ARTICLE IN PRESS

Deep-Sea Research II 51 (2004) 587–600 www.elsevier.com/locate/dsr2

Satellite observations of mesoscale eddies in the Gulfs of Tehuantepec and Papagayo (Eastern Tropical Pacific)

Adriana Gonzalez-SilveraÃ, Eduardo Santamaria-del-Angel, Roberto Milla´ n-Nun˜ ez, He´ ctor Manzo-Monroy

Facultad de Ciencias Marinas, Universidad Auto´noma de Baja California, AP. 453, Ensenada, BC - CP 22800 Mexico

Accepted 3 May 2004

Abstract

SeaWiFS high-resolution ocean-color images and AVHRR sea-surface temperature were obtained for the period from November 1998 to March 1999. Generation of mesoscale eddies was observed for the region between the Gulf of Tehuantepec and the Gulf of Papagayo (Tropical Pacific Ocean). Eighteen eddies with diameters ranging between 100 and 450 kmwere identified; 14 originated in the Gulf of Tehuantepec and mostwere cyclonic. Our results show that the frequency of cyclonic formation is higher than it has been reported and their lifetime can be longer. Using both ocean color images and daily temporal resolution, instead of sea-surface temperature alone, improved our ability to identify and follow the propagation of these eddies. The generation and fate of smaller cyclonic eddies around the periphery of the anticyclonic eddy indicate the importance of the onshore–offshore exchange of energy and biological material, which has not been considered previously. Three anticyclonic eddies were generated in the Gulf of Papagayo. These eddies have a common origin, and they travel along the main flux of the Costa Rica Coastal Current (CRCC), which later turns offshore and joins the North Equatorial Current. This behavior can be seen clearly on SeaWiFS images and suggests that the CRCC influences eddy propagation. Time series of chlorophyll-a concentration and SST were obtained fromtwo points at Tehuantepec and two at Papagayo. Cross-correlation analysis confirms an inverse relationship between these variables, showing that a decrease in SST is followed by an increase in chlorophyll-a concentration values, indicating high growth and primary production rates. r 2004 Elsevier Ltd. All rights reserved.

1. Introduction ÃCorresponding author. Tel.: +52-646-1744570x109; fax: +52-646-1744103. The eastern border of the tropical Pacific Ocean, E-mail addresses: [email protected] (A. Gonzalez-Silvera), between 51N and 211N is influenced by intense [email protected] (E. Santamaria-del-Angel), [email protected] (R. Milla´ n-Nun˜ ez), [email protected] events that produce at three sites (H. Manzo-Monroy). of Northern and (Lavı´ n et al.,

0967-0645/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.dsr2.2004.05.019 ARTICLE IN PRESS

