Download presentation
Presentation is loading. Please wait.
Published byNorman Stokes Modified over 8 years ago
1
Antarctic ice mass change estimates from GRACE: Results, uncertainties, and the combination with complementary information Martin Horwath, Reinhard Dietrich Philippe Huybrechts, Stefanie Linow SPP1257 Project “ Antarctic Ice Mass Balance“
2
Motivation / Outline Antarctic ice mass change from GRACE [Gt/a] Understand error mechanisms, assess errors Provide new ice mass change estimates (here: from GFZ Release 04) Use of GRACE in a broader methodological context
3
Basics Integrated mass signal Errors GRACE mass estimate Leakage effect (induced by any uncorrected ) GRACE error effect Errors of superimposed signal correction region functon weight function Surface mass density All error effects depend on or, respectively,.
4
Basics Choice of (and, thus, of estimation method) weight function (formally) optimal Lagrange multiplier method by Swenson and Wahr [2002] widening, scaling
5
Leakage effects weight function “from outside“ (oceans, hydrology etc.): correct from models and/or assess uncertainties “from inside“ (Antarctic ice mass variations themselves): assess using different ice variation scenarios
6
Leakage effects uniform layer assumption Note: Scenarios with positive and negative mass changes may induce surprising leakage effects: true mass change [Gt]true total estimated total location 1 with = 1 location 2 with = 0.5 -30 +20 -10 -20 overestimation +30 -40 -10 +10 sign reversal
7
Leakage effects [kg/m²/a] -200 -100 0 100 200 radar altimetry [Legrésy et. al, 2006] stochastic model of interannual accumulation fluctuations autocorrelation of relative fluctuations uniform layer assumption distance [km]
8
Leakage effects radar altimetry [Legrésy et. al, 2006] stochastic model of interannual accumulation fluctuations uniform layer assumption Relative leakage effect
9
GRACE error effects: Empiricial errors vs. propagated calibrated errors [Horwath and Dietrich, GRL, 2006] (GFZ RL04, 300km gaussian smoothed surface mass “empirical error“ STD uncorrelated error model using calibrated errors Quotient “ empiricial “ uncorrelated error model Error STD of surface mass change [mm w.e.] Quotient factor ~ 2 Correlations do matter
10
GRACE error effects: temporal correlations? Example 1: interannual errors? Example 2: error trends? Formal errors of mass trends assume uncorrelated monthly errors. How meeningful are they? -- Also with regard to interannual geophysical variations? 33% of yearly accumulation low-accumu- lation region Mass change [Gt] Time Trends in surface mass [mm w.e. p.a.] (GFZ RL04, 300km gaussian smoothed, GIA model reduced) Significance level time series at 3 points surface mass [mm w.e.]
11
GIA correction Apparent surface mass density trend from IJ05 [Ivins and James, 2005] with mean viscosity [mm w.e. p.a.] Resulting correction for Antarctic ice mass trend: 101 +/- 40 Gt/a. Note: depends not only on GIA model and its uncertainty but also on estimation method.
12
Results (GFZ RL04, 02/2003 - 11/2006) Entire grounded ice sheet Amundsen Sea sector and northwestern Marie Byrd Land Mass change [Gt] Time Mass trend [Gt/a] Time Mass trend [Gt/a] ( 36² + 40² ) 1/2 leakage and GRACE errors GIA uncertainty [average mm w.e.]
13
Results (GFZ RL04, 02/2003 - 11/2006) Time Mass trend [Gt/a]
14
Combination with complementary information GRACE Altimetry GPS SAR +optical imagery Surface mass balance from observations and atmospheric modelling PGR modelling Non-secular variations Geodetic method (GRACE +Altimetry +GPS) Budget method (Input – Output) Flow velocity method (observed – balance vel.) Ice dynamic modelling Determination of the ice mass balance of Antarctica
15
Combination with complementary information flow velocity from SAR interferometry
16
Combination with complementary information 3D thermomechanical ice-dynamic modeling to derive glaciologically realistic ice load history for GIA contribution
17
Conclusions ( -82 +/- 54 ) Gt/a Antarctic ice mass change during 02/2003 – 11/2006 ( +0.23 +/- 0.15 mm/a eustatic sea level) Insights into error mechanisms are the precondition for methodological improvements Mass balance estimates with GRACE (rather than by GRACE)
18
Thank you
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.