Estimating Hydrologic Loads from GRACE

1
Estimating Hydrologic Loads from GRACE

• Ki-Weon Seo and Clark R. Wilson
• Department of Geological Sciences, Jackson School
  of Geosciences, University of Texas at Austin

2
Estimating water content variations over basins

• GRACE stokes coefficients Clm, Slm
• Design spatial filter (Basin function) Bclm,
  Bslm
• (Swenson and Wahr 2002, Swenson et al
  2003, Famigliettie et al 2001)

3
Outline

• Generate simulated GRACE Stokes coefficients
• Design basin functions (4 different ways)
• Test the performance of basin functions
• Summary

4
Simulated GRACE Stokes coefficients

• GLDAS (Global Land Data Assimilation Scheme) and
  ECCO (Estimating the Circulation and Climate of
  the Ocean), 1987-1991
• Measurement error (50x)
• Atmospheric surface pressure error
  (NCEP-GEOS)/sqrt(2)
• (Wahr et al 1998)

5
(No Transcript)
6
4 Basin Functions

• B1 Truncated basin function to (l,m)40

• B2 Smoothed truncated basin function via
  spherical cap convolution

• B3 use load variance and autocorrelation
  (Swenson et al 2003)

• B4 Dynamic basin function varying in time, using
  climate model data (Seo and Wilson 2004)

7
Dynamic basin function
- Bclm,Bslm spherical harmonic coefficients of
B1
Static basin function
- Eclm,Eslm GRACE error
- TclmRMS,TslmRMS RMS spectrum from climate and
oceans models output
Dynamic basin function
- Tclmt,Tslmt monthly mean spectrum from
climate and oceans models output
8
Dynamic basin function

• B4 is determined from known loads (GLDAS ECCO)
• Similar to current plan to de-alias GRACE data
• Hydrologic models are reliable to use monthly
  load changes?

9

• . (l,m)40
• . B2 5 degree spherical cap
• . B3 correlation length 500km and signal
  variance 6113mm2 in decaying exponential model
• . B4 use RMS amplitude spectra for the GLDAS and
  ECCO 1987-1991

10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
RMS mass loads and test basins
14
Water mass load estimation and the associated
errors
15
RMS associated errors
16
Water mass load estimation and the associated
errors
17
RMS associated errors
18
Water mass load estimation and the associated
errors
19
RMS associated errors
20
Water mass load estimation and the associated
errors
21
RMS associated errors
22
Water mass load estimation and the associated
errors
23
RMS associated errors
24
RMS errors from different cost functions (B4)
E Leakage error Measurement error A
Leakage error Atmospheric surface pressure
error EA Leakage error Measurement error
Atmospheric surface pressure error
25
E Leakage error Measurement error A
Leakage error Atmospheric surface pressure
error EA Leakage error Measurement error
Atmospheric surface pressure error
RMS errors from different cost functions (B4)
26
Summary

• Load recovered over large basins (Amazon,
  Mississippi and Lena) with 50 x error
• B3 and B4 provide better estimates than B1 or B2
• Atmospheric surface pressure error spectrum can
  be considered in basin function design
• Aliasing error not yet considered

27
(No Transcript)