588 A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600

1992; Barton et al., 1993; Trasvin˜ a et al., 1995; sphere, a cyclonic eddy means that the sense of Chelton et al., 2000a,b; Mu¨ ller-Karger and rotation of such eddy is anti-clockwise, and when a Fuentes-Yaco, 2000; McClain et al., 2002): the vortex is developed it causes the upwelling of Gulf of Tehuantepec in Mexico, the Gulf of subsurface waters in its center and a downwelling Papagayo in the Costa Rica– border, at its extremities. The upwelling of subsurface and the Gulf of Panama´ . These strong, focused waters brings colder and nutrient-reach waters can last from1 to 15 days and blow offshore to the photic zone of the water column, promoting over the Pacific Ocean, generating filamentous phytoplankton growth. For this reason a cyclonic ocean jets and high waves along their trajectory. eddy appears as a cold and high biomass- Their effect favors strong near-shore oceanic core eddy. Anticyclonic eddies have an opposite mixing and intense lowering of sea-surface tem- behavior, appearing as a warm-core and low peratures (Barton et al., 1993; Trasvin˜ a et al., biomass-core eddy. These facts have proved the 1995). This phenomenon usually occurs in fall, use of temperature and ocean color satellite images winter and early spring when polar air masses to be efficient in visualizing and following eddy move south into the and the motion in this area (Lluch-Cota et al., 1997; Intertropical Convergence Zone (ITCZ) is at its Mu¨ ller-Karger and Fuentes-Yaco, 2000; McClain southernmost position (Chelton et al., 2000a,b). et al., 2002). Another feature observed in this zone is the Costa Mu¨ ller-Karger and Fuentes-Yaco (2000) used Rica Dome (CRD), centered at 901W and 91N, satellite images from both the ocean-color sensor which is a shoaling of the strong and shallow Coastal Zone Color Scanner (CZCS) and ad- thermocline of the eastern Pacific Ocean. Surface vanced very high-resolution radiometer (AVHRR) currents flow cyclonically around it and its to study these eddies and their effect on pigment seasonal evolution is affected by the same large- concentration in the region of Tehuantepec, scale wind patterns. The upwelling associated with Papagayo and Panama´ . From1979 to 1986, the dome is about 0.5 m dayÀ1, yielding cool sea- the formation of anticyclonic and cyclonic eddies surface temperatures (SST) (Fiedler, 2002). Anti- that moved distances greater than 1500 km from cyclonic and cyclonic eddies are generated in this their point of origin were reported (Mu¨ ller-Karger zone as a result of the effect of wind stress over the and Fuentes-Yaco, 2000); this emphasizes their ocean (Hansen and Maul, 1991; Barton et al., importance as source of energy and biological 1993; Mu¨ ller-Karger and Fuentes-Yaco, 2000; constituents fromthe continental marginto McClain et al., 2002). The result is that surface the offshore tropical Pacific. In a more recent waters are fertilized by the introduction of paper, McClain et al. (2002) confirmed the nutrient-rich waters fromsubsurface layers or importance of coastal winds and Ekman diver- fromthe coast; both sources increase phytoplank- gence for the formation of phytoplankton ton abundance, chlorophyll-a concentration (Tras- blooms along the coast of Central America vin˜ a et al., 1995; Lluch-Cota et al., 1997; Mu¨ ller- and the Costa Rica Dome (CRD). Considering Karger and Fuentes-Yaco, 2000; McClain et al., the knowledge of the region, the period from 2002) and modify trophic conditions as shown November 1998 to March 1999 was chosen, when analyzing zooplankton productivity data for to study the type (anticyclonic or cyclonic), the Gulf of Tehuantepec (Fa¨ rber-Lorda et al., number and frequency of eddy generation, propa- 2003). The overall effect of these strong winds can gation, and the relationship between chlorophyll-a be inferred on fisheries and the fact that interna- and SST for the region between the Gulf of tional pelagic fishing fleets heavily exploit this Tehuantepec and the Gulf of Papagayo. High- area. Indeed, the eastern tropical Pacific contains resolution images were used (1.1-km spatial some of the most productive waters of the world’s resolution in daily composites) from the ocean- oceans (Fiedler et al., 1991). color sensor Sea Viewing Wide Field of View An eddy is a deviation in the steady-state flow of Sensor (SeaWiFS) and complemented with SST a fluid, causing a vortex. In the northern hemi- data fromAVHRR. ARTICLE IN PRESS

A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600 589

2. Methods DAS 4.1), where chlorophyll a concentration (Chla) was estimated using the OC4 algorithm The conventional method to obtain a synoptic (O’Reilly et al., 1998), making a total of 146 view of a cloudy region is to calculate the average images. The study region was set between 102 and value (e.g., chlorophyll) for a specific location 811W and 0 and 211N, which includes both the during a certain time period (e.g., 1 week). This is Gulfs of Tehuantepec and Papagayo (Fig. 1). not ideal for studying dynamic processes, which These images were examined to obtain informa- will be blurred or not be seen at all. Daily images tion on eddy formation, size and range of Chla were used to avoid this problem, although cloudy concentration, and eddy duration. areas were obtained occasionally. The period of Sea-surface temperature (SST) images were November 1998–March 1999 was chosen, consid- obtained fromthe NOAA/NASA AVHRR Path- ering that this period has the best sequence of finder subsetting system( http://podaac.jpl.nasa. SeaWiFS images in comparison with other years, gov). These images are daily records of the and it is also the period of the year when wind jets ascending pass of the sensor at 9 kmspatial are strongest and eddy generation frequency is resolution, and they correspond to PO.DAAC’s higher. V4.1 data. SST data were used for comparison with Each image was processed using the SeaWiFS Chla maps in order to confirm sense of rotation of processing package distributed by NASA (Sea- eddies, range of temperature, and also as a way to

Fig. 1. Study region and location of points where time series of chlorophyll-a and sea-surface temperature data were taken (stars). Gray arrows indicate the location of wind jets entering the Pacific Ocean over the continent. Black arrows represent surface currents for boreal winter (NEC=North Equatorial Current, CRCC=Costa Rica Coastal Current, SEC=South Equatorial Current). ARTICLE IN PRESS

