Seasonal, Interannual, and Decadal Scale Freshwater Variability

1
Seasonal, Interannual, and Decadal Scale
Freshwater Variability in the Alaska Coastal
Current Thomas J. Weingartner 1, Seth L.
Danielson 1, and Thomas C. Royer 2 1 Institute of
Marine Science, University of Alaska, Fairbanks,
AK 99709 2 Center for Coastal Physical
Oceanography, Old Dominion University, Norfolk, VA
Appendix Calculation details
Towards Predictability Monthly anomalies of Mbc,
FWT, and FWC are predictable from easily obtained
variables!
The Ekman Freshwater Contribution
I Return flow distributed throughout the water
column II Return flow in bottom
boundary layer (30 m thick) Subsurface
salinities in I and II are within ACC but
offshore of the front (25 km lt y lt 45 km).
Surface Ekman layer salinities are averaged from
stations gt 75 km offshore.
Southeast Alaska
Normalized Anomalies

• For example, multiple linear regression models
  using anomalies of
• GAK1 Salinity _at_ 10m (current month)
• GAK1 Dynamic Height, 0-250db (current
  month)
• Discharge, (averaged over 4 preceding
  months)
• Middleton Is. NE-SW wind component (averaged
  over preceding 8 months)

x 1500 km
DE 20m Ekman depth Onshore Ekman transport
DE
ACC Front
Fig. 1
II
C.Fairfield 180m
I (interior outflow) OR II (30m bottom
boundary layer) Offshore Ekman transport
Predicted
Mbc

-H
N 31
0 lt y lt 40 km (Ly )
I
Observed
Fig. 8
Fig. 6
Other processes
explain a significant amount of the variance of
each variable for the Nov. May period. Fig. 8
shows the results for Mbc (63 of variance
explained). Similar regressions explain 75 of
the variance for FWT and FWC. Discharge and/or
GAK 1 data alone explains 50 or more of the
variance, indicating that Royers discharge
record (1930 present) could be used to hindcast
ACC transport and FWC variability. Fig. 9 shows
that Royers discharge anomaly time series is
significantly correlated with the sea level
pressure difference (DSLP) between Ketchikan and
Seward (evaluated using from Trenberth and
Paolinos 1980 SLP record) at periods gt 3
years. The correlation is physically plausible
because the sign of the sea level gradient
implies northeastward advection of moist air
masses into the Gulf of Alaska. (It is also
consistent with the regression that shows that it
is the NE-SW wind component at Middleton Island
that contributes to predictability of the ACC
variables cf. Fig. 8).
Fig. 6 shows the monthly gain (or loss) of
freshwater due to the cross-shore transport
established by wind-driven Ekman dynamics. There
are large differences in magnitude (and even
sign) between the two methods, which reflects the
uncertainty in our understanding of the
cross-shore circulation field (see Appendix).
However, both methods indicate a distinct
seasonal cycle associated with the winds (Fig. 2)
and the subsurface changes in salinity brought
about by vertical mixing and downwelling in the
ACC (Fig. 4). Both methods suggest that this
component is significant to the freshwater
balance of the ACC except in summer when it is
small relative to the discharge. The cross-shore
freshwater decreases through winter, is small in
summer, and increases in fall and is consistent
with the seasonal variations in Mbc, FWT, and
FWC. Because Ekman-induced cross-shore flow is
so feeble, models are needed to improve our
understanding of the vertical structure of this
component of the circulation field.

