SciELO - Scientific Electronic Library Online

vol.68 número2  suppl.TIProcVERTICAL TEMPERATURE AND MOISTURE STRUCTURE IN LOWER ATMOSPHERE RETRIVED FROM TERRA/MODIS índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados


Gayana (Concepción)

versión impresa ISSN 0717-652Xversión On-line ISSN 0717-6538

Gayana (Concepc.) v.68 n.2 supl.TIProc Concepción  2004 


Gayana 68(2) supl. t.I. Proc. : 325-326(e), 2004 ISSN 0717-652X



Vincent Echevin1, Olivier Aumont1, Jorge Tam2 & Jose Pasapera2

1. Laboratoire d'Océanographie Dynamique et de Climatologie (LODYC), Unité Mixte de Recherche 7617 CNRS / IRD / Université Pierre et Marie Curie (UPMC), Institut Pierre Simon Laplace, Tour 14-15 - 2ème étage - Boîte 100 4, place Jussieu 75252 PARIS Cedex 05, France. E-mail:
2. Research Center for Oceanographic and Biological Fishery Modelling (CIMOBP), Peruvian Marine Research Institute (IMARPE), Apdo. 22, La Punta, Lima, Peru.


The seasonal cycle of surface chlorophyll along the Peruvian coasts is studied with SeaWiFS satellite observations and coupled dynamical/biogeochemical ORCA/PISCES model simulations. The observed mean spatial structure of surface chlorophyll is shown to be in good agreement with that of the model. However the seasonal cycles contrast: whereas the model produces a spring bloom consistent with maximum upwelling and deepest mixed layer in the winter season, the observed seasonal cycle shows two distinct blooms one in late spring-early summer and other in fall. Several hypotheses are suggested to explain the fall bloom and its non-occurrence in the model.



The Peruvian coast (4°S-18°S) features the strongest upwelling of the Humboldt Current System (HCS) along the eastern coasts of South America. It also has a strong ENSO-related interannual variability (Carr, 2003, Carr et al., 2002) which impacts the important fisheries activities of Peru and Chile (FAO, 1999). The high biological activity off Peru is due to year- long coastal upwelling due to the trade winds. Cold, nutrient rich subsurface waters are advected vertically into the euphotic zone, and a intense biological activity develops along the coast. The maximum of upwelling occurs in winter with the reinforcement of the trade winds (Bakun and Nelson, 1991). However the seasonal variation of phytoplankton biomass appears to be out of phase with upwelling intensity (Chavez, 1995, Thomas et al., 2001, Ochoa et al., 2004) and the dynamical and/or biogeochemical processes at stake remain unexplained. Here we examine the seasonal cycle of the surface chlorophyll in SeaWiFS satellite observations during the period 1998-2003, and use a three-dimensional coupled dynamical-biogeochemical model of the upwelling area to study the dynamical forcing of the seasonal variability of surface biomass. Hypothetical explanations of the observed discrepancies between the model and the observations are presented, and some conclusions are suggested to explain the observed seasonal cycle. In the following section, we describe the data and methods used in this work, then present the results obtained in the third section. In the last section, a discussion and conclusions are given.


The SeaWiFS satellite data collected during the period 1998-2002 is used. The data is gridded at a rather low resolution (0.5°) every 8 days, in order to avoid the high concentrations nearshore, which may not correspond to actual chlorophyll values. A climatological time series is constructed by averaging the data values each month. When focus is given on the coastal values, the closest grid point near the coast is extracted from the gridded data set.

Results from a coupled dynamical/biogeochemical model are used as a guideline for the analysis of surface observations. The coupled model is composed of a dynamical component, the ORCA model (Lengaigne et al., 2002) configured at 0.5° resolution in the Pacific ocean, forced by ERS/Quickscat winds, NCEP heat fluxes, and Xie precipitation. The biogeochemical component is the PISCES model (Aumont et al., 2003), an ecosystem model which is able to represent phytoplankton growth limitations by several nutriments such as nitrate, phosphate, silicate, dissolved iron. Dissolved inorganic and organic nutrients including iron, are also provided by the river run-offs on a annual mean basis. The model is run over the period 1992-1999, the year 1992 is not considered as it is perturbated by dynamical adjustment. The mixed layer depths, vertical velocities, nutrients and phytoplankton biomass of the model are used in the following analysis.


The annual mean surface chlorophyll in the model and observations is represented Figure 1. Both display similar spatial patterns. The maximal chlorophyll is concentrated near the coast between 3°S and 18°S, with observed maximum values of ~3-4 mgChl/m3 and 0.5-1 mgChl/m3 in the model. The highest values are encountered near 6°S in the observations (respectively 7°S in the model) and between 10°S and 15°S, located with the main upwelling centers (Strub et al., 1998). The relatively low values produced by the model may be due to the weak vertical velocities generated by the ORCA dynamical model. At the model's resolution, bottom topography is not well resolved and vertical velocities, although maximum at the coast, remain noisy on the horizontal and weak (~0.5 m/day) compared to the vertical velocities of high resolution regional models (Penven et al., 2003) and reported in the literature (5-30 m/day, Strub et al., 1998).

The cross-shore gradients of chlorophyll display strong differences. If one considers the 0.4 mgChl/m3 (respectively 2 mgChl/m3 in the model) as a virtual boundary separating the most productive coastal areas from offshore waters, chlorophyll values decrease much more rapidly in the offshore direction in the model results. Very low values of ~0.15 mgChl/m3 are reached within 400-500 km from the coast. In SeaWiFS such low values are reached ~1000 km offshore. Such a discrepancy can be due to the absence of mesoscale and submesoscale processes in this fairly low resolution model, which generally act as a onshore-offshore transfer of nutrients and biomass (Marchesiello et al., 2004) in the coastal transition zone.

The seasonal cycle is calculated over the period 1993-1999 for the PISCES model and over 1998-2002 from SeaWiFS data. Latitude-time Hovmuller diagrams of the coastal chlorophyll values are shown in Figure 2. PISCES surface chlorophyll displays an expected seasonal cycle: chlorophyll is maximum in September-October at the end of winter. Indeed the model mixed layer depth increases during winter due to the stronger trade winds and the cooling of the surface water. It is maximum (~70 m) at the end of winter, with depth ranges greater than climatological values (~40 m, de Boyer Montegut et al., 2004) but with the correct seasonal phase. With this depth increase, nutrients are entrained into the surface layer and a nearshore spring bloom occurs in October-November over 4°S-16°S latitude range. Note that, except over three bands of year-long higher production and contrasted seasonal cycle near 6.5°S, 11°S and 12.5°S, the bloom persists during 1-2 months and ends at the beginning of November.

Minimum chlorophyll values are reached in February-March over most of the Peru coasts.

On the opposite, there is not only a spring bloom but also a fall bloom in the observations. The spring blooms starts in October and reach maximum values in November-December. Then the productivity decreases in January and increase to produce a second bloom in February-March, during the season of weakest winds and coastal upwelling. There is also a propagating pattern in the observations, which is not reproduced by the model. There appears to be a slow southward propagation of the bloom maximum in October-December and in February-April between 9°S and 14°S.

Figure 1. Mean surface chlorophyll in (left) PISCES model in 1993-1999 and (right) SeaWiFS data in 1998-2002. Units: mgChl/m3

Figure 2. Time-latitude diagram of the alongshore surface chlorophyll for a climatological year (left) PISCES model, (right) SeaWiFS data. Units: mgChl/m3


These results raise many questions: what are the major dynamical and biogeochemical processes which control the seasonal cycle of surface chlorophyll? what accounts for the differences in seasonal cycle phase between the model and the observations? why isn't there a late austral summer/early fall bloom in PISCES model over most of the coast? what could generate a southward propagation of the bloom as the one observed on SeaWiFS?

Light is not the limiting in the upwelling region due to the low latitude, and there is evidence of mixed layer control of productivity in the observations (Guillen and Calienes, 1980). This seems indeed to be the case in the model, since surface biomass is maximum in October right after the maximum mixed layer depth, whereas the upwelling maximum is reached in August-September (Chavez, 1995). Still, there appears a slight time delay, the maximum of the spring bloom being reached slightly earlier in the model (October) than in the observations (November). This delay may be caused by the effect of mesoscale activity which is more intense during the upwelling phase, and which may transport the upwelled nutrients offshore. After the upwelling, eddies and filaments are less active and nutrients remain close to the coast. Hence an intense biomass can develop nearshore, as confirmed by biomass measurements (Rojas de Mendiola, 1981). In the low resolution dynamical model, mesoscale processes are not resolved, the nutrients are not advected offshore, and the bloom can start earlier in the season.

Let us now investigate in more detail the biogeochemical processes at stake in the model. As noted previously, three locations along the coast show year-long high surface chlorophyll concentration (Figure 2). These high values are due to year-average iron input from three main rivers located at these nearshore model grid points. There, iron concentration is high all year round and phytoplankton biomass can grow faster and longer than elsewhere along the coast, where production of picoplankton and diatoms is iron limited. Because of this high biomass, there is a strong remineralisation following the spring bloom and the fall bloom can be triggered when nitrates are brought to the surface as the first trade wind blows take place, as for example in April 1996 (Figure 3).

