing the eddy anisotropy and strain

Precise 2D horizontal gradients needed for velocity and vorticity

​

SWOT simulator to study successive passes over a region using your model (open source : https://github.com/SWOTsimulator/swotsimulator.git

Eg : HR model of Bay of Biscay and Gironde Estuary … with SWOT swath and noise

Toublanc et al.

Model SSH

SWOT MAY 27

SWOT May 30

KaRIn LR and HR Modes: Onboard vs. Ground Processing

Raw images

(complex)

L0

L1

Focused images�(complex)

Geolocated heights, slope,

SWH, discharge, storage change…

SAR �proc.

Amplitude

Phase

Master Slave

Coherence

L2

LR: x9

LR: x9

Interferograms�(complex)

​

KaRIn/SWOT Simulations :

LR : Low Resolution (250 m - 2 km) data : over all surfaces from Onboard Processor

HR : High Resolution (5.5 m x 10-70 m) over Hydrology Land targets & 3km from coasts

Innovation in altimetry : SAR & SAR/interferometry

Signal & noise averaged over

5-7 km radius footprint

Medium noise

​

300 m alongtrack,

5 km radius crosstrack

Low noise

Jason-class & Saral

Conventional Mode

Sentinel-3 (2016) &

Sentinel-6 MF (2020)

SAR nadir Mode

250m x 250m (or 2km²),

50 km wide swaths

Very low noise

SWOT (2022)

Guanlan, S3-NGT?

SAR interferometry

Altimetric SSH observes the marine geoid & can reflect bathymetric changes

Altimetry & marine gravity field

Satellite Altimetry has always been used to determine the marine gravity field (70% of ocean surface)

Large-scale gravity anomalies linked to Earth’s internal mass distribution

Small-scale gravity anomalies can be inverted to derive bathymetry

SWOT 2D Mean Sea Surface slopes

Today’s gridded Mean Sea Surfaces and marine gravity anomaly fields have large 2D errors

• in the groundtrack diamonds
• from geodetic missions extending to higher latitudes (Geosat, ERS, Cryosat, …) have a biased N-S sampling near the equator

​

• SWOT will provide consistent N-S and E-W sampling at 2 km or 250m resolution to 78° in latitude, for improved MSS, marine gravity fields & open-ocean bathymetry

Differences between 2 state of the art MSS products, showing the groundtrack patterns, and higher E-W errors near ridges and sea-mounts

Challenge 1 : Removing�Systematic SWOT errors

Interferometry depends on a good estimate of

1. Platform Roll
2. phase errors

c), d) baseline dilation

e) Timing Errors

​

These errors can reach 20 cm !

True Ocean SSH

Residual errors after crossover calibration :

21 day orbit

1 day orbit

20 cm

2 cm

Challenge 2 : Removing Karin random noise (varies with sea-state): 

SSH

First derivative : currents

Second derivative : vorticity

Truth

Noisy data

• Users need a SSH swath product with errors removed (e.g., assimilation)
• A filtering step must be done

​

Neural Network filtering of random noise

→ Noise is correctly removed : the error made by the U-Net is around 1 cm

• No artifacts near the coastline
• The first derivative is correctly restored

• Swath example: