GRACE TimeVariable Gravity Geophysical Applications Jianli Chen Ph
GRACE Time-Variable Gravity & Geophysical Applications Jianli Chen, Ph. D. (陈剑利) Center for Space Research, University of Texas at Austin http: //www. csr. utexas. edu/personal/chen/ E-mail: chen@csr. utexas. edu September 2008 - China
个人简历: 1981 -1986 B. S. University of Science & Technology of China 1986 -1989 M. S. Shanghai Astronomical Observatory 1994 -1998 Ph. D. University of Texas at Austin (UT) 1998 -Present UT Center for Space Research
Austin, Texas
University of Texas at Austin One of the largest universities in the US with ~ 50, 000 students.
The Proud Longhorns
2005 National Champion
A Texas Hero - Vince Young
Light Up The Tower - Texas Tradition
GRACE Time-Variable Gravity & Geophysical Applications Part I: About GRACE Part II: Time-Variable Gravity Part III: GRACE Geophysical Applications
GRACE MISSION Science Goals High resolution, mean and time variable gravity field for Earth System Science applications. Mission Systems Instruments • HAIRS (JPL/SSL/APL) • Super. STAR (ONERA) • Star Cameras (�DTU) • GPS Receiver (JPL) Satellite (JPL/Astrium) Launcher (DLR/Eurockot) Operations (DLR/GSOC) Science (CSR/JPL/GFZ) Orbit Launched: March 17, 2002 Initial Altitude: 500 km Inclination: 89 deg Eccentricity: ~0. 001 Separation Distance: ~220 km Nominal Mission : 5 (extended to 8) years
Gravity Recovery and Climate Experiment Amazing GRACE!
Progress in Gravity Field Resolution Decades of tracking to geodetic satellites 13 months of GRACE data 111 days of GRACE data
The Beauty of Earth Science (Courtesy of GFZ GRACE Team) Based on the gravity model EIGEN-GL 04 C, a combination of GRACE and LAGEOS, plus gravimetry and altimetry surface data, completed to degree and order 360.
GRACE Main Products q Time-variable gravity field solutions at approximately monthly intervals. q Static mean gravity fields (e. g. , GGM 01 C, GGM 02 C, EIGEN-GL 04 C …). q In forms of fully normalized spherical harmonics (or Stokes coefficients) up to degree and order 120 (for time-variable fields). q From three processing centers, CSR, GFZ, and JPL. q Supporting data products, GAC, GAB, GAA, and etc.
GRACE Time-Variable Gravity & Geophysical Applications Part I: About GRACE Part II: Time-Variable Gravity Part III: GRACE Geophysical Applications
The Dynamical Earth System Atmosphere Solid Earth Hydrosphere Core Mantle Cryosphere Ocean
The Global Geophysical Fluids Center (GGFC) As a component of the International Earth Rotation and Reference Systems Service (IERS), GGFC and its eight Special Bureaus have been providing services to the Earth sciences community since 1998. CSR GGFC Homepage - http: //www. ecgs. lu/ggfc/
The Global Water Cycle
The Earth’s Time-Variable Gravity The geopotential field is determined by mass distribution within the Earth’s system, and is conveniently expressed in a spherical harmonic expansion as, where, r, , G M e , Re Clm , Slm are the geocentric distance, latitude, and longitude; is the gravitational constant; are the mass and equatorial radius of the Earth; are the spherical harmonics of degree l and order m; Plm (sin ) are the Legendre polynomials of degree l and order m.
The Earth’s Time-Variable Gravity Clm & Slm are computed from mass changes as Monthly Clm & Slm variations, or so called Stokes Coefficients are provided by GRACE, which can be used to infer surface mass variation or geoid height change.
Three Representations: q Spherical Harmonics Clm & Slm q Surface mass load change ∆ q Geoid height change ∆N Wl = Wl ( r ) is the Gaussian weighting as a function of spatial radius, r.
Surface Mass Changes from GRACE Main issues to solve: q Spatial smoothing or filtering is needed as high degree and order terms are dominated by noise. q Appropriate treatments of low degree harmonics (e. g. , the large uncertainty of C 20 in RL 01 & RL 04 solutions, and the missing geocenter, degree-1 terms in GRACE data). q Residual aliasing effects (from ocean and air tide model errors). q How to minimize leakage effects due to the spatial smoothing? q How to validate GRACE estimates? q What is the effective spatial resolution of GRACE?
