SciELO - Scientific Electronic Library Online

vol.30 número1  suppl.SympComparative Analysis of Warm Events in the Series 1970 - 2002 off the Coast of PeruSea Level Variations in Callao (12° S) - Peru 1942 - 2001 índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados


Investigaciones marinas

versión On-line ISSN 0717-7178

Investig. mar. v.30 n.1 supl.Symp Valparaíso ago. 2002 

Coastal Upwelling Response to
Atmospheric Wind Forcing along the
Pacific Coast of the United States.

P. Penven, J. C. McWilliams, P. Marchesiello

IGPP, UCLA, 405 Hilgard Avenue, Los Angeles, CA
90095-1567, USA, Email:


Wind stress is recognized as the most important driving force for coastal upwelling. In the California Current System, along the West Coast of the United States, although wind stress is mainly directed towards the equator, i.e. favorable to upwelling, it shows important spatial and temporal variations. Wind stress curl follows a typical pattern for Eastern Boundary Currents: Anticyclonic offshore in association with the North Pacific Subtropical Anticyclone and Cyclonic in a band 200 km to 300 km wide adjacent to the coast.

Wind stress curl is the forcing term in the Sverdrup balance equation. For the California Current System, the cyclonic wind stress curl close to the shore is held responsible for the poleward undercurrent and for the poleward coastal counter current known has the Davidson Current.

A previous numerical realistic simulation at high resolution (5 km) of the California Current System employed the COADS climatology for its upper boundary conditions. The COADS climatology is based on reports of surface marine and atmospheric observations. The objective analysis performed to obtain the gridded field uses 771 km for its smallest radius of influence. As a consequence, COADS presents a very smoothed wind field with almost no wind stress curl close to the shore. The differences between the oceanic model solution and the observations are believed to be in a great extent caused by this lack of mean wind stress curl.

Several other wind products are now available. They come from global model reanalysis, mesoscale regional model outputs and satellite observations. For this study we use NCEP for the global model product, COAMPS for the regional model output, and QuickSCAT for the satellite observations. The grid resolution of these data-sets ranges from more than 100 kilometers for COADS and NCEP to 9 kilometers for COAMPS. From this point, and since they are coming from such scat tered sources, we expect that these products present strong differences.

To understand the impact of the mean wind and wind stress curl structure on the dynamics of the California Current System, 4 numerical simulations at high resolution are conducted, forced by 4 monthly mean climatologies derived from the different wind products. This work concentrate only on spatial variations; interannual variability and higher frequency synoptic variations in the forcings are not taken into account. We focus our attention on the dynamics of the Central Upwelling of California, a portion approximately 1200 km long and 600 km wide of the California Current System, situated between Point Conception and Cape Mendocino.


At large scale, the different wind stress annual mean climatologies are relatively similar. The mean wind stress magnitude is approximately the same for each product. The main structure is an equatorward wind, almost parallel to the shore, showing a large scale anticyclonic bend offshore, a decrease near the coast, a maximum offshore of Point Conception, and a minimum near Cape Blanco. In the case of COADS, this pattern is very smooth. There is almost no decrease at the shore, and the wind maximum lies about 300 km off Point Conception. The maximum for NCEP is elongated, following the coastline at a constant distance (about 300 km) from the Southern California Bight to Cape Mendocino. From this maximum, the wind diminishes slowly towards the coast. The high resolution atmospheric model COAMPS presents a very particular behavior: it shows a sharp decrease in wind speed in a 10 to 30 km band close to the shore. This is not resolved by any of the other products. In this case, the wind stress maximum is localized in the immediate vicinity of Point Conception. The wind is also weak in the Southern California Bight. While lacking this sharp gradient at the shore, QuickSCAT observes a similar wind pattern. It also presents a secondary maximum west of Cape Mendocino.

These differences are mirrored in the wind stress curls. All exhibit the anticyclonic-cyclonic structure characteristic of the eastern boundaries. For COADS it ranges from -1e-7 Pa/m to 1e-7 Pa/m, whereas the cyclonic curl reaches 3e-7 Pa/m for NCEP, following a pattern consistent with the elongated maximum. The sharp gradient at the coast in COAMPS induces stress curls of about 10e-7 Pa/m to almost 50e-7 Pa/m! We don't know whether such values are realistic. For both COAMPS and QuickSCAT, the cyclonic wind stress curl area is confined closer to the shore in a band 150 km wide.

