strong Kamchatka Current to the west; and (d) a relatively weak (5–15 cm/s)
cyclonic circulation occupying the deep part of the Bering Sea (Figs. 2a and 3a).
According to Fig. 3a, a significant portion of this cyclonic gyre originates in the
Near Strait with the rest coming from the inflow through other Aleutian passages.
The circulation shown in Fig. 3a is in good qualitative agreement with the results of
Stabeno et al. [29] who describes a gradual leakage of the Alaskan Stream though
the passages in the Aleutian Arc.
The mean relative mismatch between the reconstructed surface velocities and
assimilated drifter velocities (Fig. 1a) is 8.2 cm/s. It is unlikely that a better
agreement could be obtained between the climatological velocity with mean
amplitude of approximately 10 cm/s and the highly variable surface currents
derived from the drifter trajectories affected by eddies and small-scale variations of
the wind stress.
The derived MDOT were validated using the standard AVISO methodology:
three regions well-covered by drifters were selected, gridded AVISO SSH
anomalies from www.aviso.oceanobs.com were added to each MDOT, and the
resulting geostrophic currents were compared with the currents deduced from
drifter trajectories for the three regions shown by black rectangles in Fig. 3. The
validation has shown that 4Dvar MDOT significantly (10–35%) outperforms all
other MDOT products [22].
Obtaining a high resolution regional MDOT and its posterior errors is a
prominent feature delivered by the 4dVar analysis of the climatological observations. In the next sections, we show how the derived MDOT benefits the analyses of
decadal variability and hindcasts of the BS circulation at interannual time scales.
4dVar Analysis of the Circulation in the Aleutian Passes
This section provides another illustration that a realistic MDOT is a key component
of a regional data assimilation system. In application to the regions with complicated topography, such as Aleutian Passes, MDOT inaccuracies may result in
significant underestimation of the volume transports. To illustrate this, the 4dVar
DA system was configured at a 6 km resolution with a domain around the Amukta
Pass shown by the smaller white rectangle in Fig. 3a. A larger rectangle shows the
domain of a similar experiment conducted in the central part of the Eastern Bering
Sea shelf (see [23]).
Amukta Pass is one of the most important pathways for the inflow of the Pacific
Water into the Bering Sea. Two assimilation experiments have been conducted in
the domain: In the first experiment, the circulation for the period between January
10 and February 5, 2002 was reconstructed by assimilating AVISO sea level
anomalies referenced to the 4dVar MDOT (Fig. 3a) in combination with the NCEP
sea surface heat, salt and momentum fluxes. Climatological (winter) temperature/
salinity in the Amukta Pass distributions were utilized as a first guess. The second
The Bering Sea Regional Data Assimilation System …
507
cyclonic circulation occupying the deep part of the Bering Sea (Figs. 2a and 3a).
According to Fig. 3a, a significant portion of this cyclonic gyre originates in the
Near Strait with the rest coming from the inflow through other Aleutian passages.
The circulation shown in Fig. 3a is in good qualitative agreement with the results of
Stabeno et al. [29] who describes a gradual leakage of the Alaskan Stream though
the passages in the Aleutian Arc.
The mean relative mismatch between the reconstructed surface velocities and
assimilated drifter velocities (Fig. 1a) is 8.2 cm/s. It is unlikely that a better
agreement could be obtained between the climatological velocity with mean
amplitude of approximately 10 cm/s and the highly variable surface currents
derived from the drifter trajectories affected by eddies and small-scale variations of
the wind stress.
The derived MDOT were validated using the standard AVISO methodology:
three regions well-covered by drifters were selected, gridded AVISO SSH
anomalies from www.aviso.oceanobs.com were added to each MDOT, and the
resulting geostrophic currents were compared with the currents deduced from
drifter trajectories for the three regions shown by black rectangles in Fig. 3. The
validation has shown that 4Dvar MDOT significantly (10–35%) outperforms all
other MDOT products [22].
Obtaining a high resolution regional MDOT and its posterior errors is a
prominent feature delivered by the 4dVar analysis of the climatological observations. In the next sections, we show how the derived MDOT benefits the analyses of
decadal variability and hindcasts of the BS circulation at interannual time scales.
4dVar Analysis of the Circulation in the Aleutian Passes
This section provides another illustration that a realistic MDOT is a key component
of a regional data assimilation system. In application to the regions with complicated topography, such as Aleutian Passes, MDOT inaccuracies may result in
significant underestimation of the volume transports. To illustrate this, the 4dVar
DA system was configured at a 6 km resolution with a domain around the Amukta
Pass shown by the smaller white rectangle in Fig. 3a. A larger rectangle shows the
domain of a similar experiment conducted in the central part of the Eastern Bering
Sea shelf (see [23]).
Amukta Pass is one of the most important pathways for the inflow of the Pacific
Water into the Bering Sea. Two assimilation experiments have been conducted in
the domain: In the first experiment, the circulation for the period between January
10 and February 5, 2002 was reconstructed by assimilating AVISO sea level
anomalies referenced to the 4dVar MDOT (Fig. 3a) in combination with the NCEP
sea surface heat, salt and momentum fluxes. Climatological (winter) temperature/
salinity in the Amukta Pass distributions were utilized as a first guess. The second
The Bering Sea Regional Data Assimilation System …
507