Along the Fe-depleted coast, phytoplankton growth is relatively weak all year round. The spring bloom is triggered when the mixed layer deepens and reaches the iron nutricline depth, which, in the model, is deeper than the nitracline and phosphate nutricline (Figure 4). Therefore, this bloom last only one-two months whereas it is more intense and longer in the observations.

In fall, the production is iron limited and, because of the greater depth of the iron nutricline than that of the nitracline, the weak upwelling generated by the first wind blows does not trigger an intense bloom as in places where iron is not limiting. Therefore it is a combination of different colimitations and different nutricline depths which explains the presence, in the model results, of a fall bloom at nitrate/phosphate limited locations and the absence of it at Fe-depleted locations.

A combination of the two co-limitations could occur in reality. Indeed, phytoplankton Fe-limitation has been recently measured in the Peru upwelling (Hutchins et al., 2002), but previous work suggest that silicate (Dugdale, 1972, Guillen et al., 1977) and nitrate could regulate productivity (Guillen and Calienes, 1980). Both hypotheses are confirmed by our model simulations depending on the location along the coast.

Two poleward propagating signals of a high chlorophyll signal at 0.2 m/s 0.1 m/s can be seen from Figure 2 at the end of summer-beginning of fall. The propagation speed is quite slow compared to measured coastally trapped waves phase speed of ~2-3 m/s (Brink, 1982), which rules out this dynamical process. The summer-fall period corresponds to the minimum of wind and upwelling. During these seasons the equatorward Peru coastal current (PCC) is weakest (Strub et al., 1998, Penven et al., 2003) and the poleward Peru undercurrent (PUC) may reach the surface (Huyer et al., 1991) and advect nutrients-rich surface water southward. The source of nutrients may be as far north as 4°S in January, as shown in Figure 5. Poleward advection due to the surfacing PUC propagative may contribute to the poleward propagation of the bloom.

Four simplified parameterizations in the coupled dynamical/biogeochemical model may have an impact on the seasonal cycle of primary productivity:

First the model river run-offs are located at only three grid points. In reality, there are numerous small rivers spread over the coastline. The coastal currents may advect nutrients along the coast and generate an alongshore continuous bloom between 4°S and 16°S, as observed in Figure 2. Second, with the low resolution (1/2°) of the dynamical model, the coastal current is weak and wide-spread. There is no advection from iron-rich to iron depleted areas along the coast, hence all the iron is consumed locally by the ecosystem, and this produces the local "grid-point" high concentration values. Third, there is a strong seasonal cycle of the river run-off along the Peruvian coasts, which was not introduced in the model. The main rivers of this very dry area between 6°S and 20°S receive water only in fall due to the rain season in the Andes. The input of iron and/or of other nutrients from river run-off in fall may enhance the fall bloom.

Fourth, there is no parameterization of the sediments input of iron in the model. This source of Fe is probably very important in the Peru upwelling, especially as it is highly soluble in the shallow suboxic waters found in this region (Bruland et al., 1991, Morales et al., 1999) . Owing to the seasonal variability of the oxycline depth, there may be an effect on the Fe input from sediments, which is not taken into account here.

Parameterization of the seasonal cycle of runoff and sediments Fe-input should further improve the model results, and indicate which process has the greatest impact on productivity. Furthermore, dynamical modelling at a higher horizontal resolution should produce more realistic upwelling rates and mesoscale circulation, which may have an impact on both the timing and amplitude of the phytoplankton blooms.

Figure 3. Temporal variation of the surface (5 m depth) chlorophyll in PISCES model at a coastal location at 12°S during the years 1993 (black), 1994 (red), 1995( green), 1996 (dark blue) and climatological (light blue). Units: mgChl/m3

Figure 4: Time-depth evolution of the Fe concentration (shading) and NO3 concentration at 12°S. Units: 104 mol/l(Fe), mol/l (NO3)

Figure 5: Time-latitude diagram of the alongshore surface NO3 concentration in Levitus climatology.


The seasonal cycle of surface chlorophyll along the Peruvian coasts was studied with SeaWiFS satellite observations and compared to outputs of a coupled dynamical/biogeochemical model implemented in the region. The spatial structure of surface chlorophyll is maximum nearshore, and maximum between 4°S and 18°S, and model and observations are in good agreement. The seasonal cycle shows a spring bloom and a fall bloom in the observations, whereas in the model, only a spring bloom occurs at most locations along the coast, where productivity is iron-limited. At coastal locations where iron is brought by river run-off, the model phytoplankton growth is no longer iron-limited and a weak nitrate-limited fall bloom takes place at the beginning of the upwelling season. Elsewhere the fall bloom does not occur because high iron concentrations are buried at greater depths than the nitracline. Furthermore the fall bloom may be enhanced by the riverine input which is concentrated in the fall season due to precipitation in the Andes. Poleward propagation of a coastal chlorophyll signal takes place between 9°S and 14°S. This signal may be linked to poleward advection of nutrients by a surfacing coastal undercurrent.


