Read the presentation of Earth Interactions (published September, 1996 in the Bulletin of the American Meteorological Society) for more information about the journal.
This article has been prepared for, and submitted to Earth
Interactions, an electronic journal published jointly by the
American Geophysical Union, the
Association of American Geographers,
and the American Meteorological
Society; with support from NASA's
Mission to
Planet Earth and
Goddard
DAAC. This should thus be treated as a reprint, and
this article may not be
distributed, or posted on the Internet, without authorization from Earth
Interactions.
|
|
Read the presentation of Earth Interactions (published September, 1996 in the Bulletin of the American Meteorological Society) for more information about the journal. |
The proper citation of this article is:
Melsom, A., S.D. Meyers, H.E. Hurlburt, E.J. Metzger and J.J. O'Brien, 1999:
ENSO Effects on Gulf of Alaska Eddies, Earth Interactions,
3. [Available online at http://EarthInteractions.org]
| Viewing figures at full resolution |
Corresponding author: Arne Melsom
E-mail: meyers@stommel.marine.usf.edu
Abstract
Generation and propagation of eddies in the coastal regions of the eastern Gulf of Alaska are examined based on ouput from a numerical ocean model. Results from a 1/8° six layer isopycnal, wind forced Pacific basin model are examined within the Gulf of Alaska during the 14 year period starting in January 1981.
Interannual variability in the upper ocean coastal circulation in the Gulf of Alaska is linked to the El Niño/Southern Oscillation phenomenon in the tropical Pacific Ocean, via coastal Kelvin waves and atmospheric teleconnections. El Niño events destabilize the Alaska Current by enhancement of the velocity shear in the vertical. The instability ultimately results in the formation of multiple strong anticyclonic eddies along the coast which slowly propagate into the Gulf of Alaska where they can survive for more than one year. A typical value for the diameters of the anticyclonic eddies is 200 km in the data and in the model. These eddies are strongly baroclinic, with a typical value for the velocity differences between layers 1 and 2 of 15 cm/s. El Viejo (La Niña) events generally reduce eddy formation.
The continental slope of the northeast Pacific is a major habitat for a number of commercially harvested fish (e.g., salmon, halibut and herring) (Cummins and Mysak, 1988). It has been suggested that year-to-year changes in fish stock recruitment and return migration routes of certain species in the Gulf of Alaska (GOA) are affected by low frequency ocean variability such as the El Niño/Southern Oscillation (ENSO) (Emery and Hamilton, 1985; Mysak, 1985). ENSO affects the extratropical oceans in two ways: locally, through changes in the surface winds, and, remotely, via tropically generated, coastally trapped Kelvin waves. Although the effect on the ocean circulation by surface winds is local, the surface wind may in itself be remotely forced by atmospheric wave motions.
Kelvin waves associated with El Niño warm events suppress upwelling of denser and nutrient-rich water masses, and those associated with El Viejo (La Niña) cold events enhance such upwelling. Both alter the biogeochemical cycle of the coastal northeast Pacific Ocean. These large changes should induce changes in the biological productivity. ENSO has been tied to changes in zooplankton biomass (McGowan, 1985; Smith, 1985) in the California Current system and interannual changes in coastal upwelling have been linked to variations in the population and health of several species of fish (Ware and Thomson, 1991). A similar relationship might be expected in the GOA.
Kelvin waves are an important oceanic teleconnection mechanism between the Tropics and higher latitudes. Theoretical (Moore, 1968) and numerical (Pares-Sierra and O'Brien, 1989) ocean models show that tropically generated Kelvin waves are a significant source of interannual variability of the near-shore ocean circulation along western America, as well as the generators of large amplitude Rossby waves in the midlatitude Pacific basin (Johnson and O'Brien, 1990; Jacobs et al., 1994) which play a vital role in decadal climate variability (Meyers et al, 1996).
Detecting these Kelvin waves has proven difficult. Poleward propagating waves at interannual timescales associated with El Niño have been found, but these waves were only at intraseasonal timescales (Enfield and Allen, 1980; Chelton and Davis, 1982; Spillane et al., 1987). This has led to the neglect of the Kelvin wave as a powerful mechanism of oceanic variability, particularly at the higher latitudes of the Pacific Ocean. However, more recent analysis of temperature and sea level data has shown that Kelvin waves generated during extremes of ENSO can propagate from the tropical Pacific up the west coast of North America to the Aleutian Island Chain (Norton and McLain, 1994; Meyers et al., 1998), and the Kamchatka peninsula (Smedstad et al., 1997; Metzger et al., 1998). Ramp et al. (1997) showed sea level along the western coast of North America during the 1991 El Niño was characterized by a northward propagating signal originating in the Tropics and a southward signal originating in the GOA. The former was attributed to a coastally trapped Kelvin wave and the latter to wind forcing. Wind anomalies outside the Tropics (Wallace and Gutzler, 1981) are another source of ENSO-induced variability in the ocean circulation.
