REGINA presentation
Ingo Sasgen& REGINA consortium
www.regina-science.eu (Mark Drinkwater, ESA); V. Klemann, L.
Petrie, P. Clarke, N. Schn, J. L. Bamber, R. Pail, M. Horwath, A.
Horvath
STSE CryoSat+ Cryosphere StudyRegional glacial isostatic
adjustment andCryoSat elevation rate corrections in
Antarctica(REGINA)Contract-Nr.: 4000107393/12/I-NBContact:
[email protected]
REGINA (www.regina-science.eu): Ingo Sasgen (PI), Martin
Horwath, Volker Klemann, Elizabeth J. Petrie, Nana Schoen, Roland
Pail, Alexander Horvath, Jonathan L. Bamber, Peter J. Clarke,
Hannes Konrad and Mark R. Drinkwater (ESA)Page Nr.ESA CliC
workshop, Troms, 20 January 20151Glacial-isostatic adjustment
Page Nr.ESA CliC workshop, Troms, 20 January
20152Glacial-isostatic adjsutment describes how the Earth deforms
under the pressure of a ice sheet and then returns to the its
initial condition after melting of the ice.For example, during the
last glacial-maximum about 20 ka BP, North America covered by ice
of about 4 km thicknessThese surface loads deformed the lithosphere
several hundred meters downward.After melting of the ice sheet
Earth started to rebounded. Since its mantle acts like a high
viscous liquid, this process is strongly delayed and still ongoing.
It is visible today as land uplfit measureable by GPSBut GIA also
represents a large signal in the GRACE data as due to the inflow of
mantle material to the formerly glaciated regions.Modelling of the
GIA signal is done with viscoelastic earth models.
Ice sheetBedrockOceanShelfGlacial-isostatic adjustment
(GIA)Flow-line GIA simulation, example[1] Hannes Konrad, Ingo
Sasgen, Volker Klemann (GFZ)Page Nr.ESA CliC workshop, Troms, 20
January 20153Approaching GIA in AntarcticaHybrid GIA estimate
Numerical modellingEarth / Ice models, Paleo observations etc.
Earth observationGPS, GRACE, AltimetryGIA estimateGIA
predictione.g. Whitehouse et al. 2012Ivins & James
2005(IMBIE-models)e.g. Riva et al. 2009, Gunter et al. 2014Sasgen
et al. 2013
REGINA aims:Determine GIA elevation rate correction for
CryoSat-2Investigate Crustal Low Viscosity ZoneEmploy multiple
space-geodetic data setsPage Nr.ESA CliC workshop, Troms, 20
January 2015Two principal approaches towards determining GIA in
Antarctica1) Numerical modelling; thermomechanical ice sheet
models, viscoelastic Earth models, constraints: geomorphological
data, relative sea-level, Paleoclimate, etx.2) Geodetic estimate:
separate GIA and ice-mass balance by the different signatures in
the geodetic data sets. Gunter et al. 2014, altimetry + gravimetry
but without GPSMix of both approaches Way forward
Present and past ice-mass change Accumulation eventDeglaciation
since LGMPage Nr.ESA CliC workshop, Troms, 20 January
2015Conceptual separation of ongoing and past ice-mass
change.Present-day: i.e. occurring during measurement period. e.g.
accumulation event: mass increase increase in gravitational
potential (GRACE), increase in ice sheet surface elevation
(altimetry) and subsidence of bedrock (GPS)Past, i.e. occurring
before the time of measurement. But signature in observations due
to delayed Earth response. Loading relevant for few hundred to
thousands of years, depending on viscosity of the Earth mantle.
Some GRACE and GPS complementary w.r.t. present-day and past
ice-mass change. Therefore great for separation.Observation
equationsSystem of linear equationsGRACEGPSICESat/Env.Present mass
changePast mass changeCompaction (density
ch.)Kernels*:Ice-elevation changeGravitational potential Bedrock
displacement
ElasticViscousSatellite observationsEarth response* Filtered
according to observationConsideration of Earth structures &
FilteringReprocessed dataSolution with discs on geodesic gridPage
Nr.ESA CliC workshop, Troms, 20 January 2015Response linear w.r.t.
load, therefore problem can be formulated as simple system of
linear equations. Lefthandside: data sets (individually
filtered)Righthandsie: unknowns (present-day ice-mass change, past
ice-mass change, and pure density change = no signal in GPS and
GRACE)Matrix consists of response functions to individual forcing.
