GRACEHydrology Workshop, Irvine, CA

1
GRACE-Hydrology Workshop, Irvine, CA

• Characterizing Signals and Errors in GRACE
• Time-variable Gravity Solutions
• Paul F. Thompson
• Dept. of ASE/EM, University of Texas at Austin
• GRACE team at UT-CSR
• March 22, 2004

2
What does a GRACE gravity solution represent?

• time variable background models
• solid earth tide
• ocean tide
• rotational deformation
• non-tidal atmosphereocean (AOD)

Static geopotential
updates to static and time variable GEO

• Errors that impact estimated GEO during data
  processing
• instrument measurement noise
• parameterization
• aliasing of short-period variability
• spacecraft events
• and so on

• Impact on interpretation due to average unmodeled
  GEO
• continental hydrology
• solid earth tide
• ocean tide
• rotational deformation
• AOD

3
What does a GRACE gravity solution represent?

• Primary product from GRACE is a monthly gravity
  field

• Update found that minimizes residuals of
  (observed-predicted) measurements

4
What does a GRACE gravity solution represent?

• Where, the nominal model contains a static and
  time-variable geopotential

• Update consists of more than unmodeled
  continental hydrology

• Difference between two GRACE solutions for time A
  and time B, assuming Gstatic is unchanged

5
Annual Variability 600 km smoothing radius
GRACE - cosine
GRACE - sine
NCEP hydro - cosine
NCEP hydro - sine
6
2003 Variability, 600 km smoothing radius
GRACE
April
August
October
NCEP hydrology
April
August
October
7
GRACE and NCEP continental hydrology

• Annual cycles comparable in amplitude
• Some correlation between GRACE and hydrology up
  to degree 30 (600 km wavelengths)
• Some months agree very well, some not at
  allhighlights difference in phase between GRACE
  and NCEP hydro
• Deviations from annual are comparable to annual
  signal for some monthsnon-sinusoidal
• Preliminary comparison with other hydrology
  models show better agreement (e.g., GLDAS)

8
GRACE Variability About Average - 2003
March, 2003
April, 2003
May, 2003
July, 2003
9
Background Model Variability
10
Calibrated error estimates
Example 1
Example 2
Example 4
Example 3
11
Short-period aliasing and the AOD model
Aliasing error due to unmodeled atmosphere
Perturbation in time-variable gravity solutions
due to AOD model--August to November, 2002
Thompson, et al., 2000 2003 2004
12
Mass variability compared to GRACE errors
13
Summary

• Analyzed 14 monthly solutions - 4 in 2002, 10
  in 2003
• GRACE gravity solution variability
• Strong annual cycle of 8 mm amplitude
• NCEP hydrology shows reasonable agreement in
  amplitude, but disagreement in phase. Other
  models show better agreement
• Interpretation of time-variable gravity results
• 600 km smoothing radius is reasonable with 2003
  solutions
• Averaging needs to account for definition of
  monthly
• Regional scale variability is evident (e.g., the
  two basins bordered by Amazon river)
• Aliasing/de-aliasing most significant for the
  highest degrees, particularly the sectorials
• Further work required to isolate unmodeled mass
  variability (e.g., hydrology, ocean, atmosphere)
  in the presence of the processing related errors
  (e.g., instrument noise, temporal aliasing)