The general ocean circulation in the GOA is dominated by the cyclonic Pacific subpolar gyre, as revealed by dynamic topography charts (Lagerloef, 1995). The current associated with this gyre forms the Alaska Current (AC), a broad and weak eastern boundary current flowing northwards in the eastern GOA. The AC becomes narrower and stronger as it approaches the northern GOA, ultimately leaving the region as the Alaskan Stream, an intense southwestward flowing current along the southern coast of Alaska (west of the panhandle) and the Aleutian Island chain. Furthermore, there is a wave guide at the eastern margin of the GOA where the flow is highly variable due to variations in wind forcing and coastally trapped waves propagating in from the south (Chelton and Davis, 1982).
In this investigation, we examine the results from a 1/8° 6-layer isopycnal model with realistic coastline geometry and bottom topography. For model details, see Hurlburt et al. (1996). The numerical simulation is driven by realistic daily wind stress from to 1981 to 1994. An annual cycle of eddy activity is found in the ocean circulation of the eastern GOA. This numerical experiment indicates strong interannual variability in this region. The existence of these seasonal and interannual variations is supported by observations of the "Sitka eddy" (Tabata, 1982).
The model results are made available by the Naval Research Laboratory (NRL), Stennis Space Center, MS. The model experiment is a multilayer Pacific basin simulation using the NRL Layered Ocean Model (NLOM) (Wallcraft, 1991). NLOM is based on the semi-implicit, free surface model of Hurlburt and Thompson (1980), and it is formulated using an Arakawa C-grid (Arakawa, 1966). The daily 1000 mb wind product from the European Centre for Medium-Range Weather Forecasts (ECMWF) [with the 1981-1991 mean replaced by the Hellerman and Rosenstein (1983) (HR) annual mean] drives the ocean circulation. Recent results from this class of models and additional model description are given by Hurlburt et al. (1996).
|
|
Fig. 1,
one of two panels: NLOM bottom topography in the eastern GOA. |
The Pacific model domain extends from 20°S to 62°N and has a horizontal resolution of 1/8° x 45/256° (latitude x longitude) for each variable, which is approximately 14 x 10 km in the Alaska Gyre region. Isopycnal outcropping is an essential feature of ocean models covering the GOA. This is included by entrainment from the layer below whenever a layer becomes thinner than a prescribed minimum thickness. Mass is conserved within the layers, so that an accumulation of entrained mass in one layer is balanced by an equal amount of detrained mass elsewhere in the model domain; see Shriver and Hurlburt (1997) for a detailed discussion.
|
|
Mov. 1.
Mean model upper layer thickness in the eastern GOA. |
3.1 General circulation and eddy formation
In the model GOA, the thicknesses of the first and second layers are typically 50-60 m and 50-150 m, respectively. However, there are frequently large deviations from these means. Along the coast, the interface between the upper two layers can be deflected up to 100 m below its local mean depth. Outside the coastal region, this deflection may have the same deviation magnitude at the center of westward-drifting eddies. A poleward flowing current corresponding to AC is observed in the simulation results. The model AC has a seasonal cycle similar to that reported by Lagerloef et al., (1981), who attributes the cycle to a seasonal shift in the atmospheric circulation from the intense Aleutian low during winter to the North Pacific high pressure system during summer (Royer, 1975). In the model, the transport, upper layer thickness and vertical shear of the velocity associated with the coastal current reach maximum values in December or January. At this time, the magnitude of the velocity differences between layer 1 and layer 2 is typically 10 cm/s in coastal regions (approximately 15 and 5 cm/s in layers 1 and 2, respectively); with maximum values exceeding 40 cm/s.
In most winters the model AC meanders. The alongshore wavelength and offshore amplitude of the meanders are typically 200 km and 40 km, respectively. However, the amplitude may become ~100 km, after which the current usually breaks into predominantly anticyclonic eddies. The eddies are seen to drift slowly southwestward (i.e., offshore), with an estimated speed of ~1 cm/s, or ~300 km/year. Generally, the horizontal pressure gradient is significantly larger in layer 1 than in layer 2. Hence, the accompanying velocity shear is significant and the motion in the meanders is strongly baroclinic.
Fewer cyclonic eddies are generated and they decay more rapidly. The generation of cyclonic eddies may be artificially suppressed due to their relatively small horizontal extent, since horizontal gradients are dampened by a Laplacian friction which selectively dampens small scales. (This is necessary for numerical stability). Further, the production of cyclonic eddies may be reduced due to the limited space inshore of the current. Cyclonic eddies may also be weakened by the isopycnal outcropping in the model, since such outcropping reduces thickness gradients of shallow model layers. The thickness gradients are the model's representation of the horizontal density gradients. Isopycnal outcropping is characteristic of this region.
