Freshwater variability and predictability in the Alaska Coastal Current

We constructed the annual cycle of the baroclinic, geostrophic component of the mass and freshwater transports and freshwater content (FWC) in the Gulf of Alaska's Alaska Coastal Current (ACC) from CTD data to assess the processes controlling the freshwater budget of the ACC. In most months the coastal freshwater discharge is balanced by the along-shelf freshwater transport (FWT). Freshwater is trapped to the ACC because offshore eddy freshwater fluxes appear to buffer onshore Ekman transport of low-salinity surface waters due to the cyclonic winds. The annual average FWT is 880 km(3) yr(-1), which compares favorably with the annually averaged runoff of 760 km(3) yr(-1). The barotropic component of the ACC might transport an additional 380 km(3) yr(-1) of freshwater, with the imbalance due to underestimates in runoff, the neglect of freshwater influx from the British Columbian shelf, or the reference salinity (33.8). The mean FWC is 540 kin 3, so that the freshwater flushing time over the 1500 kin portion of the ACC considered is <= 1 yr. Our conclusions are tentative because of uncertainties in discharge estimates, structures of the cross-shelf flow and along-shelf winds, neglect of flow-topography interactions, and the barotropic transport. If our results hold, they imply that the ACC is an important freshwater source for the Bering Sea shelf and Arctic Ocean. Mass transports and FWT in late winter 1998 (El Nino) were nearly twice as large as in winter 1999 (La Nina), and the springtime onset of stratification was earlier in 1998 than in 1999. These differences were due to the winter 1997-98 being subject to anomalously large runoff and strong downwelling over the coastal Northeast Pacific Ocean. The comparison suggests that climate changes leading to warmer, rainier winters will advance the onset of springtime stratification on the inner shelf. November-May anomalies of ACC mass transport, FWT, and FWC are significantly correlated with runoff anomalies and the latter are significantly correlated with anomalies of Ketchikan-Seward atmospheric sea level pressure gradient. These results lead to a 20th century runoff record that permits retrospective examination of the ACC. It appears that the wettest (driest) decade was the 1920s (1900s) during which time ACC transports decreased (increased). While the Pacific Decadal Oscillation is correlated with this "runoff" time series, it explains < 25% of the runoff variance. Coastal measurements of salinity and/or dynamic height are better predictors of these ACC variables than runoff, suggesting that along-shelf monitoring of the ACC might be achieved using simple measurements at only a few coastal stations. (c) 2004 Elsevier Ltd. All rights reserved.