The Challenge of Spatial Smoothing
The Challenge of Low Degree Terms
Spatial Filterings of GRACE Data q Gaussian filtering (Jekeli 1981, Wahr et al. 1998) q Non-isotropic Gaussian filtering (Han et al. 2005) q Optimized filtering (signal-to-noise ratio) (Chen et al. 2006) q Decorrelation + Gaussian filtering (Swenson and Wahr 2006) q Wiener optimal filtering (Sasgen et al. 2006) q Statistical filtering (Davis et al. 2008) q Wavelets & EOFs q…
GRACE Time-Variable Gravity & Geophysical Applications Part I: About GRACE Part II: Time-Variable Gravity Part III: GRACE Geophysical Applications
GRACE Geophysical Applications (1) q Hydrology q Land water storage q Ground water q Evapotranspiration q Runoff q Oceanography q Non-steric sea level change q Ocean bottom pressure q Heat content q Ocean tides q Glaciology q Polar ice sheets mass balance q Mountain glaciers mass balance
GRACE Geophysical Applications (2) q Solid Earth Geophysics q Post glacial rebound (PGR) q Coseismic & postseismic deformation q Volcano activities q Atmosphere q Atmospheric pressure q Air tides q Climate change q Floodings q Droughts q Sea level rise q Earth Rotation q…
The Global Water Cycle Hydrological Parameters P q Precipitation (P) q Evapotranspiration (E) q Runoff (R) q Water storage change ( S) q Basic Conservation Equation S = P - E - R S E R
Hydrological Interests q Terrestrial water storage change ( S) - soil moisture & ground water q q Snow water equivalent ( SWE) q q SWE = P - E - R - S Evapotranspiration (E) q q S=P-E-R P E E = P - R - S Runoff (R) q R = P - E - S S R
World Major River Basins Mississippi Amazon Bay of Bengal OB
Example 1 - Global Water Storage Change from GRACE & Model q GRACE time-variable gravity data q CSR constrained RL 01 solutions up to degree 120 q 35 monthly solutions covering the period April 2002 - June 2005 q C 20 excluded, truncation at degree 60 q 800 km Gaussian smoothing q q Global land data assimilation system (GLDAS) q GLDAS estimated TWS changes q Converted into spherical harmonics of degree 100 q Drop C 20 and geocenter terms (to be consistent with GRACE) q 800 km Gaussian smoothing & truncation at degree 60 Comparisons q Global scale q Basin scale
Global Water Storage Change from GRACE
GRACE/GLDAS Comparison in Amazon
GRACE/GLDAS Comparison in Mississippi
GRACE/GLDAS Comparison in Ganges
GRACE/GLDAS Comparison in Zambezi
Example 2 - River Runoff in the Amazon Basin Estimated from GRACE & Models q Basic equation: R = (P - E) - S q S from GRACE time-variable gravity q q 22 CSR RL 01 unconstrained solutions April 2002 - July 2004 q C 20 excluded q 1000 km Gaussian smoothing P-E from ECMWF model estimates q q Based on land-atmosphere water mass balance Comparison with gauge measurements (at Obidos, Brazil)
Land-Atmosphere Water Mass Balance q For land water mass balance div. Q W q For atmosphere water mass balance P E S R q For combined land-atmosphere water mass balance W - Vertically-integrated precipitable water ( ECMWF) Q - Divergence of the vertically-integrated average atmospheric moisture flux vector (ECMWF)) S - Water storage (GRACE)) Details in Syed, T. H. , J. S. Famiglietti, J. L. Chen, M. Rodell, S. I. Seneviratne, P. Viterbo, C. R. Wilson, Amazon and Mississippi River Discharge Estimated from GRACE and a Land-Atmosphere Water Balance, Geophys. Res. Lett. , 32, L 24404, doi: 10. 1029/2005 GL 024851, 2005.
Runoff Estimates in the Amazon River Basin
Runoff Estimates in the Amazon River Basin River Gauge Observations GRACE without leakage correction GRACE with leakage correction
Example 3 - Antarctic Ice Sheet Long-Term Mass Balance From GRACE The Antarctic ice sheet has a total area of ~ 14, 000 km 2 and averaged ice sheet thickness of ~ 2. 16 km, accounts for 90% of the world’s ice and 75% of the world’s fresh water resources, and has the potential to raise the global sea level by over 70 meters if completely melt.
GRACE Data Processing q q GRACE time-variable gravity data q CSR RL 04 solutions up to degree 60 q 43 monthly solutions covering the period Jan. 2003 - Sept. 2006 q P 4 M 6 decorrelation + 300 km Gaussian smoothing Forward Modeling q A numerical simulation technique to more effectively quantity leakage effects from smoothing. q Successfully applied in a number of recent studies. q The main purpose is to determine what original mass change signals could generate the changes observed by GRACE. q Postglacial rebound (PGR) effects (IJ 05 PGR model) q Comparison with remote sensing measurements
Global long-term mass change rates from GRACE Based on 58 CSR RL 04 monthly gravity solutions from April 2002 to May 2007.