3.2 Instability processesInstability processes at the oceanic mesoscale are traditionally divided into barotropic and baroclinic instability. Barotropic instability is the process where mesoscale features (meanders, filaments, eddies) develop from a basic state by conversion of kinetic energy from the basic state to the eddy field (Rayleigh, 1880; Kuo, 1949; Fjørtoft, 1950). In the case of an f-plane, a necessary condition for instability is the presence of an inflection point in the basic state flow. In contrast, baroclinic instability is the process in which energy is converted from (basic state) potential energy to mesoscale energy (Eady, 1949; Phillips, 1951). In a two-layer system, the wavelength of the most rapidly growing disturbance is a function of mean flow, velocity shear in the vertical, buoyancy, layer thickness and latitude.
The instability mechanism is investigated here by examination of a cross-section at 56.75°N. Using the surface elevation and deflection of the interface between layers 1 and 2 as the basic state on a straight geostrophic f-plane channel, the growth rate of unstable perturbations may be found.
|
|
Fig. 2.
Perturbation growth rate as a function of wavenumber. |
|
|
|
|
Fig. 4,
one of four panels: Eigenvector solution for upper layer thickness. |
The stability of the cross-section may be studied in more detail by solving an eigenvalue/ eigenvector problem using a finite difference formulation for perturbation variables (EV analysis); this is also described in Appendix A. Assuming a two-layer system, the growth rates computed by this method are depicted as a function of wavenumber by the green line in Figure 2. From Figure 3 we note that the growth rate of perturbations increases by almost a factor of 2 from January 22 to February 3, 1983. This is probably due to potential energy added by the local winds, or a combination of this and a deepening of the Kelvin signal.
|
|
Fig. 3,
one of four panels: Model upper layer thickness in the GOA following the
1982-83 El Niño. |
|
|
Fig. 5,
one of two panels: Perturbation growth rate as a function of wave number,
on Jan 28, 1983 (red), and on Nov. 7, 1988 (blue). |
|
|
Fig. 6,
one of two panels: Anomaly of the combined thickness of the upper two
layers on Mar. 2, 1983 in the GOA. |
The EV analysis was repeated for a case of downwelling in November 1988, i.e., during the 1988-89 El Viejo episode. The cross-section at 56.75°N was again investigated, and the result of the analysis is that the perturbations with the highest growth rates have wavelengths of around 100 km. In the present simulation, one wavelength is then spanned by 7 grid nodes, and the numerical damping of small horizontal scales will have a negative effect on the generation of eddies. The growth rate on November 7, 1988 is displayed as a function of wave number in Figure 5 (top panel) where it is compared to the results for January 28, 1983. Also in Figure 5 (bottom panel), the duration of the unstable perturbations is depicted. Here, the red line shows the growth rate of the 200 km-long wave perturbation in 1983, and the blue line is the corresponding development of the 100 km-long wave perturbation in 1988. Note the relatively small changes of high growth rates in the 1983 case, which indicates that instability-favorable conditions lasted for at least two e-folding periods of 11.5 days.
Feliks and Ghil (1993) performed a detailed instability analysis of a downwelling front along the southern coast of Asia Minor in the Eastern Mediterranean. They found that the f-plane approximation worked well until eddies had developed, except for cases with very small basic state currents. They also found that increasing the number of vertical modes beyond two does not significantly affect the most unstable wave. This explains the relative success of the fairly simple EV analysis above.
We end this discussion by noting that the present analysis is unfit to describe the later stages of the eddy generation, and their eventual drift westwards. Anticyclonic eddies will become larger, and cyclonic eddies become smaller, due to non-linear (semi-geostrophic) effects, as described by Hoskins (1975). This is also observed in the model results, see e.g., Movie 3 below.
3.3 ENSO effects
|
|
Mov. 2.
Model eddies in the GOA after the 1982-83 El Niño. |
|
|
Mov. 3.
Offshore propagation of eddies from spring to summer 1983. |
One of the two major El Niño events in recent history occurred during the 1982-83 boreal (northern) winter (the other event being that of 1997-98). The most intense eddy formation event in the model run occurs at this time. Movie 2 displays the eddy generation in the GOA following the 1982-83 event. In the movie, the evolution of the model upper layer is shown and Lagrangian vectors for the circulation of the upper layer are included. The evolution and offshore propagation of the eddies that are generated in Movie 2 are animated in Movie 3. One observes that the anticyclonic eddies remain as the dominating feature of the Gulf of Alaska ocean circulation throughout 1983. In 1984, the model circulation in the GOA is dominated by the lasting effect of the strong eddies that were generated the year before.
|
|
Mov. 4.
Meandering of the coastal jet in the GOA after the 1982-83 El
Niño. |
|
|
Mov. 5.
