phd qualifying presentation

45
Interannual and decadal variations of Antarctic ice shelves using multi-mission satellite radar altimetry, and links with oceanic and atmospheric forcings Fernando S. Paolo PhD Qualifying, May 20, 2013 Scripps Institution of Oceanography University of California, San Diego

Upload: fernando-paolo

Post on 07-Jul-2015

294 views

Category:

Documents


0 download

DESCRIPTION

Interannual and decadal variations of Antarctic ice shelves using multi-mission satellite radar altimetry, and links with oceanic and atmospheric forcings

TRANSCRIPT

Page 1: PhD Qualifying Presentation

Interannual and decadal variations ofAntarctic ice shelves using multi-missionsatellite radar altimetry, and links with oceanic and atmospheric forcings

Fernando S. PaoloPhD Qualifying, May 20, 2013Scripps Institution of OceanographyUniversity of California, San Diego

Page 2: PhD Qualifying Presentation

Presentation outline

1. Background

Importance, Hypothesis, Evidence

2. Thesis

Chapter 1, Chapter 2, Chapter 3

3. Summary

Results and Big Picture

Page 3: PhD Qualifying Presentation

Why do we care?

Ice-sheet mass loss Sea-level rise

Page 4: PhD Qualifying Presentation

Why do we care?

Shepherd et al., 2012

Ice-sheet mass loss Sea-level rise

Antarctica

GreenlandGlaciers

Ice volume3 mm/yr (~1.8 from Cryosphere)

Page 5: PhD Qualifying Presentation

Why Antarctica?

The marine Ice-sheet instability

Bed above sea levelIncreased discharged with grounding-line retreat → unstable condition!

Vaughan and Arthern, 2007

Fig. M. HelperData BEDMAP

Page 6: PhD Qualifying Presentation

Why ice shelves?

Fig. Ice-shelf coverageby satellite altimetrymissions

ice shelves are the “interface” between the ice sheet and the ocean

Page 7: PhD Qualifying Presentation

Ice-shelf buttressing

Compressive stress is a result of ice-shelf buttressing

Hughes, 2011

OCEAN GROUNDED ICE

Ice rise

Confined embayment

Ice rumple

Calvingfront

Page 8: PhD Qualifying Presentation

Ice-shelf-ocean interaction100s km

1-2 km

Three modes of basalt melt (Jacobs et al., 1992)

Fig. M. Craven, AAD

Page 9: PhD Qualifying Presentation

Ice-shelf-ocean heat exchange

Jenkins et al., 2010

Melt rates of 10s m/yr

Page 10: PhD Qualifying Presentation

Ice-shelf-ocean heat exchange

Jacobs et al., 2011Jenkins et al., 2010

Melt rates of 10s m/yr

CDW all the way to the sub-ice-shelf cavity

Page 11: PhD Qualifying Presentation

Ice-shelf and grounded-ice thinning

Pritchard et al., 2012

Page 12: PhD Qualifying Presentation

Evidence on ice-shelf buttressing

Rignot et al., 2004

Page 13: PhD Qualifying Presentation

Previous studies

Zwally et al., 2005 Shepherd et al., 2010 Pritchard et al., 2012

To detect climate signals we need long and continuous records!

ERS-1/2 1992-01 ERS-2/Envisat 1994-08 ICESat 2003-08

9 years50 km

DurationSpt. Res.

14 yearsOne trend per ice shelf

5 years30 km

Page 14: PhD Qualifying Presentation

My contribution

1. Derive reliable time series of elevation

change over the longest possible time period

2. Quantify long-term trends

3. Quantify interannual-to-decadal variability

4. Identify causes of temporal and spatial

variability

Page 15: PhD Qualifying Presentation

Thesis structure

Chapter 1 → The methodology

(Generate the dataset)

Chapter 2 → Radar-Laser comparison

(Validate the dataset)

Chapter 3 → Ice-shelf variability

(Analyze the dataset)

Page 16: PhD Qualifying Presentation

Chapter 1Constructing time series of elevation change

Page 17: PhD Qualifying Presentation

Satellite altimetry missions

Twenty years of continuous data over the ice shelves

Page 18: PhD Qualifying Presentation

The challenge of multi-mission integration

Differences between missions:

– RA systems, orbit configurations, time spans...

Radar interaction with variable surf. properties:

– Surface density,

– Penetration depth,

Spatial and temporal dependent corrections:

– Ocean tide + load (for high lat)

– Atm pressure (IBE)

– Regional sea-level rise

ρs(x , t )

k e( x , t)

Page 19: PhD Qualifying Presentation

The challenge over ice shelves

Due to hydrostatic equilibrium the altimeter only see

10-15% of the grounded ice signal (in elevation

change)

So to increase signal-to-noise ratio requires lots of

averaging both in time and space

Page 20: PhD Qualifying Presentation

Averaging in time

Monthly averages

Seasonalaverages

Time steps → 3-month blocks of data

Page 21: PhD Qualifying Presentation

Averaging in space

3 x

One month of data

~750 bins with 15 to 200 observations (for FRIS)

Page 22: PhD Qualifying Presentation

Averaging time series

82 time series per bin (x 2)

61,500 time series for FRIS (x 2)

Matrix before

Matrix after

Page 23: PhD Qualifying Presentation

Inter-mission cross-calibration

ERS-1 ERS-2 Envisat

What happens when there are no data in the overlapping period?

Page 24: PhD Qualifying Presentation

The backscatter problem

ρs(x , t )

k e(x , t )

Remy et al., 2012

Penetration:

Densification:

Page 25: PhD Qualifying Presentation

Backscatter correction

hc (t)=h(t)−s g (t)−h0

Amplitude series

Differenced series

Done for each grid-cell

Elevation

Backscatter

Page 26: PhD Qualifying Presentation

Time-varying backscatter

hc (t)=h(t )−s(t )g (t )−h0(t )

ERS-1 ERS-2 Envisat

Done for each grid-cell

Page 27: PhD Qualifying Presentation

Different corrections, different results?

Amplitude ts

Differenced ts

Page 28: PhD Qualifying Presentation

Different corrections, different results?

Different fluctuation and trend

Constant correlation

Variable correlation

Amplitude ts

Differenced ts

How significant are these differences?

Page 29: PhD Qualifying Presentation

Chapter 2Envisat (radar) vs ICESat (laser) inter-comparison

Page 30: PhD Qualifying Presentation

Two altimeters, one purpose

Envisat (Radar)

– microwave (λ ~ 2.5 cm)

– wide footprint (3 km)

– all weather

– continuous sampling

– penetrates into snow

ICESat (Laser)

– visible (λ ~ 650 nm)

– narrow footprint (70 m)

– cloud interaction

– campaign mode

– top-of-snow reflection

Page 31: PhD Qualifying Presentation

Do they measure the same thing?

First time this comparison is done in this way

Page 32: PhD Qualifying Presentation

Do they measure the same thing?

Envisat ICESat

We need an explanation for such differences!

First time this comparison is done in this way

Page 33: PhD Qualifying Presentation

Two ways of estimating elevation changes

1) Eulerian (fixed):

2) Lagrangian (moving):

∂h∂ t

D hD t

=∂h∂ t

+ u⋅∇ h

A A' B(t1) (t2)

A'-A = Euler B-A = Lagrange

Page 34: PhD Qualifying Presentation

Footprint differences

ICESat footprint (70 m) is about 0.05% of RA-2 footprint (3 km)

Page 35: PhD Qualifying Presentation

Is radar Eulerian?∂h /∂ t

Page 36: PhD Qualifying Presentation

Is radar Eulerian?∂h /∂ t

Page 37: PhD Qualifying Presentation

What is signal and what is noise?

ICESat data are very noisy! How much can we trust?

Pritchard et al., 2012

Cross-over analysis Along-track analysis

Two different techniques, same pattern → features are in the data!

Page 38: PhD Qualifying Presentation

Chapter 3Variability of Antarctic ice-shelf elevations

Page 39: PhD Qualifying Presentation

Our main goal

– Search for mechanisms that could explain the

observed variability in h (x , t )

Page 40: PhD Qualifying Presentation

Ice-shelf mass balance

∂h∂ t

=∂Δ∂ t

−M∂∂ t

ρw−1

+∫0

Mdm

∂∂ t

ρ f−1

(m)

+ (ρi−1−ρw

−1) ( M s+ M b+ u⋅∇M + M ∇⋅u )

Altimeterobservation

Sea-level variations

Ocean-densitychanges Firn compaction

Ice-ocean density contrast

Surfaceaccumulation rate

Basalaccumulationrate

Advection ofthickness gradientand flow divergence

Shepherd et al., 2003; Padman et al., 2012

Page 41: PhD Qualifying Presentation

Variability within an ice shelf

We are able to resolve the “fine” spatial scales

Page 42: PhD Qualifying Presentation

Spatio-temporal change in ∂h /∂ t

Why aren't thinning/thickeningregions “fixed”?

Page 43: PhD Qualifying Presentation

Correlations, correlations...

Fig. J. Allen, NASAData NSIDC

– Sea ice protects ice shelves by cooling air temperatures and dampening waves.– Also affects mode 1of basal melt.

What is the relationto sea-ice variability?

Is there any relation to climate Indices (ENSO, SAM, ZW3)?

– EOF analysis on h(x,t)

Page 44: PhD Qualifying Presentation

Large-scale coherent events?

AMERYFRIS

ROSS

Decadal oscillationIn phase?

Page 45: PhD Qualifying Presentation

Thesis summary

Generate a 20-year long and high resolution dataset of thickness variation for all Antarctic ice shelves.

Better understand the radar altimeter signal interaction with ice surfaces, and its effect in the final estimates.

Estimate long-term trends and explain the variability in Antarctic ice-shelf thickness for the last two decades.