420
ERIC P. CHASSIGNET ET AL.
remainder of this section, we outline some of these metrics and provide
examples for the HYCOM Atlantic forecasting system.
Large-scale circulation features: These tests evaluate whether the global
and basin-scale models correctly place the large-scale features of ocean
circulation, such as gyres, strong fronts, and currents. It is indeed necessary
to know the oceanic mean SSH over the time period of the altimeter
observations before one can assimilate the SSH anomalies determined from
satellite altimeter data. Furthermore, at the scales of interest (tens of
kilometres), it is also necessary to have the mean of major ocean currents
and associated SSH fronts sharply defined. This is not feasible from coarse
hydrographic climatologies (~1º horizontal resolution) and from present
geoid measurements since the geoid is not yet known accurately on the
mesoscale. The approach taken by the HYCOM-based system is to use a
model mean generated by a previous 1/12º North Atlantic simulation
performed with MICOM (Chassignet and Garraffo, 2001).
Eddy kinetic energy/SSH variability: These tests evaluate whether the
models have a realistic level and distribution of energy (mean and
variability) at depths where observations are available.
Sea Surface Height (analysis, forecast): Provide an assessment of the
models’ ability to represent observed sea surface heights.
Sea Surface Temperature (analysis, forecast): These tests evaluate
whether the models are producing acceptable nowcasts and forecasts of sea
surface temperature. The near real-time system is routinely compared to
buoy observations of SST.
Vertical profiles, time series of profiles and vertical cross sections
(analysis, forecast): Since the present forecasting system assimilates only
surface quantities (SSH, SST), quantitative comparisons of model
temperature and salinity to unassimilated profile data from XBTs, CTDs,
and ARGO floats, and moored buoys can be used to assess the model’s
performance in the ocean interior. In Figure 2, model temperature sections
are compared to XBT measurements obtained from the Marine
Environmental Data Service (MEDS) dataset. A quantitative assessment
using the RMS difference between the model and data profiles is shown in
Figure 3. With assimilation of surface data only, the Atlantic HYCOM
forecasting system has, overall, larger RMS error than climatology or
MODAS-3D. MODAS-3D (Fox et al., 2002) uses the statistics of the
historical hydrographic data base to downward project the same MODAS
SSH anomaly and SST analyses assimilated by HYCOM, indicating superior
performance for a data-based method of downward projection than the
Cooper and Haines (1996) technique used in HYCOM, at least in this
application.
ERIC P. CHASSIGNET ET AL.
remainder of this section, we outline some of these metrics and provide
examples for the HYCOM Atlantic forecasting system.
Large-scale circulation features: These tests evaluate whether the global
and basin-scale models correctly place the large-scale features of ocean
circulation, such as gyres, strong fronts, and currents. It is indeed necessary
to know the oceanic mean SSH over the time period of the altimeter
observations before one can assimilate the SSH anomalies determined from
satellite altimeter data. Furthermore, at the scales of interest (tens of
kilometres), it is also necessary to have the mean of major ocean currents
and associated SSH fronts sharply defined. This is not feasible from coarse
hydrographic climatologies (~1º horizontal resolution) and from present
geoid measurements since the geoid is not yet known accurately on the
mesoscale. The approach taken by the HYCOM-based system is to use a
model mean generated by a previous 1/12º North Atlantic simulation
performed with MICOM (Chassignet and Garraffo, 2001).
Eddy kinetic energy/SSH variability: These tests evaluate whether the
models have a realistic level and distribution of energy (mean and
variability) at depths where observations are available.
Sea Surface Height (analysis, forecast): Provide an assessment of the
models’ ability to represent observed sea surface heights.
Sea Surface Temperature (analysis, forecast): These tests evaluate
whether the models are producing acceptable nowcasts and forecasts of sea
surface temperature. The near real-time system is routinely compared to
buoy observations of SST.
Vertical profiles, time series of profiles and vertical cross sections
(analysis, forecast): Since the present forecasting system assimilates only
surface quantities (SSH, SST), quantitative comparisons of model
temperature and salinity to unassimilated profile data from XBTs, CTDs,
and ARGO floats, and moored buoys can be used to assess the model’s
performance in the ocean interior. In Figure 2, model temperature sections
are compared to XBT measurements obtained from the Marine
Environmental Data Service (MEDS) dataset. A quantitative assessment
using the RMS difference between the model and data profiles is shown in
Figure 3. With assimilation of surface data only, the Atlantic HYCOM
forecasting system has, overall, larger RMS error than climatology or
MODAS-3D. MODAS-3D (Fox et al., 2002) uses the statistics of the
historical hydrographic data base to downward project the same MODAS
SSH anomaly and SST analyses assimilated by HYCOM, indicating superior
performance for a data-based method of downward projection than the
Cooper and Haines (1996) technique used in HYCOM, at least in this
application.