Weak GOA eddies after the 1988-89 El Viejo. |
Model results indicate that upwelling perturbations stabilize the flow and yield few and weak eddies. There are two El Viejo events (Meyers et al., 1999) during the model study, late 1984 and early 1988. No deep eddies are generated during the 1988 event. Figure 6 (right panel), however, clearly shows the generation of some cyclonic eddies (although not as apparent given the choice of color scales). Mesoscale motion is also weak following the 1984 event. (The 1984 El Viejo was of moderate size, and is not recognized as an ENSO event by all indices.) Eddy generation is common during the winter and the sparsity of eddies suggests these ENSO events may suppress normal eddy development. Movie 5 displays the thickness of the upper layer in the model following the 1988-89 event. Some upper layer anticyclones of relatively small size can be seen in this animation.
4.1 Sea surface height
|
|
Fig. 7.
Position of IGOSS station Sitka, AK (pink), and model point Sitka
(yellow). |
|
|
Fig. 8b.
Observed (blue) and modeled (red) SSH anomaly at Sitka, AK for one year
centered at newyear 1983, after applying a two month LPB filter. |
The model results have been sampled at a frequency of 1/(3.05 days). Hence, we interpolate the daily SSH observations to model output dates using a low-pass boxcar filter (LPB filter). In panel a of Figure 8, the resulting SSH anomalies for a one year period centered at January 1 1983 are depicted. The observed SSH values exhibit much larger fluctuations than the model results. This discrepancy is likely due to the exclusion of the continental shelf in the model domain. Thus, generation of shelf waves are suppressed in the numerical model, if not eliminated. Inclusion of these waves is instrumental in the onshelf description of SSH during storm surges (Martinsen et al., 1979).
In order to discriminate the high frequency onshelf fluctuations, the SSH time series were subjected to a two month LPB filter for the same period. The resulting SSH variations are depicted in panel b of Figure 8. This figure clearly demonstrates the accuracy of the model results. However, the model SSH is unquestionably lagging the data. The highest correlation is achieved with a lag of 12-18 days, when the correlation exceeds 0.99, whereas the no-lag correlation is 0.94. (The corresponding correlation values in the non-filtered case are 0.78 at a lag of 3 days, and 0.72, respectively.) Note also that the 1982-83 SSH mean is well above the zero level (which is equal to the 1981-1994 mean value). This figure depicts the period where the model-observation correlations reach their highest values, see Table 1.
| Period | 3.05 sample | 2 month LPB filter | ||
| Lag | Correlation | Lag | Correlation | |
| 1981-82 | 3 | 0.67 | 24 | 0.93 |
| 1982-83 w | 3 | 0.78 | 15 | 0.99 |
| 1983-84 | 3 | 0.58 | 12 | 0.83 |
| 1984-85 c | 3 | 0.67 | 43 | 0.90 |
| 1985-86 | 3 | 0.73 | -6 | 0.94 |
| 1986-87 w | 9 | 0.75 | 12 | 0.97 |
| 1987-88 | 3 | 0.75 | 12 | 0.93 |
| 1988-89 c | 3 | 0.42 | 6 | 0.33 |
| 1989-90 | 46 | 0.61 | 46 | 0.94 |
| 1990-91 | 6 | 0.58 | 9 | 0.61 |
| 1991-92 w | 3 | 0.79 | 18 | 0.97 |
| 1992-93 | 3 | 0.64 | 6 | 0.86 |
| 1993-94 w | 9 | 0.66 | 15 | 0.96 |
| Tbl. 1. Correlations of observed and modeled SSH at Sitka. Lags are in days, positive when observations lead model results. Years denoted by w (in red) and c (in blue) correspond to El Niño events and El Viejo events, respectively. | ||||
From Table 1, we see that the 2 month LPB filtered correlations are usually between 0.85 and 1. However, it is obvious that the model is unsuccessful in reproducing the observed interannual SSH variability in 1988-89 when the correlation drops to 0.33. The reason for this discrepancy lies mainly in inaccurate model reproduction of two strong spikes in the observed SSH for this period. A positive anomaly of more than 30 cm is observed in November 1988, and a negative anomaly reaching almost 40 cm is measured in February 1989. By examination of the 2 month LPB filtered time series, we find that the model lags the observed positive anomaly by about half a month, and leads the negative anomaly by approximately one month. Furthermore, the amplitudes of the modeled spikes are much smaller than what the observations reveal. This combination leads to the low correlation value for 1988-89.
From Table 1, we also note the substantial model-data lag at which the highest correlation value occurs in 1989-90. The model has also been run using the National Centers for Environmental Prediction's (NCEP) reanalysis winds. Using this alternative forcing, the model-data intercomparison for 1989-90 is more favorable. However, results for some of the other years are better with the present forcing by ECMWF winds. Thus, the main source of inaccurate model results for SSH is most likely the quality of the forcing fields, i.e., the winds. Another possible source for model errors with respect to negative SSH anomalies is the model's management of thin layers by discrimation of sharp coastal fronts in regions of strong upwelling. This may lead to coastal gradients that are too small in the model results, and the largest misrepresentation occurs at the coast.