Response functions to disc-load functions calculated with
viscoelastic Earth modelNext, preparation of data
setsAltimetryICESat / Envisat combination for 2003-2009 [1]
Combination maskICESat available onlyICESat (smaller
errors)Envisat (smaller errors)
N. Schn, J. Bamber, Univ. Bristol10 km gridded dataPage Nr.ESA
CliC workshop, Troms, 20 January 2015Decision for combining Envisat
/ ICESat based on errors. Get signal at ice sheet margin and
tighten polar gap (omission area of radar) Reduces some of the bias
of LA within East Antarctica
7Gravimetry
A. Horvath, R. Pail (TU Mnchen), M. Horwath (TU Dresden)Page
Nr.ESA CliC workshop, Troms, 20 January 2015GRACE linear trends for
time period of ICESatFiltering based on Swenson & Wahr,
optimized to effectively reduce noise, while minimizing signal
degradation. Monthly error levelsGaussian 200 km smoothing after
destriping8GPS displacements
availableunavailablepartially available[1] L. Petrie (now Univ.
Glasgow), P. Clarke (Newcastle Univ.)Trend estimation accounting
for colored noiseReview of metadataTest of processing optionsPage
Nr.ESA CliC workshop, Troms, 20 January 2015Re-analysis of GPS data
for AntarcticaReview of metadataTest of processing options (e.g.
atmospheric correction)Trend estimation accounting for coloured
noise.9
Geometrical set-upShear velocity anomalies[1] with respect to
PREM[2]
Spatial distributionof discsEast Ant. Rheology (200 km Elast.
lith.)West Ant. Rheology (variable)1175 discs
Deviation (%)
[1] Danesi & Morelli (2001); [2] Dziewonski & Anderson
(1981)Earth structure dependent response kernels[3]
[3] V. Klemann, I. Sasgen (GFZ)Possibility: GOCE constraintsPage
Nr.ESA CliC workshop, Troms, 20 January 2015Geometric set up
showing the placement of the disc response functions in Antarctica.
Rheological difference between East and West is considered, as
indicated by seismic imagingEast Antarctic rheology fixed. West
Antarctic varied to assess influence of Earth structure.Step 1:
first-order GIA estimateRemoval of surface-mass contributions
GRACE [1]Altimetry [1]
[1] Post-processing I: Swenson & Wahr, 2006 + Gaussian 200
km (GFZ RL05)=
Different scalePage Nr.ESA CliC workshop, Troms, 20 January
2015First-order GIA estimate.Assumption: all height changes caused
by ice in altimetry. Altimetry filtered identical to
GRACESubtraction gives first-oder gravity signal of GIA (not
bedrock displacement) GIA estimate for Antarctica
Page Nr.ESA CliC workshop, Troms, 20 January 2015Results: Left
first order GIA estimate. Right after correction with GPS.
Pronounced improvement: Antarctic Peninsula. Signal from altimetry
too large positive. Density of ice leads to overestimation of
positive mass. Residual signal (GIA) strongly negative. After
correction with GPS uplift as expected.
12Influence of Earth structure
WeakStrongWeak+FilteringIMBIEPage Nr.ESA CliC workshop, Troms,
20 January 2015Sensitivity to Earth model. Upper left: extreme case
weak lithosphere + ductily layer. GIA gravity changes produce
uplift up to 3 cm/yr, locally.Upper right: extreme case thick
lithosphere, signature much smoother und only up to 10 mm/yr. Good
agreement in magnitude and location with W12a. But differences
remainLower left: Same as upper left, but with a posteriorio
filtering applied. More robust as inversion for uplift rates can
produce artefacts, away from GPS stations
Separation of ice mass and GIAIce-mass balance 2003-2009GIA
estimatePage Nr.ESA CliC workshop, Troms, 20 January 2015Approach
allows separation of GIA and ice-mass balance. And the components
can individually be quantified. As by product ice-mass balanceMass
balances next slide14GIA apparent mass change (selection)
But again: spread GIA corrections > GRACE signalREGINA
2003-2008:-63 to -75 Gt/yr (IMBIE: -72 Gt/yr)Page Nr.ESA CliC
workshop, Troms, 20 January 2015IOM validation of REGINA
estimates
IOM [1] mass balance: 66 Gt/yr (c.f. Rignot, IOM-IMBIE: 142
Gt/yr) [1] Depoorter, M. A., J. L. Bamber, J. A. Griggs, J. T. M.
Lenaerts, S. R. M. Ligtenberg, M. R. van den Broeke, and G. Moholdt
(2013), Calving fluxes and basal melt rates of Antarctic ice
shelves, Nature, 502(7469), 89-92.REGINA mass balance: 63 to -75
Gt/yr (Envisat/ICESat & GRACE combination)
Page Nr.ESA CliC workshop, Troms, 20 January 201516REGINA Phase
1 conclusionsCombination algorithmIncludes GRACE, Envisat/ICESat,
GPSAccounts for different filtering / resolutionsSolves for local
density changes (no a priori model req.)Refined Earth structure
considersCrustal low-viscosity zone East and West Antarctic
rheologyLow mantle viscosities (beyond IMBIE range; not
shown)ResultsGIA apparent mass change of 26 to 38 Gt/yrIce-mass
balance of 63 to 75 Gt/yr (IMBIE 72 Gt/yr)Next steps: final GIA
product in REGINA Phase 2 (08/2015)
Page Nr.ESA CliC workshop, Troms, 20 January 201517
www.regina-science.eu
Page Nr.ESA CliC workshop, Troms, 20 January 201518Apparent mass
change of GIAIMBIE (not used)IMBIE (used)Post IMBIEGIA
correctionICE-5GIJ05_R2W12aAGE1REGINAICE-6GIJ05_R2_5kTypeNumerical
predictionNumerical predictionNumerical predictionGeodetic
estimateGeodetic estimateNumerical predictionNumerical
predictionData set usedGeomorphology, Paleoclimate, relative
sea-level data (GIA)Geomorphology, PaleoclimateIce-dynamics,
Paleoclimate, geomorphology, relative sea-level data (GIA)GPS, GIA
ensemble modellingICESat/Envisat, GRACE, GPS, GIA
kernelsGeomorphology, Paleoclimate, relative sea-level data
(GIA)Geomorphology, PaleoclimateSpatial resolution (SH degree)
provided-2566012096-256Earth model assumptionsYes
(implicit)YesYesNoNoYes (implicit)YesReferenceSasgen et al. 2013;
doi:10.5194/tcd-6-3703-2012Ivins et al. 2013;
doi:10.1002/jgrb.50208Whitehouse et al. 2012b;
doi:10.1111/j.1365-246X.2012.05557.xSasgen et al. 2013;
doi:10.5194/tcd-6-3703-2012www.regina-science.euArgus et al. 2014;
doi: 10.1093/gji/ggu140 Velicogna et al. AGU 2014,
G51A-0343Apparent mass change (Gt/yr)140 to 18040 to 65ca. 604926
to 38107ca. 120 GRACE minus REGINA GIA, 2003-2008GRACE minus REGINA
GIA, 2003-2014-63 to -75 Gt/yr (IMBIE: -72 Gt/yr)-100 to -112
Gt/yrPage Nr.ESA CliC workshop, Troms, 20 January 2015
CSR RL05ICESatCSR RL05@ ICESatCSR RL05& ICESat[selected]
Data availabilityToo noisy for GRACEBestAdditional signalsPage
Nr.ESA CliC workshop, Troms, 20 January 2015
Geoid rate (e) to disc loadhl=90 km (thick)Ast.= 1 1018 Pa
s(weak)
Load + Elastic responseLoad + viscoelastic responseRelaxation,
without loadJmax=2048(Model)Load + Elastic responseRelaxation,
without loadEach line: response after +t = 10 yrs;Sim. period: 2
kyrsPage Nr.ESA CliC workshop, Troms, 20 January 201521Displacement
rate (u) to disc load
Elastic responseViscoelastic responseRelaxation, without
loadingEach line: response after +t = 10 yrs;Sim. period: 2
kyrsLoad dimensionElastic responseRelaxation, without loading
hl=90 km (thick)Ast.= 1 1018 Pa s(weak)Page Nr.ESA CliC
workshop, Troms, 20 January 2015Each line -> time interval of 10
yr until 2000 yr.Time (ka) -> Time (kyr)
Inversion ohne elastische Lastaenderung und transiente
Phase.Dies ist motiviert, da bereits nach < 100 yr der Zustand
schon fast relaxiert ist.
22Separation ice-mass and GIA (geoid)
Filter 2:Statistical filter+ Wiener filterFilter
1:REGINAPresent-day ice-mass changeGIAPage Nr.ESA CliC workshop,
Troms, 20 January
2015C:\Users\sasgen\Dropbox\cryoitt\deliverable\D5.1 Impact
Assessment Report\Apparent_Volume_change.xlsx23Radial displacement
rate for disc load
3x1019 Pa s(stiff)90 km 3x1019 Pa s(stiff)30 km(strong)(weak)V.
Klemann, I. Sasgen, GFZWeaker lithosphere more localized &
greater amplitudes
Viscosity not important within REGINA context standard West
Antarcticaweak West AntarcticaMention paper: GOCE Page Nr.ESA CliC
workshop, Troms, 20 January 2015These are example response
functions. Varied parameters are: thickness of the lithosphere
between 30 km (very thin) and 90 km (standard). Also in- and
excluding ductile layer (possible thin viscous layer in the elastic
lithosphere), further weakening the lithosphere. Viscosity also
varied but not important for steady state runs
REMOVE24
Procedure of separation
Page Nr.ESA CliC workshop, Troms, 20 January 2015REMOVE
Step 2: Refinement with GPSPage Nr.ESA CliC workshop, Troms, 20
January 2015Next, correct the first order GIA estimate using GPSWhy
should this work?Y-axis: GPS uplift rate corrected for mass change
observed with altimetryX-axi
s: Same but GRACE minus altimetry (plot before)
REMOVE 14-1626Step 2: Refinement with GPS
Viscous response(past loading)Viscous response(past
loading)Elastic response (present loading)Elastic response (present
loading)East Ant. rheol.
West Ant. rheol.Jmax=2048Jmax= 90 & 200 km GaussianPage
Nr.ESA CliC workshop, Troms, 20 January 2015Viscoelastic response
should plot along orange line (and depend on the Earth model
assumptions)Elastic on green line (only one set of elastic Earth
parameters: PREM)Advantage: each dataset can be considered in the
adequate resolution27GRACE e vs. GPS u (Altim. removed)
East Ant. rheol.
West Ant. rheol.Page Nr.ESA CliC workshop, Troms, 20 January
2015This is how the GPS data plot (red circles without error bars
>10 mm/yr, estimated manually)Some structure visible. But data
reveals mixed signals (remaining elastic, i.e. presnet-day ice-mass
change and GIA).Estimation using system of equations shown
earlier.28Elastic and viscoelastic kernels
Axisymmetric disc load: 62 km radius, mass gain 5.6 Gt/yr
Temporal evolutionJmax=2048~ 10 kmRadius constantLoad rate constant
Viscoelastic Earth model: Martinec, 2000V. Klemann, I. Sasgen,
GFZPage Nr.ESA CliC workshop, Troms, 20 January 2015Viscoelastic
response function computed for disc load experiment run to steady
state. Disc radius 62 km (geodesic grid) at mass rate of 5.6 Gt/yr.
Earth model varied to account for special conditions in West
Antarctica.
REMOVE.29Sheet1Ocean sectorIce shelvesGLF
SMBCFdh/dtBMBAreaSBMBMRGtyr-1Gtyr-1Gtyr-1Gtyr-1Gtyr-1103
km2myr-1%West Indian OceanAR, NE, AIS,
W*2353049815522-118-14038174-0.800.2247West Indian
Ocean+32431-20429--17943--47East Indian OceanSHA*, VAN, TOT*, MU,
POR*, 3331648721344-5120-2194865-3.350.7351East Indian Ocean+ADE*,
MER, NIN*, COO, REN*50826-30675--30080--50Ross SeaDRY, RIS,
SUL1491671171531000-6726492-0.140.0530Ross
Sea+17516-16715--7928--32Amundsen SeaLAN*, GET*, CD*, THW*, PI*,
COS*38319551119843-15613-3954856-7.110.8767Amundsen
Sea+50524-23250--48457--68Bellingshausen SeaABB*, VEN*, GEO*,
WOR1391182163110-6543-2552286-2.980.2689Bellingshausen
Sea+17412-4113--28123--87Weddell SeaLBC, FRIS, BRL,
JFL33435139233553100-11852608-0.190.0925Weddell
Sea+36335-37133--13153--26Fringing West Antarctica SUL, LAN, GET,
CD, THW, PI, 542231471923254-22145-67853154-4.400.3574Fringing West
Antarctica+COS, ABB, VEN, GEO, WOR70027-27563--79262--74Total
Surveyed 1,57356444361,106141-28250-1,1931631,481-0.810.1152Total
Upscaling 47667-21633--2613474-3.530.4755Total
Antarctica2,04987-1,321144--1,4541741,555-0.940.1152
Sheet2
Sheet3