590 A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600 complement information on eddy translation when In order to ascertain the relationship between no SeaWiFS images were available. Chla concentration and SST, time series were Eddies were first identified by looking for eddy- extracted fromthe locations indicated in Fig. 1: shaped features on ocean-color images. Next, the Gulf of Tehuantepec, the Gulf of Tehuantepec profiles of Chla concentration and SST for each and 430 kmoffshore; the Gulf of Papagayo and eddy were compared to confirm the sense of 450 km offshore. Some time-series data are missing rotation. SST images were occasionally unavail- due to cloud cover; the number of data (or days) able, so the alternative was to use the image of used for this study is shown in Fig. 2. In order to 1 day before or after the Chla image we had in have a better sequence of data in time series, order to confirmour observations. The isolines of missing data were interpolated using local regres- 0.25 and 0.5 mÀ3 of Chla were used as a guide to sion models, lowess for short (Cleveland and set the external limits and the diameter of eddies. Devlin, 1988; Chambers and Hastie, 1991). These With this procedure, we tried to be objective and series were used for cross-correlation analysis. consistent in our observations and measurements. There are no ocean-wind data for the period Eddies were labeled according to their sense of analyzed in this study because NSCAT generated rotation (anticyclonic, A or cyclonic, C), origin data for only 1996 and 1997 and QuikSCAT wind (Tehuantepec, T or Papagayo, P) and date of data are only available since September 1999. generation. However, and to have reference, wind-speed data

Fig. 2. Number of images data per pixel considering the total set of images used (from November 1998 to March 1999). ARTICLE IN PRESS

A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600 591 were obtained froman on-shore station at The generation of cyclonic eddies occurs Huatulco Airport (Oaxaca, Me´ xico), located near after the generation of an anticyclonic eddies, the Gulf of Tehuantepec (Fig. 1), and these data and Table 1 was arranged in such a way that the were used for comparison with images from the name of the cyclonic eddies follows the name of Gulf of Tehuantepec. the anticyclonic ones. For example, AT08nov98 has four cyclonic eddies associated to it; their generation dates were November 13, December 1, 18 and 25 (Fig. 5). Their shape can be circular or 3. Results elongated, and in most situations they occur east of the main wind jet (Fig. 4). Cyclonic eddies The analysis of SeaWiFS and AVHRR images last between 4 and 30 days, whereas anticyclonic showed a very dynamic system over the study eddies can last for more than 100 days (see region, where Chla ranged between 0.1 and Table 1). An exception was CT25dec98 (Fig. 5), 10 mg mÀ3, and SST between 23 and 31 1C. which survived for almost 2 months (see Table 1). Monthly maps of mean Chla and SST illustrate This eddy had an interesting behavior also temporal and spatial variability of the study region observed for eddy CT14feb99 (Fig. 6). Both were (Fig. 3). During November it is possible to observe smaller (80–174 km) in diameter and were gener- the presence of high concentrations of Chla in the ated close to an anticyclonic eddy (AT08nov98 Gulf of Tehuentepec extending offshore as a broad and AT07jan99); their movement followed the filament and cooler waters dominating the region. sense of rotation of the main eddy until their FromNovemberto Decemberthere is an increase dissipation, which occurred when they approached on phytoplankton bloomand a decrease on SST the coast. that also can be observed in Papagayo. An eddy- Because of their long life span, anticyclonic like feature can be seen on both places, although it eddies fromTehuantepec movelong distances to is more visible at Tehuantepec. In January, bloom the west or southwest. In this study, we observed conditions are stronger in both Tehuantepec and distances as long as 700 kmfromtheir point of Papagayo, and a jet of higher Chla concentrations origin; these distances have probably been under- extends farther offshore until February when these estimated, considering that some eddies were still strong extensions show a connection. In March, inside the study area in our last image (March 26). SST values rise along the Gulf of Tehuantepec and Indeed, previous studies reported distances as far the strong jet developed in February begins to as 1500 km(Mu ¨ ller-Karger and Fuentes-Yaco, move back. The Papagayo region, however, 2000). Considering the distance and time taken presents higher concentrations of Chla during by each eddy fromits point of origin to its final this month. position, we estimated their translation speed. We Eighteen eddies were identified using daily first set the diameter of the eddy using the images of Chla and SST; their characteristics are 0.5 mg mÀ3 isoline, and fromthis imaginary line listed on Table 1. Fourteen of those eddies (the diameter) we considered its central point to be originated in the Gulf of Tehuantepec and most the center of the eddy. We found velocities from are cyclonic. Anticyclonic eddies are larger and 4.7 to 8.02 cms À1 for anticyclonic eddies of last longer than cyclonic, which have a life span of Tehuantepec region and 6.8 cms À1 to 13.3 cms À1 approximately 30 days. The generation of cyclonic for those of the Papagayo region (Table 1). eddies generally occur east of the 951 meridian, at AT14dec98 (Fig. 5) did not present a significant the Gulf of Tehuantepec (Fig. 4). Anticyclonic propagation, and for this reason we did not eddies, however, occur west of 951W, which calculate its speed. Cyclonic eddies have speeds corresponds to the western region of the gulf. varying from3.99 to 15.78 cms À1. The exceptions were AT14dec98 and AP14dec98, Three anticyclonic eddies were generated at their origin being observed in the area between the Gulf of Papagayo. Their origin was centered Tehuantepec and Papagayo. on 86.921W and 11.241N, and they followed a ARTICLE IN PRESS