We end this discussion by noting that the model upper layer thickess is strongly correlated to the model sea level at seasonal time scales. The no-lag correlation of the non-filtered time series is 0.940, and the correlation of the 2 month LPB filtered time series is 0.943. However, at interannual time scales the correlation between upper layer thickess and sea level is smaller. After application of a one year LPB filter, this correlation is 0.783.
4.2 Eddies
Strong eddy formation due to a propagating internal Kelvin signal also seems consistent with observations. In an extensive examination of oceanographic data collected during 1927-1977 in the GOA, Tabata (1982) concluded that ``baroclinic eddies occur frequently in this region. Among these is the recurring, well developed, anticyclonic eddy situated within a few hundred kilometers of Sitka'', now commonly called the Sitka eddy. The primary characteristics of the Sitka eddy are reproduced qualitatively and quantitatively by the numerical model (diameter 150 km - 300 km, isopycnal deflection of up to 100 m, initial location ~200 km off Baranof Is., lifetime of order one year).
|
|
Fig. 9,
one of ten panels: Model upper layer thickness on Feb. 28 1988. |
Tabata (1982) refers to the possible presence of the Sitka eddy with various degrees of certainty. The uncertainty may be attributed to the sparsity of available data as well as interannual variability of the mesoscale ocean circulation off Sitka. As can be seen from Figure 9, the simulation results contain a significant amount of interannual variability in this region.
In the datasets that are surveyed by Tabata (1982), the spring of 1958, the summers of 1960 and 1961, and the spring of 1977 are the times when the Sitka eddy was undoubtedly present. Both 1957-58 and 1976-77 were seasons with an El Niño event in the tropical Pacific Ocean.
Repression of eddy activity following El Viejo tropical events is supported by the measurements off Sitka. There are two such events reviewed in Tabata (1982), which occurred in 1955-56 and 1956-57. In the spring and summer of 1956 and 1957, Tabata concludes that the Sitka eddy is either weak or non-existent.
The simultaneous presence of multiple anticyclonic eddies in the GOA similar to the present results from the NLOM (e.g., Movie 5) was recently established from observations by Thomson and Gower (1998). Examinating thermal imagery from the winter of 1995, they detected five well-defined warm eddies along the coast from just south of Queen Charlotte Islands in the south to the apex of the GOA in the north. The diameter of the eddies was approximately 160 km, and they were spaced about 250 km apart. Thomson and Gower (1998) attribute the eddy generations to strong velocity shears in the vertical due to a sudden wind reversal. In the tropical Pacific Ocean, 1994-95 was an El Niño event according to some indices, but not all.
|
|
Fig. 10.
Topex/Poseidon satellite altimetry for late April, 1998. |
The altimeter data have been processed using the Geophysical Data Records of each satellite. The SSH measurements have been corrected for atmospheric effects, removal of tides, inverse barometer pressure loading and orbit errors. The Topex/Poseidon satellite has an alongtrack resolution of approximately 6 km, and a between-track resolution of about 170 km at these latitudes. The corresponding numbers for the ERS-2 sattelite are 7 km and 23 km, respectively. The repeat cycles are 10 days for Topex/Poseidon and 35 days for ERS-2.
|
|
Fig. 11,
one of four panels: Model SSH in the eastern Pacific Ocean in November
1982. |
|
|
Fig. 12.
Observed (blue) and modeled (red) SSH anomaly at Sitka, AK after
application of a one year LPB filter. |
|
|
Fig. 13,
one of four panels: Observed (blue) and modeled (red) SSH anomaly at
Neah Bay, WA after application of a one year LPB filter. |
|
|
Fig. 14.
ECMWF pseudo wind stress alongshore component (top) and observed SSH
anomaly (bottom) at Sitka, AK, after application of a one year LPB
filter . |
Next, we consider the question of whether the SSH variability in this region is generated by winds (local forcing), or by oceanic teleconnections. In Figure 14, the alongshore component of the pseudo wind stress at 57.5° N, 137.5° W, and the observed SSH anomalies at Sitka are depicted after application of a one year LPB filter. It is obvious from this figure that the anomalously high SSH values reached in early 1983 have not been generated by local winds. Reexamining Figure 13, the 1982-83 SSH anomalies appear to have propagated along the North American coast in a south to north direction, with the largest anomalies in the south. This observation is consistent with a recent study of oceanic teleconnections by coastally trapped Kelvin waves in the equatorial and northeastern Pacific (Meyers et al., 1998). SSH anomalies during 1982-83 are lag correlated up the coast at speeds indicative of Kelvin waves, suggesting the major SSH anomalies of this event are remotely forced in the Tropics and not locally forced by winds. In this way, the major 1982-83 El Niño has affected the ocean circulation along the coasts of the entire American continent. Huyer and Smith (1985) presented results from observations off Oregon, and concluded that initial anomalies in the sea level and in the hydrography following the 1982-83 El Niño were most plausibly explained by coastal Kelvin waves. Moreover, the high sea level observed at the Oregon coast was subsequently enhanced by local winds. Thus, the Kelvin signal that reached the GOA may have been reinforced by local winds on its path from the Equator.
