High-frequency and high-wavenumber variability in the

California Current: Evaluating model requirements for SWOT assimilation

Sarah Gille1, Matthew Mazloff1, Jinbo Wang2,

Teresa Chereskin1, Bruce Cornuelle1, Dimitris Menemenlis2, Marcello Passaro3,

Cesar Rocha4

1Scripps Institution of Oceanography, 2Jet Propulsion Laboratory

3Technischen Universität Munchen 4Woods Hole Oceanographic Institution

SC

RIP

PS

IN

STIT

UTION OF OCEAN

OG

RA

PH

Y

UCSD

California Current: Test bed for SWOT Goal: Develop regional version of MITgcm to assimilate SWOT’s high-wavenumber measurements Build on existing regional ECCO machinery and network of observations • SWOT (swath boundaries) • Nadir altimetry (Jason) • Moorings • HF radar • Buoys (NDBC) • Glider lines

Regional MITgcm built to match MITgcm (llc4320) global model

• ~2 km resolution • Tidal forcing on

boundaries and surface

• 90 vertical levels allows internal waves to propagate

Tide in 2016 for Los Angeles replicates major features

of tide gauge observations

MBARI M2 mooring (June-Sept) Dynamic height

Time (days from 1 June)

Frequency spectra

Regional models

Global model Observations

Can a regional model generate enough internal wave energy?

Regional tests

• Mooring has high-frequency energy

• Global model (llc4320 MITgcm) replicates mooring energy

• Regional MITgcm and ROMS missing high-frequency energy

Hypotheses: • Mooring data noisy; global model too energetic • Interannual variability in observations • Open boundaries don’t let in enough energy

Regional models

Global model Observations

CCE1 mooring (June-Sept)

Time (days from 1 June)

• 2016 and 2017 differ, • … but spectra similar

• Global model replicates mooring,

• … but regional model lacks high-frequency variability

Can a regional model generate enough internal wave energy?

Regional tests

• Mooring has high-frequency energy in 2016 and 2017

• Global model (llc4320 MITgcm) replicates mooring energy

• Regional MITgcm missing energy at high frequency

Hypotheses: • Mooring data noisy; global model too energetic • Interannual variability in observations • Open boundaries don’t let in enough energy

Sea surface height wavenumber spectra 500 100 20 10 5

Wavelength (km)

10-3

10-2

10-1

Along-track wavenumber (cpkm)

10-7

10-6

10-5

10-4

10-3

10-2

10-1

po

wer

sp

ectr

al d

en

sit

y (

m2/c

pkm

)

-2

-4

global model (line 90)

model daily-averaged

regional model (line 90)

• Global model: spectra from hourly output vs daily averages

• Regional model: less energetic than global model at high wavenumbers---more like daily averages Adapted from Chereskin et al, submitted, JGR-Oceans, 2018

Sea surface height wavenumber spectra 500 100 20 10 2 1 0.2

Wavelength (km)

10-2

100

Along-track wavenumber (cpkm)

10-4

10-3

10-2

10-1

100

po

wer

sp

ectr

al d

en

sit

y (

m2/c

pkm

)

-2-4

global model (line 90)

model daily-averaged

regional model (line 90)

• Global model: spectra from hourly output vs daily averages

• Regional model: less energetic than global model at high wavenumbers---more like daily averages Adapted from Chereskin et al, submitted, JGR-Oceans, 2018

Sea surface height wavenumber spectra 500 100 20 10 2 1 0.2

Wavelength (km)

10-2

100

Along-track wavenumber (cpkm)

10-4

10-3

10-2

10-1

100

po

wer

sp

ectr

al d

en

sit

y (

m2/c

pkm

)

-2-4

95%

Sentinel 3

AltiKa

Jason 1-2 (track 195)

global model (line 90)

model daily-averaged

regional model (line 90)

• Global model: spectra from hourly output vs daily averages

• Regional model: less energetic than global model at high wavenumbers---more like daily averages

• Altimeter spectra more energetic than models from 100-50 km and flatten out (implying “noise”) for scales smaller than ~50 km.

Adapted from Chereskin et al, submitted, JGR-Oceans, 2018

Vertical velocity

Global model: llc4320 Regional model: MITgcm with open boundaries

Larger vertical velocity variance in global model

W variance at 500 m; black contour = 3000 m bathymetry

Global model (llc4320 MITgcm)

Regional model (MITgcm)

W variance at 500 m; black contour = 3000 m bathymetry

Global model (llc4320 MITgcm)

Regional model (MITgcm)

Mendocino escarpment generates tidal beam

Mendocino escarpment tidal beam less pronounced

Larger vertical velocity variance in global model

Global model (llc4320 MITgcm)

Regional model (MITgcm)

Mendocino escarpment generates tidal beam

Mendocino escarpment tidal beam less pronounced

Zoom in on Santa Barbara Channel

Larger vertical velocity variance in global model

Vertical velocity comparable in SoCal Bight, in lee of islands

Global model (llc4320 MITgcm)

Regional model (MITgcm)

Wave energy flux (u’p’)

Regional

Global 90

22

-38

-59

-55

-51

Megawatts (MW)

74

-165

Positive: energy into the domain. Negative: energy out of the domain.

North West South

Summary and Conclusions • Small-scale and high-frequency

processes occur in the California Current region in observations and global model, but not in regional model

• Energy originates outside of regional domain (e.g. Hawaii and western Pacific).

• Future work: Regional models that represent internal waves will need a new strategy to input energy at open boundaries.

Global

Regional

Vertical velocities

Global model: llc4320 Regional model: MITgcm with open boundaries

Characterizing the California Current

Acoustic Doppler Current Profiler data: 1993-2004, 39 cruises

Helmholtz decomposition to separate rotational and divergent components

Chereskin et al, submitted, JGR-Oceans, 2018