592 A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600

Fig. 3. Monthly composites of chlorophyll-a concentration (mg mÀ3) fromSeaWiFS and sea-surface temperature(SST, 1C) from AVHRR. The mean position of the Costa Rica Dome (from Fiedler, 2002) is superimposed. ARTICLE IN PRESS

A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600 593

Table 1 Description of eddies identified in this work

Name Date end Days Lat/Lon begin Lat/Lon end Diameter (km) Chla (mg mÀ3) Speed (cms À1)

AT08nov98 2/22 106 14.75 12.10 270–400 0.18–3.5 8.02 À95.81 À102 CT13nov98 12/10 27 13.15 11.99 67–173 0.18–1.35 10.5 À95.12 À97.04 CT01dec98 12/10 10 14.75 14.75 210 0.5–6.8 — À94.47 À94.47 CT25dec98 01/23 54 13.256 14.81 80–175 0.2–6.6 6.34 À97.485 À99.72 AT14dec98 12/25 11 13.76 13.89 197 Â 51 0.1–0.48 — À92.34 À91.97 AT18dec98 12/23 5 15.07 14.78 230–350 0.3–8.7 7.45 À95.37 À96.17 CT18dec98 01/03 16 14.52 14.99 210 0.5–10.3 6.52 À94.81 À95.49 AT07jan99 3/19 71 14.96 14.02 270–400 0.2–2.3 4.7 À96.19 À98.72 CT07jan99 01/18 12 13.15 13.204 223 Â 61 0.3–2.9 4.69 À94.39 À94.83 CT29jan99 02/07 10 13.78 13.83 225 Â 68 0.1–4.2 3.99 À94.36 À94.68 CT14feb99 03/11 19 13.62 15.38 89–100 0.1–0.7 12.19 À99.14 À98.74 CT25feb99 Last 30+ 13.52 12.68 158 Â 257 0.2–1.2 15.78 À96.99 À100.66 CT08mar99 03/11 4 12.94 12.86 413 Â 89 0.1–4.1 7.84 À94.94 À95.18 CT05feb99 02/16 12 12.34 12.521 À94.89 À93.99 170–250 0.2–2.3 9.55 AP14dec98 02/04 52 12.57 12.76 À90.34 À93.49 220–450 0.1–2.5 7.5 AP20dec98 Last 96+ 11.54 10.45 255–430 0.5–6.8 À87.14 À95.06 10.4 AP29jan99 Last 57+ 11.81 10.83 206–430 0.3–9.7 6.8 À87.76 À90.69 AP19mar99 Last 8+ 11.11 11.55 170–200 0.27–9 13.28 À86.52 À87.22

The name includes type (anticyclonic, A or cyclonic, C), origin (Tehuantepec, T or Papagayo, P) and date of generation. Date of disappearance is also indicated, number of days each one lasts (+ means they possibly last more), latitude and longitude of beginning and end, diameter (km), range of chlorophyll-a concentration and speed of each eddy.

northwards trajectory along the coastline, turning Near the Gulf of Papagayo, the Costa Rica southwest at approximately 88.71W and 12.71N Dome (CRD) is an important feature that and traveling about 874 kmoff the coast. The influences phytoplankton pigment distribution origin of AP14dec98, however, was north of (Muller-Karger and Fuentes-Yaco, 2000; Fiedler, Papagayo (Fig. 4) presenting lower concentrations 2002). Its mean position (from Fiedler, 2002), of Chla and a northwest propagation to the Gulf superimposed over monthly Chla and SST images, of Tehuantepec. No cyclonic eddy was observed in is shown on Fig. 3. The CRD is closer to the coast this region. in February–March than in November–December. ARTICLE IN PRESS

594 A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600

20

18

16

14

12

10

8

6 -102 -100 -98 -96 -94 -92 -90 -88 -86 -84 -82

Fig. 4. Location where eddies started forming. Circles correspond to anticyclonic eddies and triangles to cyclonic eddies.