Local wind forcing appears to have affected the interannual variability significantly in the late 80's and early 90's. During these years, the SSH anomalies are also somewhat stronger in the northern GOA. This discussion of SSH variability along the GOA coastline supports the results of Emery and Hamilton (1985), who found that variations in coastal SSH may be attributed to local winds in some winters, but not all. In particular, they list 1968-69, 1971-72 and 1972-73 as winters with anomalously large SSH without correspondingly strong local winds. It is interesting to note that both 1968-69 and 1972-73 correspond to El Niño events.
The model results for interannual variability of SSH (Figure 12) and mesoscale activity (Figure 9) indicate a relationship between these quantities. In a study of anticyclonic eddies in the Eastern Mediterranean, it was shown that the growth rate of the most unstable wave is proportional to the maximum speed of the basic state current (Felix and Ghil, 1993). This observation may explain the relationship between SSH and mesoscale activity: The model currents are predominantly geostrophic, so large SSH values at the GOA coast correspond to strong currents.
The possible relationship between ENSO events and eddies in the Gulf of Alaska was noted by Swaters and Mysak (1985), who suggested that this may be due to variations in the atmospheric circulation of this region linked with such events. The results of the numerical simulation strongly supports the hypothesis of a link between ENSO events and intense mesoscale motion in the GOA. The 1982-83 El Niño was one of the major events of the ocean-atmosphere system in the 20th century and appears to have produced deep eddies in the GOA. The possible link is also seen in the mesoscale activity following the 1991-92 El Niño. On the other hand, the 1986-87 El Niño, which was of moderate size, did not spawn deep eddies in the GOA in the numerical model. Note that the 1986-87 El Niño lasted for more than a year (Meyers and O'Brien, 1995), so the 1988 winter circulation depicted in Figure 9 may also be influenced by ENSO. Note also that the model underrepresents the amplitude of the 1986-87 El Niño (but not the other two) when forced by the operational ECMWF winds.
From the results for SSH and winds above, we suggest that the deep model eddies generated in early 1983 are predominantly due to oceanic ENSO teleconnections, whereas the deep eddies in 1992 could be influenced by atmospheric teleconnections. The lack of a significant ENSO response in 1986-87 may be due to inadequate ECMWF winds over the equatorial Pacific during this time (Hundermark et al., 1996). Moreover, the local wind forcing contains variability on a decadal time scale that almost certainly has a counterpart in the oceanic variability (Lagerloef, 1995). The present simulation period is unfortunately too short to examine this variability. However, we recall from the discussion in Section 4.2 that the correlation between sea level and upper layer thickness is lower at interannual timescales than at seasonal timescales. This may reflect a decadal variability in the local wind forcing.
Having established NLOM as a numerical model that is suitable for examination of interannual variability in the GOA, our future plans include performing and analyzing a decade-long simulation of the Pacific Ocean with the tropical latitudes excluded. In this way we will be able to discriminate oceanic teleconnections from the equatorial Pacific ocean during ENSO events. Then, we should be in a position to be able to distinguish between effects of oceanic and atmospheric teleconnections on the GOA ocean circulation.
The long-term effect of an El Niño event in the North Pacific Ocean circulation was reported based on results from another version of the present numerical model (Jacobs et al., 1994). There it was demonstrated that the 1982-83 El Niño triggered planetary waves that crossed the North Pacific basin and caused a partial northward rerouting of the Kuroshio Extension in 1992-93. Computing the empirical orthogonal functions of Geosat altimetry data from the Exact Repeat Mission, Bhaskaran et al. (1993) detected a mode in the GOA region for which no cause was found in the local forcing mechanisms. However, the mode, which was strongly reflected in hydrographic data, correlated well with a Southern Oscillation index. This paper demonstrates that the high-latitude ocean responds strongly to ENSO events, having memory over thousands of kilometers and many years. How remotely forced interannual changes in upper ocean circulation affect marine life or local weather patterns is not fully understood.
Acknowledgements
Financial supported is provided by the Research Council of Norway (Norges
Forskningsråd, NFR) and by the Strategic Environmental Research and
Development Program (SERDP). The Center for Ocean-Atmosphere Prediction
Studies receives its base funding through the Office of Naval Research
Secretary of the Navy Chair awarded to Prof. James J. O'Brien. Funding
from NFR is provided under contracts 100674/410 and 122230/720. Funding
for the Naval Research Laboratory is provided by the Office of Naval
Research under program element 62435N and Advanced Research Projects
Agency grant SC25046 and SERDP. The model simulation was performed on the
Cray YMP at the Naval Oceanographic Office, Stennis Space Center, Miss.,
USA. The near real-time and reanalysed altimeter data were provided by
the Colorado Center for Astrodynamics Research (CCAR) at the University
of Colorado, Boulder. Thanks to Dr. Detlev Müller for many helpful
discussions, and to Prof. Robert R. Leben for generously providing the
altimetry image in Figure 10.