GRACE-observed mass change rates (in cm/year of water height) over Antarctica, P 4 M 6 + 300 km Gaussian smoothing. PGR effects (in mass equivalent) over Antarctica from the IJ 05 PGR model, P 4 M 6 + 300 km Gaussian smoothing. (Please note the different color scales, ± 8 vs. ± 4. )
Antarctic long-term ice mass change rates (GRACE - PGR) Enderby Land (EL) Antarctic Peninsula (AP) Amundsen Sea Embayment (ASE)
Forward Modeling q Step-1: To choose six regions (shaded) where GRACE data show prominent signals. In each region (on 1 x 1 grid), a trial mass rate (in units of km 3/yr) is distributed uniformly. The remainder of the grid (outside Antarctica) retains GRACE mass rates. Therefore, spatial leakage from the 6 regions to areas outside Antarctica should be evident when comparing the model and GRACE rate maps. This is a variation of the forward modeling technique in our previous studies. q Step-2: To convert the constructed mass rate grids into spherical harmonics. q Step-3: To replicate procedures (used to transform GRACE data) to compute surface mass changes (no degree-1 terms, same truncation and same 2 -steps filtering, …). q Step-4: To adjust model rates and region shapes until there is general agreement with the GRACE map. As a final constraint, we force model integrated mass rate for each region (sum over grid points with cosine of latitude weights within boundaries where magnitude exceeds 1 cm/year) to agree with the GRACE map.
GRACE Observations (GRACE-PGR) Modeling Scheme Forward Modeling Estimates
Time series at point A Time series at point B Time series at point E
GRACE time series at point C GRACE time series at point D D
Comparison Between GRACE RL 01 & RL 04 Solutions RL 01: GRACE - PGR RL 04: GRACE - PGR (2002. 04 - 2005. 11) (2003. 01 - 2006. 09) (800 km Gaussian Filtering) (Decorrelation + 300 km Gaussian) Improvement of spatial resolution of GRACE RL 04 solutions is evident.
Example 4 - Coseismic & Postseismic Deformation From GRACE q The Sumatra-Andaman earthquake (Mw=9. 1) on December 26, 2004, the 2 nd largest in modern history. q Special processing of GRACE range rate data (Han et al. 2006) revealed co-seismic gravity changes. q Examine GRACE RL 04 spherical harmonic products for evidence of coseismic change. q RL 04 time series may be examined for postseismic deformation.
GRACE Data and Processing q q q CSR GRACE RL 04 solutions q April 2002 through February 2007; q 55 monthly solutions, up to degree & order 60 Applied Filters q De-correlation filtering [Swenson and Wahr, 2006] q 300 km Gaussian smoothing Coseismic Change q [mean of (2005 + 2006)] minus [mean of (2003 + 2004)] ; q 21 solutions in each mean; q (2003 + 2004): Jan. 2003 - Nov. 2004; q (2005 + 2006): Jan. 2005 - Sept. 2006; ( Dec. 2004 excluded); q Postseismic Change q Use entire 55 point time series; q Examine pre-seismic and post-seismic periods
GRACE-observed mass change: [mean of 2005+2006] - [mean of 2003+2004] From 4 different filtering schemes (in cm of equivalent water thickness change).
Coseismic Deformation Estimate (cm water equivalent mass change) Epicenters of Sumatra-Andaman (Mw=9. 1) and Nias (Mw=8. 7) earthquakes are marked by pink and white triangles. GRACE time-series at point A & B. B
Preseismic Mass Rate Map Postseismic Mass Rate Map
Mass Rate Time Series at Points A-F At points in uplift zone.
Postseismic Mass Rates at Points G-L At points in subduction zone.
Conclusions (co- & postseismic deformation) q GRACE RL 04 gravity data are of sufficient spatial resolution to capture mass redistribution of Sumatra-Andaman earthquake. q After appropriate filtering, RL 04 data clearly delineate effects of the rupture, extending over 1800 km, consistent with previous GRACE estimates, deformation model estimates and other geodetic measurements. q RL 04 time series show steady mass increase, presumably related to post-seismic uplift, equivalent to ~ 8 cm/year of water mass in a broad region northwest of the epicentered on the Andaman-Sunda fault zone. q Time series of gravity change (mass rates) provide a new coseismic/postseismic measure of major earthquakes. q Mass rate estimates (for tectonic, seismic and other non-climate studies) may be contaminated by interannual variability in regions with large seasonal signals
GRACE Other Applications - Sea Level Change Global Non-Steric Sea Level Change From GRACE (See Chen et al. 2005 (JOG, Vol. 79, No. 9, 532 -539) for details. )
GRACE Other Applications - Mountain Glaciers Patagonia Ice Fields Melting From GRACE
GRACE Observed Mass Rates in the PIF Region +A GRACE estimated PIF ice melting rate based on forward modeling is 25. 0 ± 9. 2 km 3/year first successful ‘direct’ measurement of PIF mass balance of recent years [Chen et al. 2007, GRL].
GRACE Time-Variable Gravity & Geophysical Applications Part I: About GRACE Part II: Time-Variable Gravity Part III: GRACE Geophysical Applications
Amazing GRACE
Thank you ! Questions: chen@csr. utexas. edu
- Slides: 66