Fig. 4
The Alaska Coastal Current (ACC Fig. 1) is a
year-round, wind- and buoyancy-forced current
circumscribing the Gulf of Alaska Coast between
British Columbia and the Bering Sea. Low
pressure systems (storms) supply the seasonally
varying downwelling-favorable wind and
precipitation/runoff (Fig.2) that force the ACC.
Herein we consider seasonal and longer-term
variations in the freshwater transport and
content of the ACC and suggest ways by which this
variability might be efficiently monitored.
Fig. 4 consists of representative seasonal
salinity sections from the Cape Fairfield Line.
This transect extends 50 km offshore with
stations spaced 4km apart ( lt the Rossby radius
of deformation). In winter (Feb. Apr.), the ACC
is narrow (10 km), with a bottom-advected plume
structure and weak vertical and cross-shore
salinity gradients. In summer (Jun. Sept.),
the ACC is broad (35 km), with a
surface-advected plume structure (plume depth 30
m) and strong vertical and cross-shore salinity
gradients. In fall (Oct.- Dec.), the ACC is
adjusting toward winter conditions. It has an
intermediate type plume structure, with moderate
vertical and strong cross-shore salinity
gradients.
Fig. 2
Table 1. Cross-shore Eddy Freshwater Flux (see
Appendix for details)
ABOVE AVERAGE ACC TRANSPORT
Month Plume Depth (zp m) U (m s-1) DSf Eddy Flux
April 100 -.15 0.7 -4500
August 30 -.15 2 -4000
October 50 -.20 2.5 -11,000
December 75 -.30 1.8 -18,000
BELOW AVERAGE ACC TRANSPORT
r .71 3-yr lowpass filter
Fig. 9
From Fig. 9 we conclude that the early 1900s were
anomalously dry compared to the succeeding 100
years and that ACC transport and freshwater
content were also anomalously small. Although not
shown here, we find relatively weak coherence
between the Pacific Decadal Oscillation and
freshwater runoff anomalies.
We use data from the Cape Fairfield Line (Fig. 3)
collected in 55 different months between 1980 and
2001 to compute monthly estimates of alongshore
baroclinic transport (Mbc), freshwater transport
(FWT), freshwater content (FWC), and the
wind-induced (Ekman) net cross-shore freshwater
transport (FWE). We also use Gulf of Alaska
discharge data (following Royer, 1982) and
temperature and salinity data at station GAK 1
(Fig. 3) as predictors of ACC FWC, FWT, and Mbc
along the Cape Fairfield transect. GAK 1 was
always occupied coincident with Cape Fairfield
sampling.
The eddy freshwater fluxes are offshore
throughout the year and vary seasonally in
magnitude with maximum eddy flux in fall when the
ACC plume is thick and the front strong. Eddy
fluxes are comparable in size, but opposite in
sign to FWE, suggesting that these processes may
buffer one another!! If, in fact, the difference
between these two terms is small on average, how,
then is freshwater carried offshore?? One
possibility, is through ageostrophic deflection
of the ACC as it encounters changes in the
coastline. This process is suggested in Fig. 7,
which shows offshore deflection of the ACC in
Nov. 2001 by Kayak Island. The importance of
such processes to the freshwater budget is
unknown, but should depend upon the orientation
of the coastal object and the bathymetry and
varies seasonally with ACC velocity and
stratification.
References Royer, T. C., Coastal freshwater
discharge in the Northeast Pacific, J. Geophys.
Res., 87, 2017-2021, 1982. Spall, M. A. and D. C.
Chapman, On the efficiency of baroclinic eddy
heat transport across narrow fronts, J. Phys.
Oceanogr., 28, 2275 2287, 1998. Stabeno, P.J.,
R. K. Reed, and J. D. Schumacher, The Alaska
Coastal Current continuity of transport and
forcing, J. Geophys. Res., 100, 2477-2485,
1995. Trenberth, K. E. and D.A. Paolino, Jr.
1980. The Northern Hemisphere Sea Level Pressure
data set Trends, errors, and discontinuities.
Mon. Weather Rev. 108 855-872.
Conclusions 1. The seasonally-varying baroclinic
component of ACC freshwater transport (FWT) is
in-phase with the coastal discharge, suggesting
that alongshore advection balances the coastal
buoyancy flux in most months. 2. The cross-shore
flux of freshwater due to Ekman transport and
eddies (via baroclinic instability) varies with
seasonal changes in wind stress, discharge, and
ACC frontal structure. These terms have similar
magnitudes throughout the year but opposite
signs, suggesting that they tend to balance. The
estimated fluxes are extremely uncertain, but
improved estimates will likely be obtained from
realistic numerical models of the Gulf of Alaska
shelf. Models can also offer guidance on the
importance to the freshwater budget of ACC
deflection by coastal features. 3. Monthly
anomalies of the baroclinic components of
alongshore mass and freshwater transport and
freshwater content of the ACC can be efficiently
monitored from a single hydrographic station (GAK
1) and/or Royers discharge model. 4. The
barotropic component of the alongshore freshwater
transport is likely substantial and cannot be
neglected in monitoring. Stabeno et al. (1995)
find that the total transport is coherent with
the alongshore wind. We hypothesize that this
component could be monitored efficiently either
from the wind field or with Seward sea-level
(after accounting for the thermosteric effect)
after careful calibration with direct current and
salinity measurements along the Cape Fairfield
Line.
Fig. 5
Fig. 5 shows the mean monthly values and ranges
of Mbc , FWT, and FWC. The annual cycle of Mbc
and FWT are in-phase and both are westward
throughout the year. Mbc varies from a maximum of
-0.1 Sv in August to -0.6 Sv in December. The
magnitude of FWT is only 5 - 8 of Mbc. Maximum
mass and freshwater transports lag the annual
discharge maximum by 2 months (Fig. 2). The FWC
annual cycle differs from the annual transport
and discharge cycles. FWC is a minimum in
May-June, increases rapidly through August, and
maintains steady values through fall. The
tendency for FWC to remain relatively constant in
late winter and fall while FWT changes suggests
that alongshore advection buffers the changing
buoyancy flux associated with runoff and the
onshore Ekman transport of freshwater (Fig. 6).
On annual average FWT is 25000 m3 s-1 although
our calculation omits the seasonally varying
barotropic contribution to the freshwater
transport, which is unknown. Stabeno et al.
(1995) suggest a mean annual barotropic speed of
.03 m s-1, which is 0.25 Sv when integrated
over 200 m depth and the 40 km width of the ACC.
Assuming that 6 of this barotropic transport
consists of freshwater, then the barotropic
component of FWT is 15000 m3 s-1 on annual
average.
Fig. 3
Fig. 7
Kayak Is. ACC
Acknowledgements The authors are grateful for
the support provided by the Exxon Valdez Oil
Spill Trustees Council, the National Oceanic and
Atmospheric Administration, and the National
Science Foundation.
MODIS, Nov. 7, 2001