Arakawa, A., 1966: Computational design for long-term numerical integration of the equations of fluid motion: Two-dimensional incompressible flow. Part I. J. Comput. Phy., 12, 12-35.
Bhaskaran, S., G. S. E. Lagerloef, G. H. Born, W. J. Emery and R. R. Leben, 1993: Variability in the Gulf of Alaska from Geosat Altimetry data. J. Geophys. Res., 98C, 16311-16330.
Chelton, D. B. and R. E. Davis, 1982: Monthly mean sea-level variability along the west coast of North America. J. Phys. Oceanogr., 12, 757-784.
Cummins, P. F. and L. A. Mysak, 1988: A quasi-geostrophic model of the northeast Pacific. Part I: A preliminary numerical experiment. J. Phys. Oceanogr., 18, 1261-1286.
Eady, E. A., 1949: Long waves and cyclone waves. Tellus, 1, 33-52.
Emery, W. J. and K. Hamilton, 1985: Atmospheric forcing of interannual variability in the northeast Pacific ocean: Connections with El Niño. J. Geophys. Res., 90C, 857-868.
Enfield, D. and J. S. Allen, 1980: On the structure and dynamics of monthly mean sea level anomalies along the Pacific Coast of North and South America. J. Phys. Oceanogr., 10, 557-578.
Feliks, Y. and M. Ghil, 1993: Downwelling-front instability and eddy formation in the Eastern Mediterranean. J. Phys. Oceanogr., 23, 61-78.
Fjørtoft, R., 1950: Application of integral theorems in deriving criteria of stability for laminar flows and for the baroclinic circular vortex. Geofys. Publ., 17 No 6, 52 pp.
Hellerman, S. and M. Rosenstein, 1983: Normal monthly wind stress over the world ocean with error estimates. J. Phys. Oceanogr., 13, 1093-1104.
Hoskins, B. J., 1975: The geostrophic momentum approximation and the semi-geostrophic equations. J. Atmos. Sci., 32, 233-242.
Hundermark, B. W., H. E. Hurlburt, E. J. Metzger and J. F. Shriver, 1999: A comparison of wind stresses derived from ECMWF 1000 mb winds and FSU pseudo stresses over the tropical Pacific Ocean, 1981-1993. Tech. Rep. NRL/FR/7323--95-9643, Naval Research Laboratory, Stennis Space Center, MS, in press.
Hurlburt, H. E. and J. D. Thompson, 1980: A numerical study of loop current intrusions and eddy shedding. J. Phys. Oceanogr., 10, 1611-1651.
Hurlburt, H. E., A. J. Wallcraft, W.J. Schmitz, Jr., P. J. Hogan and E. J. Metzger, 1996: Dynamics of the Kuroshio/Oyashio current system using eddy-resolving models of the North Pacific Ocean. J. Geophys. Res., 101C, 941-976.
Huyer, A. and R. L. Smith, 1985: The signature of El Niño off Oregon, 1982-1983. J. Geophys. Res., 90C, 7133-7142.
Jacobs, G. A., H. E. Hurlburt, J. C. Kindle, E. J. Metzger, J. L. Mitchell, W. J. Teague and A. J. Wallcraft, 1994: Decade-scale trans-Pacific propagation and warming effects of an El Niño anomaly. Nature, 370, 360-363.
Johnson, M. A. and J. J. O'Brien, 1990: The northeast Pacific Ocean response to the 1982-1983 El Niño. J. Geophys. Res., 95C, 7155-7166.
Kuo, H. L., 1949: Dynamic instability of two-dimensional nondivergent flow in a barotropic atmosphere. J. Meteor., 6, 165-185.
Lagerloef, G. S. E., 1995: Interdecadal Variations in the Alaska Gyre. J. Phys. Oceanogr., 25, 2242-2258.
Lagerloef, G. S. E., R. D. Muench and J. D. Schumacher, 1981: Low-frequency variations in currents near the shelf break: Northeast Gulf of Alaska. J. Phys. Oceanogr., 11, 627-638.
Martinsen, E. A., B. Gjevik and L. P. Røed, 1979: A numerical model for long barotropic waves and storm surges along the western coast of Norway. J. Phys. Oceanogr., 9, 1126-1138.
McGowan, J. A., 1985: El Niño 1983 on the southern California bight. In: El Niño north: El Niño effects in the eastern subarctic Pacific Ocean, W. S. Wooster and D. L. Fluharty, editors (Univ. of Washington Sea Grant Publ.), 166-184.
Metzger, E. J., H. E. Hurlburt, J. C. Kindle, R. C. Rhodes, G. A. Jacobs, J. F. Shriver and O. M. Smedstad, 1998: The 1997 El Niño in the NRL Layered Ocean Model. 1998 NRL Review, Naval Research Laboratory, Washington, DC, 63-71.