Lower SST values are observed, as a result of and a decrease on Chla when the strong jet coastal shoaling of the thermocline off the Gulf of developed in February moves back. In Papagayo, Papagayo, forced by Ekman pumping on the coastal waters show strong negative anomalies of side of the Papagayo wind jet. Upwelling Chla until the end of December, when the first of low-temperature and nutrient-rich waters pro- eddy developed (AP20dec98). The following motes phytoplankton growth and Chla accumula- strong positive anomalies correspond to the tion. However, the position of the CRD does not formation of the other eddies, with the exception always coincide with Chla increase, which is more of the peak at the end of February, which is related related to mixing by the wind jet itself (Fiedler, to an increase on Chla concentrations in AP29- 2002). In order to find the relationship between jan99. Offshore, at the point located at the average SST and Chla distribution of both the Gulf of position of the CRD, negative anomalies of SST Tehuantepec and Papagayo and infer the influence developed at the end of December when the CRD of the CRD in the latter, we analyzed time series deepened. Fromthen on, cycles of negative and for those points indicated in Fig. 1. Time series for positive anomalies developed and were followed anomalies of Chla concentration and sea-surface by a slight increase on Chla that attained positive temperature, which were calculated using as values fromthe middle of February, when the reference the mean of the entire series, are shown thermocline shoaled near the coast (see Fig. 3) and in Fig. 7. The Gulf of Tehuantepec exhibits a the Papagayo wind jet was stronger. Conse- strong variability, with higher concentrations in quently, long jets of low SST and high Chla are December and January. Short but strong events of observed. negative anomalies for SST can be observed in Results of cross-correlation analysis (Table 2) various occasions, and in some occasions they are confirmthe inverse relationship between SST and followed by positive anomalies of Chla. There is Chla. Although the correlation coefficients are not an offshore decrease on SST fromNovember to high, they are significant (95% of confidence level January, which corresponds to the extension of the and 145 degrees of freedom) and indicate that in jet fromthe coastal region, but variability on Chl a most situations, Chla increases following SST is smaller. In March, there is an increase on SST lowering with no observable lag. The exception ARTICLE IN PRESS

A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600 595

Fig. 5. SeaWiFS LAC images of chlorophyll-a concentration (mg mÀ3) indicating identified eddies. Eddies were labeled according to type (anticyclonic, A or cyclonic, C), origin (Tehuantepec, T or Papagayo, P) and date of generation. was the point in Tehuantepec, where the highest We also calculated the anomalies of wind speed correlation coefficient had a lag of 1 day. These at Huatulco Airport, using as reference the mean low, although significant, values can be associated of the entire series. It is possible to observe (Fig. 7) to differences in spatial resolution of images (1.1 the gradual increase on wind speed fromNovem- and 9 km) and differences on time of image ber to December, when short bursts of stronger retrieving between sensors. Besides, the presence winds become more frequent, while the strongest of clouds also could influence these results. values are observed during January and February. However, at the time the interpolation process We looked for some relationship between these was performed, no more than five continuous days data and the date of generation of cyclonic and were interpolated. The interpolation process cho- anticyclonic, eddies but we did not have any sen explore the relationship between two variables satisfactory result. However, it is possible to without fitting a specific model (such as a straight observe that the period of strongest winds line or predefined distribution) and it yields a correspond to the period of higher Chla concen- better and clearer picture of the relationship tration (January) at the inshore station (Gulf of between the x and y variables (Cleveland and Tehuantepec). A more detailed analysis was done Devlin, 1988; Chambers and Hastie, 1991). through a cross-correlation between time series of ARTICLE IN PRESS

596 A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600

Fig. 6. SeaWiFS LAC images of chlorophyll-a concentration (mg mÀ3) showing identified eddies. Eddies were labeled according to type (anticyclonic, A or cyclonic, C), origin (Tehuantepec, T or Papagayo, P) and date of generation. wind speed and time series of Chla concentration could be seen on images and time series analyzed and SST, but correlation coefficients were very low in this study, and the subsequent higher concen- and were not significant. trations of Chla can extend farther offshore, probably as a result of nutrient enrichment. However, no significant relationship was found 4. Discussion and conclusions between wind events and eddy generation or between wind speed and Chla concentrations in The generation of eddies in the Gulf of time series. This fact could be the result of the Tehuantepec is explained as the result of the stress origin of wind data, which were obtained froma produced by the strong wind events that act over shore station. Besides, it is necessary to consider the ocean pushing surface waters away fromthe that during the months analyzed in this study coastal region. Entrainment induced by vertical (November 1998–March 1999) El Nin˜ o Southern shear in the offshore jet penetrates deeper than Oscillation’s cold phase (La Nin˜ a) affected the elsewhere, producing vertical mixing with subse- Pacific, caused mainly by the intensification of quent surface cooling and nutrient enrichment (Murtugudde et al., 1999). This (Trasvin˜ a et al., 1995). The gradual surface cooling phenomenon is characterized by unusually cold ARTICLE IN PRESS

A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600 597

Tehuantepec Papagayo 8 8 6 (A) 6 (D)

s

e 4 4

i

l a 2 2

m

o 0 0

An -2 -2 -4 -4

8 8 (B) 6 6 (E)

s 4 e 4

i

l a 2 2

m o 0 0

An -2 -2 -4 -4 N DJ FM 8 (C) 6

4

2

0

Wind Anomaly -2

-4 ND JFM

Fig. 7. Time series of anomalies of chlorophyll-a concentration (mg mÀ3, dashed line) and SST (1C, continuous line) for those points indicated in Fig. 1, fromDecember(D) to March (M). (A) Tehuantepec inshore, (B) Tehuantepec offshore, (D) Papagayo inshore, (E) Papagayo offshore. The anomaly of wind speed (m sÀ1) for Tehuantepec (Huatulco Airport) is also shown (C).