We thank staff from Instituto del Mar del Peru (IMARPE) for valuable discussions. Jorge Tam acknowledges the financial support from the Institut de Recherche pour le Développment (IRD).


Aumont, O., E. Maier-Reimer, S. Blain & P. Monfray, 2003. An ecosystem model of the global ocean including Fe, Si, P colimitations. Global Biogeochemical Cycles , Vol. 17, No. 2, 1060. [ 1]

Bakun A. & C.S. Nelson, 1991. The seasonal cycle of wind stress curl in sub-tropical eastern boundary current regions. Journal of Physical Oceanography, 21, 1815-1834. [ 2]

Brink K.H., 1982. A comparison of long coastal trapped waves theory with observations off Peru. Journal of Physical Oceanography, 12, 897-913. [ 3]

Bruland , K. W., J.R. Donat & D.A. Hutchins, 1991. Interactive influences of bioactive trace metals on biological production in oceanic waters, Limnology and Oceanography. 36:1555-1577. [ 4]

Carr M.-E., 2003. Simulation of Carbon pathways in the planktonic ecosystem off Peru during the 1997-1998 El Nino and La Nina. Journal of Geophysical Research, Vol. 108, C12,3380. [ 5]

Carr M.-E., P.T. Strub, A.C. Thomas & J.L. Blanco, 2002. Evolutions of the 1996-1999 La nina and EL Nino conditions off the western coast of south America: a remote sensing perspective, Journal of Geophysical Research, 108, C12, 3236. [ 6]

Chavez F. P., 1995. A comparison of ship and satellite chlorophyll from Califronia and Peru. Journal of Geophysical Research, Vol. 100, C12, 24,855-24,862. [ 7]

de Boyer Montégut C., G. Madec, A. S. Fischer, A. Lazar & D. Iudicone, 2004. A global mixed layer depth climatology based on individual profiles, Journal of Geophysical Research, under revision. [ 8]

Dugdale, R.C., 1972. Chemical oceanography and primary productivity in upwelling regions, Geoforum, 2, 47-61. [ 9]

FAO, 1999. The state of world fisheries and aquaculture 1998. FAO documentation group, Rome, Italy. [ 10]

Guillen, O., R. Calienes & R. de Rondan, 1977. Medio ambiante y production primaria en el area Pimentel-Chimbote, Boletin del Instituto del Mar del Peru, 3, 107-159. [ 11]

Guillen, O. & R. Calienes, 1980. Biological productivity and El nino, Fisheries management and environmental uncertainties: el nino and the anchovy, Wiley Interscience, 600 pp. [ 12]

Huyer A., M. Knoll, T. Paluszkiewicz & R.L. Smith, 1991. The Peru undercurrent: a study in variability. Deep Sea Research, Vol. 38(S), S247-S271. [ 13]

Hutchins D.A., et al., 2002. Phytoplankton iron limitation in the Humboldt current and Peru upwelling. Limnology and Oceanography, 47,997-1011. [ 14]

Lengaigne, M., J.-P. Boulanger , C. Menkes, S. Masson, P. Delecluse & G. Madec, 2002. Ocean response to the March 1997 westerly wind event, Journal of Geophysical Research, 107. [ 15]

Marchesiello P., S. Herbette, L. Nykjaer & C. Roy, 2004. Eddy-driven dispersion processes in the Canary Current upwelling system: comparison with the California system, submitted to GLOBEC Newsletter. [ 16]

Morales, C.E., S.E. Hormazabal & J.L. Blanco, 1999. Interannual variability in the mesoscale distribution of the depth of the upper boundary of the oxygne minimum layer off northern chile (18S-24S): Implications for the pelagic system and biogeochemical cycling. Journal of Marine Research. 57:909-932. [ 17]

Ochoa, N., E. Ramos & M. Baylon, 2004. Variabilidad temporal de la comunidad fitoplanctonica marina en un area somera de la bahia de Ancon durante 1992-2003. XIII RC ICBAR, April 2004. [ 18]

Penven, P., J. Pasapera, J. Tam & C. Roy, 2003. Modeling the Peru Upwelling System seasonal dynamics Mean circulation, seasonal cycle and mesoscale dynamics, GLOBEC International Newsletter, 9, 23-25. [ 19]

Rojas de Mendiola B., 1981. Seasonal phytoplankton distribution along the Peruvian coast. In Coastal Upwelling, F.A. Richards, ed. American Geophysical Union, Washington,D.C., 348-356 [ 20]

Thomas A.C., M.E. Carr & P. T. Strub, 2001. Chlorophyll variability in eastern boundary currents, Geophysical Research Letter. 28: 3421-3424. [ 21]


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