Meyers, S. D., M. A. Johnson, M. Liu, J. J. O'Brien and J. L. Spiesberger, 1996: Interdecadal variability in a numerical model of the northeast Pacific Ocean. 1970-1989. J. Phys. Oceanogr., 26, 2635-2652.
Meyers, S. D., A. Melsom, G. T. Mitchum and J. J. O'Brien, 1998: Detection of the Fast Kelvin Wave Teleconnections due to El Niño-Southern Oscillation. J. Geophys. Res., 103C, 27655-27663.
Meyers, S. D. and J. J. O'Brien, 1995: Pacific Ocean influences atmospheric carbon dioxide. EOS, 76, 1.
Meyers, S. D., J. J. O'Brien and E. Thelin, 1999: Reconstruction of monthly SST in the tropical Pacific Ocean during 1868-1993 using adaptive climate basis functions. Mon. Weather Rev., in press.
Moore, D. W., 1968: Planetary-gravity waves in an equatorial ocean. Ph.D. thesis, Harvard University, Cambridge, MA. 207 pp.
Mysak, L. A., 1985: On the interannual variability of eddies in the northeast Pacific Ocean. In: El Niño north: El Niño effects in the eastern subarctic Pacific Ocean, W. S. Wooster and D. L. Fluharty, editors (Univ. of Washington Sea Grant Publ.), 97-106.
National Oceanic and Atmospheric Administration, 1988: Digital relief of the Surface of the Earth. Data Ann. 88-MGG-02, NOAA, Nat. Geophys. Data Center, Boulder, CO. [Available online at http://www.ngdc.noaa.gov:80/mgg/global/seltopo.html].
Norton, J. G. and D. R. McLain, 1994: Diagnostic patterns of seasonal and interannual temperature variation off the west coast of the United States: Local and remote large-scale atmospheric forcing. J. Geophys. Res., 99C, 16019-16030.
Pares-Sierra, A. and J. J. O'Brien, 1989: The seasonal and interannual variability of the California Current system: A numerical model. J. Geophys. Res., 94C, 3159-3180.
Phillips, N. A., 1951: A simple three-dimensional model for the study of large-scale extratropical flow patterns. J. Meteor., 8, 381-394.
Ramp, S. R., J. L. McClean, C. A. Collins, A. J. Semtner and K. A. S. Hays, 1997: Observations and modeling of the 1991-1992 El Niño signal off central California. J. Geophys. Res., 102C, 5553-5582.
Rayleigh, Lord, 1880: On the stability, or instability, of certain fluid motions. Proc. London Math. Soc., 9, 57-70.
Royer, T. C., 1975: Seasonal variations of water in the northern Gulf of Alaska. Deep Sea Res., 22, 403-416.
Shriver, J. F. and H. E. Hurlburt, 1997: The contribution of the global thermohaline circulation to the Pacific to Indian Ocean throughflow via Indonesia. J. Geophys. Res., 102C, 5491-5511.
Smedstad, O. M., D. N. Fox, H. E. Hurlburt, G. A. Jacobs, E. J. Metzger and J. L. Mitchell, 1997: Altimeter data assimilation into a 1/8° eddy resolving model of the Pacific Ocean. J. Met. Soc. Japan, 75, 429-444.
Smith, P. E., 1985: A case history of an anti-El Niño to El Niño transition on plankton and nekton distribution and abundances. In: El Niño north: El Niño effects in the eastern subarctic Pacific Ocean, W. S. Wooster and D. L. Fluharty, editors (Univ. of Washington Sea Grant Publ.), 121-142.
Spillane, M. C., D. B. Enfield and J. S. Allen, 1987: Intraseasonal oscillations in sea level along the west coast of the Americas. J. Phys. Oceanogr., 17, 313-325.
Swaters, G. E. and L. A. Mysak, 1985: Topographically-induced baroclinic eddies near a coast, with application to the Northeast Pacific. J. Phys. Oceanogr., 15, 1470-1485.
Tabata, S., 1982: The anticyclonic, baroclinic eddy off Sitka, Alaska, in the Northeast Pacific Ocean. J. Phys. Oceanogr., 12, 1260-1282.
Thomson, R. E. and J. F. R. Gower, 1998: A basin-scale oceanic instability event in the Gulf of Alaska. J. Geophys. Res., 103C, 3033-3040.
Walker, G. T., 1928: World weather. Q. J. R. Meteorol. Soc., 54, 79-87.
Wallace, J. M. and D. S. Gutzler, 1981: Teleconnections in the geopotential height field during Northern Hemisphere winter. Mon. Weather Rev., 109, 784-812.
Wallcraft, A. J., 1991: The Navy layered ocean model users' guide. NOARL Rep. 35, Naval Research Laboratory, Stennis Space Center, MS. 21 pp.
Ware, D. M. and R. E. Thomson, 1991: Link between long-term variability in upwelling and fish production in the Northeast Pacific Ocean. Fish. Ocean., 48, 2296-2306.