Sea-Level Change Driven by Recent Cryospheric and Hydrological Mass Flux

Mark TamisieaHarvard-Smithsonian Center for Astrophysics

James DavisEmma HillErik IvinsGlenn Milne

Jerry MitrovicaHans-Peter PlagRui PonteBert Vermeersen

Thanks to:

Extracting Source InformationFrom Geographic Sea Level Variations

• Introduction– Terminology– Physics– Patterns for Greenland, Antarctica and glaciers

• Obtaining Greenland and Antarctic Ice Mass Balance– Select set of tide gauges– Binning of many tide gauges

• Future Directions– Improvements to fingerprints– Focus on near field

• New data types• Geoid better discriminator?

– Integration with ocean modeling• Large oceanic variability• Hydrological example

Introduction Sea Level Variations Due to Loads

References:• Farrell and Clark [1976]• Clark and Primus [1987]• Nakiboglu and Lambeck [1991]• Conrad and Hager [1997]• Mitrovica et al. [2001]• Plag and Jüttner [2001]

Load

Ocean

Possible Loads:• Ice Sheets• Glaciers• Water Stored on the Continents

Assumptions:• Static Ocean Response• Elastic Earth (generally)

Ice sheet melts -- or --

River basin loses water

Load Changes

• More water in ocean• Crust and sea surface adjust to the changing mass load

UniformMelting

Meier, 1984

Melting Scenarios

RSL Fingerprints from Melting Ice

Sheets and Glaciers

Antarctica

Greenland Mountain Glaciers

1.0 corresponds to value of globally-averaged sea level rise.

Obtaining Greenland and Antarctic Ice Mass Balance

ΔRSL (at a given point) = Contributions from Glacial Isostatic Adjustment (GIA)+ Antarctica + Greenland + Glaciers + Steric Effects + Atmospheric Effects + Currents + Hydrology + Tectonics + Sedimentary Loads + …

Adding up the Contributions

Assume large spatial scales and long time scales leave only a few contributions.

First Example: Small Number of Tide Gauges

Mitrovica et al., 2001Tamisiea et al., 2001

Douglas, 1997

Select Set of Tide Gauges

Raw Tide Gauge Data

GIA CorrectedTide Gauge Data

Second Example: Binning of Many Tide Gauges

• Tide gauge data binned• Numerous regression

estimates generated by varying binning resolution, GIA model, and steric model

Results:Antarctic Contribution: 0.4 ± 0.2 mm/yrGreenland Contribution: 0.10 ± 0.05 mm/yrGlobal Average: 1.05 ± 0.75 mm/yr10 to 15% Variance Reduction

Plag, 2006.

Also, see poster by C.-Y. Kuo and C.K. Shum

Future Directions

1. Improvements to fingerprints

2. Focus on near field– New data types– Geoid better discriminator?

3. Integration with ocean modeling– Large oceanic variability– Hydrological example

1. Fingerprint ImprovementsUniform Melting

Mass balance scenario adapted by James and Ivins, 1997 from Jacobs, 1992.

Tamisiea et al., 2001

2. Focus on Near Field

• The impact of different melting scenarios greatest in near field.

• Saltmarsh proxy records with uncertainties of 0.25 mm/yr would still resolve difference in models to the right.

Milne and Long

Glacier model based on Arendt et al., Science, 2002

Alaska – Earth Model Dependence

mm/yr

Effects of Earth Model on Sea Surface and RSL

Tamisiea et al., 2003

3. Integration with Ocean Modeling• Interannual variability large• Incorporate fingerprinting technique into models to

perform integrated analysis

MIT/AER ECCO-GODAE solution

range (0-10 cm)

Altimeter

Source: Ponte et al.

Comparison of Tide Gauge Time Series with Ocean Model

Hill, Ponte, and Davis, 2006A combined time series including

a) Inverted barometer time series [Ponte, 2006]

b) Ocean model time series [courtesy of D. Stammer]

were compared to the time series of 380 globally-distributed PSMSL tide gauges

While removing the model time series significantly reduces the mean global variance, an annual signals remains.

Example time series for stations with high variance reduction(red=tide gauge, blue=model)

[Figure removed]

Example: Annual SignalLaDWorld Hydrology Dataset

• Long time series• Predicted GMSL close to observed

Milly and Shmakin, 2002Milly, Cazenave, and Gennero, 2003

[Figure removed]

Variance Reduction of Tide Gauge Data

• Hydrology model time series removed from residual time series (TG-OM-IB)

• Variance reduced

[Figure removed]

Conclusions

• Fingerprinting offers another method of constraining the sources of sea level rise.

• Large regional effects could provide more effective test of regional mass variation scenarios.

• Inclusion into dynamic ocean models should improve the ability to recover these static signals from the tide gauge and altimetry data.