Table 2 necessary to introduce nutrients that promote the Results of cross-correlation analysis between time series of sea- increase on Chla concentrations, which also could surface temperature and chlorophyll-a concentration taken fromthe points indicated on Fig. 1 be the reason of the poor correlation on statistics. However, in order to confirm these mechanisms, Lag R Lag R more observations and comparisons with other time periods (El Nin˜ o and normal years) are Tehuantepec 1 À0.28 Papagayo 0 À0.23 T—offshore 0 À0.30 P—offshore 0 À0.34 needed. The axis of the Tehuantepec wind jet almost Correlation coefficients (R) are significant (a=0.05) and Lag is always turns anticyclonically toward the west. the time interval (days) related to the highest R in the time With increasing distance fromthe coast, the radius series. of curvature of the jet axis becomes progressively larger than the inertial radius, indicating the ocean temperatures in the equatorial Pacific, and importance of geostrophic adjustment of the jet along the coast of Mexico it promotes negative (Chelton et al., 2000b). In the ocean, water masses anomalies of sea-surface height (Zamudio et al., respond forming anticyclonic and cyclonic eddies 2001, Caldero´ n-Lopez, 2002). These negative that can be easily observed on ocean-color images. anomalies were related to a lower thermocline The period analyzed in this study presented more and positive anomalies of Chla concentration eddies than those observed in previous studies; (Caldero´ n-Lopez, 2002). For this reason we think especially cyclonic eddies, considering that we that lower wind conditions can produce the mixing analyzed only one season. In fact, for each ARTICLE IN PRESS

598 A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600 anticyclonic eddy, one to four cyclonic eddies were with the others. In fact they presented an average generated (see Table 1). The origin of most of them speed fluctuating between 3.99 and 15.78 cms À1 was at 951W, which corresponds to the average where the values above 12 cms À1 fall within the position of the wind jet. Anticyclonic eddies are range of speed reported on previous studies generated on the anticyclonic (westward) side of (Stumpf and Legeckis, 1977; Hansen and Maul, the wind jet while cyclonic eddies on the cyclonic 1991; Willett, 1996;Mu¨ ller-Karger and Fuentes- (eastward) side. The generation of anticyclonic Yaco, 2000; McClain et al., 2002). eddies west of the main wind jet was frequently Papagayo anticyclonic eddies have a common observed on previous studies fromfield, SST and origin, and after their generation they travel on predictions made by numerical models, but the following the main flux of the Costa Rica Coastal counterpart of a cyclonic eddy generation was not. Current (CRCC), which then turns offshore to join Indeed, the question of why significant cyclonic the North Equatorial Current (NEC). This beha- eddies do not spin up in this area has been raised vior is clearly followed in SeaWiFS images and (Barton et al., 1993). However, our results show suggests an influence of the CRCC on eddy that the frequency of cyclonic eddy formation is propagation. In fact, lateral boundary friction in higher than previously thought and that their the CRCC might provide a secondary source of lifetime can be even longer. The use of both ocean- anticyclonic vorticity (Hansen and Maul, 1991), color images and daily temporal resolution instead and the observed translation may be explained by of SST alone improved our capability of identify the eddy being embedded in the westward moving and follow the propagation of these eddies. NEC. An example of how far the influence of these Besides, the generation and fate of the smaller eddies can be followed in open ocean was cyclonic eddies around the periphery of the anti- presented by Boyle et al. (1981), who described a cyclonic eddy indicate their importance in the patchy distribution of near-surface waters that onshore-offshore exchange of energy and biologi- contain enriched concentrations of copper, ex- cal material, a fact not considered previously. tending 4000 kmor morewest of Costa Rica. They It is interesting to note that the anticyclonic attributed this enrichment to coastal processes and eddy AT14dec98 has an elongated shape, and it is speculated that the intermittency was probably located closer to the coast (Fig. 5). The same due to variations of the ocean surface circulation. feature was observed by Mu¨ ller-Karger and The Papagayo anticyclonic eddies that have been Fuentes-Yaco (2000) on their images from CZCS described herein are a likely mechanism for the (eddy 80LINDA9T+), suggesting that it is also a transport of coastally enriched water offshore. The recurrent feature, like those eddies fromthe same suggestion was presented by Mu¨ ller-Karger western region of the Gulf. This shape can be and Fuentes-Yaco (2000) regarding the impor- explained as a result of friction due to their tance of both the Gulf of Papagayo and Tehuan- position over the continental shelf. In fact, its tepec as source of mass, energy, nutrients, plants generation begins as a filament that moves off- and animals from the ocean margin to the ocean shore and starts spinning towards the left with a interior. The origin of AP14dec98, however, was circular shape; as the spin starts closing in, the north of Papagayo (Fig. 4), presenting lower shape elongates and moves closer to the coast concentrations of Chla and a northwest propaga- where it remains in the same position until it tion to the Gulf of Tehuantepec. In fact, the region disappears. This behavior can be the result of the between Tehuantepec and Papagayo seems to be interaction with the continental shelf, which affects an intermediate zone where winds and surface the speed differently in different parts of the eddy, currents are weaker than along the main axes of (Mu¨ ller-Karger and Fuentes-Yaco, 2000). More the Tehuantepec and Papagayo jets, and it is observations would be necessary in order to characterized by the presence of a series of confirmthe recurrence of this eddy and the causes meanders, filaments and eddies, as AP14dec98. of its behavior. Moreover, this eddy did not Anticyclonic eddies in Tehuantepec are observed present a significant propagation in comparison since November, whereas those at Papagayo start ARTICLE IN PRESS

A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600 599 at the end of December. This behavior is a result of propagation even though some parts of it were the effect of winds on the oceanic system. The covered. For this reason, we think that the Tehuantepec wind jet is triggered by sub-tropical identification of eddies was not seriously affected cold, winter high-pressure systems originating in by clouds during the period analyzed in this study. the Gulf of Mexico, which force strong winds Ocean-color observations give relevant and new through the low-altitude passes in the Isthmus information about the evolution and frequency of of Tehuantepec and Nicaragua’s lake district eddy generation and propagation in the Tropical (see Fig. 1). These systems may move farther to Pacific Ocean. This study focuses on a short period the south–east on the Caribean Sea generating the of time, with the intention of analyzing the largest Papagayo jet sequentially after a Tehuantepec number of sequential images without clouds event, as well as fromintensified trade-winds during the lifetime of the sensor SeaWiFS and (Chelton et al., 2000a). Therefore, there is a lag coinciding with the period when eddies are more between the strong winter winds on both regions frequent (boreal winter). In doing so, we were (Fiedler, 2002). The monthly evolution of both able to track the evolution of more eddies than systems can be observed on Fig. 3, on which it was previous studies. However, it is clear that we still possible to infer the influence of the Costa Rica have several unanswered questions. The influence Dome (CRD). High Chla concentrations off of La Nin˜ a is one of them, and this subject will Papagayo are generated by a combination of jet- be addressed in future studies. In addition, to driven mixing near the coast and the shoaling of better quantify the relationship between wind the thermocline inshore (CRD). The CRD is far events and chlorophyll concentration, the use of fromthe coast fromNovember to January, but ocean-wind data (as those fromQuikSCAT) seems despite this high concentrations of Chla develop necessary. and the generation of anticyclonic eddies are observed. This suggests the importance of wind mixing and/or Ekman pumping for the introduc- Acknowledgements tion of nutrients in the photic zone instead of the influence of the CRD. The approximation to the The authors thank Universidad Autono´ made coast of the CRD (February/March) coincides with Baja California (UABC) and Facultad de Ciencias lower SST values over the region as a result of the Marinas (Ensenada, Mexico) for their support. We lowering of the thermocline. However, high con- thank P. Fiedler for geographical coordinates of centrations of Chla related with eddy generation the Costa Rica Dome, who besides Mati Kharu are also observed north to the position of the provided pertinent criticism, helping to improve CRD, which confirms the observation of Fiedler this paper. (2002) that phytoplankton bloomdevelopmentis more related to the wind jet than to the thermocline lowering due to the CRD. In fact, it has been stated References that Papagayo anticyclonic eddies erode the CRD while they may be reinforced by cyclonic eddies Barton, E.D., Argote, M.L., Brown, J., Kosro, P.M., Lavin, (Mu¨ ller-Karger and Fuentes-Yaco, 2000; Fiedler, M., Robles, J.M., Smith, R.L., Trasvin˜ a, A., Velez, H.S., 2002). However, we did not observe the generation 1993. Supersquirt: dynamics of the Gulf of Tehuantepec, Mexico. Oceanography 6, 23–30. of cyclonic eddies at Papagayo, which could be the Boyle, E.A., Huested, S.S., Jones, S.P., 1981. On the distribu- result of cloud cover in this zone where less images tion of copper, nickel, and cadmium in the surface waters of were taken (see Fig. 2). In fact, the image on Fig. 2 the North Atlantic and North Pacific oceans. Journal of gives an idea about the influence of clouds on the Geophysical Research 86, 8048–8066. entire study area and makes it clear that their Caldero´ n-Lopez, J.M., 2002. Clorofila-a como indicador de aguas productivas y su asociacio´ n con altimetrı´ aenel influence is stronger on the area of Papagayo and Oce´ ano Pacifico Oriental. Undergraduation Thesis. Uni- the Costa Rica Dome. However, after the identi- versidad Auto´ noma de Baja California, Ensenada, Mexico, fication of an eddy, it was possible to track its 96pp. ARTICLE IN PRESS

