DATA ASSIMILATION
339
illu
inve
strates an example of such comparison. Differences between different
model residuals are comparable to their respective theoretical expectations in
both overall amplitude and spatial distribution. The absolute magnitudes of
these estimates are statistically consistent, as the model-data difference of
the simulation is also comparable with its theoretical expectation (Figure 2).
The fidelity of the assimilated analyses permits diverse studies and
applications of not only ocean circulation (e.g., Fukumori et al., 2004) but
also of ocean biogeochemical processes (e.g., McKinley, 2002) and geodetic
stigations (e.g., Dickey et al., 2002.) For instance, Figure 6 illustrates
such an application and an assessment of the data assimilated model estimate.
Figure 6. Coherence between observed and modeled excitation of Earth’s wobble (polar
motion); NCEP atmosphere reanalysis (thin black; “no ocean”), ECCO simulation plus NCEP
atmosphere (gray; “simulation”), ECCO assimilation plus NCEP atmosphere (thick black;
ive to the terrestrial frame)
nd that estimated by atmospheric (National Centers of Environmental
“assimilation”). Also shown are the 95% and 99% confidence levels. (Gross, 2003, personal
communication. See Gross et al., 2003, for related results.)
The figure shows coherence between observed excitation of Earth’s polar
motion (the wobble of Earth’s rotation axis relat
a
Prediction (NCEP) Reanalysis, Kalnay et al., 1996) and oceanic models.
While changes in atmospheric circulation (thin black curve) account for
most of the observed polar motion, adding the ocean estimate (gray curve)
significantly improves the coherence at almost all frequencies. Moreover,
the ocean assimilation (thick black curve) further improves the coherence
illustrating the impact of ocean data assimilation in improving the estimate
of ocean circulation. Satellite navigation employs estimates of polar motion
and thus would benefit from forecasts as well as near real-time ocean
analysis systems.
339
illu
inve
strates an example of such comparison. Differences between different
model residuals are comparable to their respective theoretical expectations in
both overall amplitude and spatial distribution. The absolute magnitudes of
these estimates are statistically consistent, as the model-data difference of
the simulation is also comparable with its theoretical expectation (Figure 2).
The fidelity of the assimilated analyses permits diverse studies and
applications of not only ocean circulation (e.g., Fukumori et al., 2004) but
also of ocean biogeochemical processes (e.g., McKinley, 2002) and geodetic
stigations (e.g., Dickey et al., 2002.) For instance, Figure 6 illustrates
such an application and an assessment of the data assimilated model estimate.
Figure 6. Coherence between observed and modeled excitation of Earth’s wobble (polar
motion); NCEP atmosphere reanalysis (thin black; “no ocean”), ECCO simulation plus NCEP
atmosphere (gray; “simulation”), ECCO assimilation plus NCEP atmosphere (thick black;
ive to the terrestrial frame)
nd that estimated by atmospheric (National Centers of Environmental
“assimilation”). Also shown are the 95% and 99% confidence levels. (Gross, 2003, personal
communication. See Gross et al., 2003, for related results.)
The figure shows coherence between observed excitation of Earth’s polar
motion (the wobble of Earth’s rotation axis relat
a
Prediction (NCEP) Reanalysis, Kalnay et al., 1996) and oceanic models.
While changes in atmospheric circulation (thin black curve) account for
most of the observed polar motion, adding the ocean estimate (gray curve)
significantly improves the coherence at almost all frequencies. Moreover,
the ocean assimilation (thick black curve) further improves the coherence
illustrating the impact of ocean data assimilation in improving the estimate
of ocean circulation. Satellite navigation employs estimates of polar motion
and thus would benefit from forecasts as well as near real-time ocean
analysis systems.