A new model of the Central California Upwelling System has been developed to assess the ocean response to the different wind forcings. A nesting capability has been integrated into the Regional Ocean Modeling System (ROMS) to obtain local solutions at high resolution while preserving the large-scale circulation at affordable computational cost. It has been applied to a domain at 5 km resolution that covers the central upwelling region of the United States West Coast, around Monterey Bay, embedded into a domain at 15 km resolution including the whole US Pacific Coast. Long term simulations (10 years) have been conducted to obtain yearly cyclic statistical equilibria. Four experiments are pursued: EXP_COADS, EXP_NCEP, EXP_COAMPS, and EXP_QSCAT; forced by the 4 differents wind stress monthly climatologies. The solutions are analyzed from year 3 to year 10.

After a short spin-up, each of the 4 simulations produces a well developed upwelling. During the summer months, a sharp front separates the cold upwelled water from the warmer ocean. Filaments and eddies are expelled from the front in a realistic manner. They are predominantly generated in the proximity of capes as observed in nature. While the solutions are highly turbulent, no significant discontinuity is observed at the connection between the large scale and the embedded grids.

At the surface, all the simulations present a meandering equatorward California Current. On average, it generates 3 large scale meanders: offshore of Cape Mendocino, Point Reyes and Point Conception. The offshore wind maximum induced more pronounced meanders in EXP_NCEP, whereas they are more confined to the shore (as the wind maximum) for EXP_COAMPS and EXP_QSCAT. The impact of the mean wind stress curl is most noticeable in the vertical structure of the alongshore mean currents. For EXP_COADS, at the surface, the current is equatorward all year round with a maximum at the coast. In a relatively narrow band (50 km), a poleward undercurrent flows along the shore at a depth of about 50 to 150 m. In the case of EXP_NCEP, the poleward undercurrent attains a width of 250 km. It surfaces in winter to give a surface coastal counter current 150 km wide. This surfacing of the poleward flow is also observed in EXP_COAMPS, but it is much narrower (50 km) and almost two times stronger. In the case of EXP_QSCAT, we get a pattern comparable to EXP_COADS.

The alongshore currents are compared to the Accoustic-Doppler Current Profiler (ADCP) data collected from MBARI's M1 buoy near the mouth of Monterey Bay. These data show the shoaling of the Undercurrent and its surfacing as the Davidson Current. While all the model solutions compare fairly well with the data, only EXP_COAMPS exhibits this characteristic pattern.

The seasonal anomalies can be directly compared to satellite altimetry sea surface height measurements. Going from the coast towards the open Ocean, the winter anomalies have a typical maximum - minimum - maximum pattern, the maximum at the shore being the signal of the Davidson Current. All the simulations show an equivalent structure. The amplitude is too weak for EXP_COADS, and the smooth wind structure induced a pattern too broad for EXP_NCEP. Both EXP_COAMPS and EXP_QSCAT compare very well with the satellite measurements. For summer, the sea surface height anomaly is reversed to give a minimum - maximum - minimum structure associated with upwelling. In this case, the amplitude for EXP_COAMPS is too large at the coast north of 40 N. Here EXP_QSCAT is here in very good agreement with the observations.

When comparing with seasonal temperature cross-sections derived from 50 years of data along Calcofi line 67 (offshore of Monterey Bay), the model solutions did not present a strong sensibility to the different wind products. Although they are very close to the observations, all solutions are slightly too warm (0.5 C to 1 C) at the shore. They are also slightly less stratified in winter and too cold offshore in summer (0.5 C to 1 C). This might be induced by a weakness in the heat flux parameterization or a missing process like the tides or the wind synoptic variability.


A new model of the Central California Upwelling System has been set up to test the impacts of the mean wind stress structure on the dynamics of the California Current System. Four products are tested: COADS, NCEP, COAMPS and QuickSCAT. They differ strongly in their representation of the wind stress close to the shore.

As one would expect, a correct representation of the wind stress curl at the shore is mandatory to obtain a realistic Davidson Current. Because COADS and NCEP are too coarse and because the scatterometer measurements of QuickSCAT can not be achieved in the vicinity of the coastline, this is accomplished solely by COAMPS. Meanwhile, the experiments forced by QuickSCAT were the closest to the observations away from the shore.

This work focused on the impact of the mean local forcing on the California Current System. Other constraints could also affect the ocean along the Pacific Coast of the United States, such as the synoptic and interannual variations of the surface forcing or oceanic teleconnection via coastal trapped waves. The tools developed for this project can be adapted to include these processes.

Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons