Complex Antarctic ice-shelf height changes revealed by eighteen years of satellite radar altimetry

Complex Antarctic ice shelf thickness change revealed by 18 years of satellite radar altimetry

Fernando S. Paolo*Helen A. FrickerLaurie Padman*fpaolo@ucsd.edu

AGU Fall Meeting 2014

18 years of continuous satellite radar altimetry over the ice shelves

1994 2012

ERS-1 ERS-2 Envisat

How and what satellite radar altimetry measures

ocean tides atm pressuresea-level risesurf. densificationsurf. penetration ...thickness change

∂h/∂t :

High crossover density3-month time bins30-km spatial cells

We averaged many observations in time and space

Averaging data points

35 days of crossovers

# of crossovers per cell per 3 months

We constructed several records and stacked them to enhance the signal-to-noise ratio

Incoherent

The spatial distribution of crossing points changes with time due to “repeat” satellite tracks not being exact, miss-pointing errors, etc.

Averaging time series

1994 2012

18 years of volume change in East Antarctic ice shelvesVo

lum

e

1994 2012

Is the trend a

Line?

Quadratic?

Cubic?

Volu

meHow do we represent trends?

Tibshirani, 1996

misfit

bias-variance tradeoffWe select λ by

Cross-Validation

min || h − t β ||2 + λ || β ||1Using Lasso:

constrains bias

constrains variance

We fit polynomials: ĥ(t) = ∑ βn tn + ε

Regional time series of thickness change

Sustained ice loss since 1994

East Antarctica ice shelves in phase?

Ice-loss acceleration

5-year straight-line fit (ICESat period)

18-year polynomial fit

We derived average rate of change from the18-year polynomial fits

%-Thickness change 1994-2012

5%loss

5%gain

500 km

25 10 0 10Rate of thickness change

(m/decade)

19941997

20002003

20062009

2012

¡1500

0

+1500

Volume change (km3 ) West East All

Queen Maud

Amery

WilkesAmun

dsen

Bellin

gsha

usen

Larsen

Filchner-Ronne

RossROE

ROW

WIT

SUL

NICGET

DOT

CRO

THW

PINCOS

ABBVEN

STA

BAC

WIL

GEO

LAB

LACLAD

RONFIL

BRU

RII

FIM LAZBAU

PRI

AME

WES

SHA

DIBMERCOO

REN

MAR

DRY

WestAntarctica

EastAntarctica

TOT

MOS

HOL

Some ice shelves have lost up to 18% of their thickness in 18 years

%-Thickness change 1994-2012

5%loss

5%gain

500 km

25 10 0 10Rate of thickness change

(m/decade)

19941997

20002003

20062009

2012

¡1500

0

+1500

Volume change (km3 ) West East All

Queen Maud

Amery

WilkesAmun

dsen

Bellin

gsha

usen

Larsen

Filchner-Ronne

RossROE

ROW

WIT

SUL

NICGET

DOT

CRO

THW

PINCOS

ABBVEN

STA

BAC

WIL

GEO

LAB

LACLAD

RONFIL

BRU

RII

FIM LAZBAU

PRI

AME

WES

SHA

DIBMERCOO

REN

MAR

DRY

WestAntarctica

EastAntarctica

TOT

MOS

HOL

West Antarctic ice shelves:rate of volume loss increasedby ~70% in the past decade East Antarctic ice shelves:

earlier increase in volume stopped in the past decade

All Antarctic ice shelves: rate of volume change -25 km3/yr (pre-2003) ➔ -310 km3/yr (post-2003)

Average rate for different

4.5-year intervals

Short-term ratesare highly variable, and depend on the chosen time interval!

Volume loss from Antarctic ice shelves accelerated in the last decade

Rate of volume loss from WAIS increased by ~70% in the 2000s

West Antarctic ice shelves have lost up to 18% of their thickness in less than 2 decades

Single satellite missions are insufficient to infer the long-term state of the ice shelves

Summary

-9 deg isotherm moving southward

Cook & Vaughan, 2010

Regional atmospheric warming trend?

Different forcings within each environmental setting?

Pritchard et al., 2012

Characteristic signature of ocean-induced basal melting

Faster thinning rates near the grounding lines ➔ increased CDW inflow

A reliable and complete ice shelf mask is a problem.So we (Geir Moholdt) created our own using all data available: MOA (Scambos et al. 2007), ASAID (Bindschadler et al. 2011), InSAR (Rignot et al. 2011), ICESat (Fricker/Brunt et al. 2006-10)

We explore all possible time combinations

These are elevation changes with respect to

different epochs

D J F M A M J J A D J Fsummer autumn winter spring

h(t0) h(t1) h(t2) h(t3) h(tN)

Δh0,1 Δh0,2 Δh0,3 Δh0,... Δh0,N

Δh2,3 Δh2,... Δh2,N

Δh1,2 Δh1,3 Δh1,... Δh1,N

Δh(t,t0)

Δh(t,t1)

Δh(t,t2)

...

Why can we do this? The spatial distribution of crossovers changes with time

As a consequence of hydrostatic balance...

This limits detection of changes in the vertical component

≈