600 A. Gonzalez-Silvera et al. / Deep-Sea Research II 51 (2004) 587–600

Chelton, D.B., Freilich, M.H., Esbensen, S.K., 2000a. Satellite E1 Golfo Tehuantepec y a´ reas adyacentes: variacio´ n observations of the wind jets off Central America. Part I: espacio-temporal de pigmentos fotosinte´ ticos derivados de case studies and statistical characteristics. Monthly Weather sate´ lite. Ciencias Marinas 23, 329–340. Review 128, 1993–2018. McClain, C.R., Christian, J.R., Signorini, S.R., Lewis, M.R., Chelton, D.B., Freilich, M.H., Esbensen, S.K., 2000b. Satellite Asanuma, I., Turk, D., Dupouy-Douchement, C., 2002. observations of the wind jets off Central America. Part II: Satellite ocean-color observations of the tropical Pacific regional relationships and dynamical considerations. Ocean. Deep-Sea Research II 49, 2533–2560. Monthly Weather Review 128, 2019–2043. Mu¨ ller-Karger, F., Fuentes-Yaco, C., 2000. Characteristics of Chambers, J.M., Hastie, T.J., 1991. Statistical Models. Chap- wind-generated rings in the eastern tropical Pacific Ocean. man & Hall/CRC, London, pp. 309–376. Journal of Geophysical Research 105, 1271–1284. Cleveland, W.S., Devlin, S.J., 1988. Locally-weighted regres- Murtugudde, R.G., Signorini, S.R., Christian, J.R., Busalacchi, sion: an approach to regression analysis by local fitting. A.J., McClain, C.R., Picaut, J., 1999. Ocean color Journal of the American Statistical Association 83, 596–610. variability of the tropical Indo-Pacific basin observed Fa¨ rber-Lorda, J., Lavı´ n, M.F., Guerrero-Ruiz, M.A., 2003. by SeaWiFS during 1997–1998. Journal of Geophysical Effects of wind forcing on the trophic conditions, zoo- Research 104, 18351–18366. plankton biomass and krill biochemical composition in the O’Reilly, J.E., Maritorena, S., Mitchell, B.G., Siegel, D.A., Gulf of Tehuantepec. Deep-Sea Research II, this issue [doi: Carder, K.L., Garver, S.A., Kahru, M., McClain, C., 1998. 10.1016/j.dsr2.2004.05.022]. Ocean color chlorophyll algorithmfor SeaWiFS. Journal of Fiedler, P.C., 2002. The annual cycle and biological effects of Geophysical Research 103, 24937–24953. the Costa Rica Dome. Deep-Sea Research I 49, 321–338. Stumpf, H.G., Legeckis, R., 1977. Satellite observations of Fiedler, P.C., Philbrick, V., Chavez, F.P., 1991. Oceanic mesoscale eddy dynamics in the eastern tropical Pacific upwelling and productivity in the eastern tropical Pacific. Ocean. Journal of Physical Oceanography 7, 648–658. Limnology and Oceanography 36, 1834–1850. Trasvin˜ a, A., Barton, E.D., Brown, J., Velez, H.S., Kosro, Hansen, D.V., Maul, G.A., 1991. Anticyclonic rings in the P.M., Smith, R.L., 1995. Offshore wind forcing in the Gulf eastern tropical Pacific Ocean. Journal of Geophysical of Tehuantepec, Me´ xico: the asymmetric circulation. Research 96, 6965–6979. Journal of Geophysical Research 100, 20649–20663. Lavı´ n, M.F., Robles, J.M., Argote, M.L., Barton, E.D., Smith, Willett, C.S., 1996. A study of anticyclonic eddies in the eastern R., Brown, J., Kosro, M., Trasvin˜ a, A., Ve´ lez, H.S., Garcia, tropical Pacific Ocean with integrated satellite, in situ, and J., 1992. Fı´ sica del Golfo de Tehuantepec. Ciencia y modeled data. Ph.D. Thesis, University of Colorado, 127pp. Desarrollo 18, 97–108. Zamudio, L., Leonardo, A.P., Meyers, S.D., O’Brien, J.J., Lluch-Cota, S.E., Alvarez-Borrego, S., Santamaria del Angel, 2001. ENSO and Eddies on the Southwest Coast of Mexico. E.M., Muller-Karger, F.E., Herna´ ndez-Vazquez, S., 1997. Geophysical Research Letters 28, 13–16.