Submitted to Remote Sensing of Environment May 2000

Transcription

1 Submitted to Remote Sensing of Environment May 2000 Patterns of Seasonal Dynamics of Remotely Sensed Chlorophyll and Physical Environment in the Newfoundland Region Yakov D. Afanasyev 1, Nikolay P. Nezlin 2, and Andrey G. Kostianoy 2 1 Memorial University of Newfoundland, St. John's, Canada 2 P.P.Shirshov Institute of Oceanology, Moscow, Russia Corresponding author: Yakov D. Afanasyev Department of Physics and Physical Oceanography Memorial University of Newfoundland, A1B 3X7, St. John's, Canada yakov@physics.mun.ca 1

2 ABSTRACT The patterns of seasonal variation of surface plant pigment concentration (Chl) in the Newfoundland Region were studied using the remotely sensed data from CZCS ( ) and SeaWiFS radiometers (from September 1997 to October 1999). Sea Surface Temperature (SST) data obtained from AVHRR radiometers and Sea Surface Height (SSH) data obtained from TOPEX/POSEIDON altimeter were used then to interprete the observed patterns in terms of physical factors which influence the growth of phytoplankton. Stable seasonal cycles of both SST and Chl were observed in all parts of the region under study (Labrador Current, Newfoundland Bank, Flemish Pass, frontal zone between Gulf Stream and Labrador Current). The SST values in summer season during two years under study (September 1997 October 1999) were up to 3 C higher as compared with climatically averaged values. The seasonal pattern of Chl in Labrador Current zone was typical of Arctic regions (one maximum in summer), in Gulf Stream zone it was typical of subtropical regions (smoothed maximum during winter), and in between these zones it was typical of mid-latitudes (two maxima in spring and autumn). Over the Grand Newfoundland Bank the seasonal pattern had one spring maximum, typical of shallow regions. The patterns of seasonal phytoplankton cycles resulted mainly from the meteorological factors influencing water stratification; the latter seems to be a crucial factor in either light or nutrient limitation of phytoplankton growth. 2

3 1. INTRODUCTION The region under study (40 53ºN, 40 70ºW, Figure 1) is located in the northwestern part of the Atlantic Ocean, near the eastern coast of Canada. The global-scale pattern of ocean currents makes this region especially interesting from the point of view of plankton ecology and biogeography. In particular, this zone appears to be under strong influence of both the cold Labrador Current flowing from the north and the warm Gulf Stream flowing from the southwest. The physical and biological gradients are extremely pronounced in this region, and distinct features typical to different geographical zones occur over the small area. From the point of view of biogeography several distinct regions (so-called provinces) sharply different in terms of both physical environment and biology can be identified within the Newfoundland Region. Sathyendranath et al. (1995) proposed the classification of the waters of North Atlantic in terms of their biological productivity. Their classification of domains and provinces is based on differences in physical environment, which influence regional phytoplankton dynamics. According to this classification, the relatively small region we study contains four biogeographic provinces from three domains: - the Arctic Province from the Polar Domain, bounded by the oceanic Polar Front; - the Gulf Stream Province, and - the North Atlantic Drift Province from the West-Wind Domain, which comprise the flow of the Gulf Stream and the North Atlantic Current; - the North-West Atlantic Province from the Coastal-Boundary Domain, which includes shelf and slope regions and adjacent waters. The concept of dynamic biogeography implies that the unique biogeochemical provinces have seasonally-predictable algal growth dynamics (Platt and Sathyendranath, 1988; Platt et al., 1991). The basic concept is that the regional differences in algal cycles and other aspects of pelagic ecology must reflect differences in mixed-layer dynamics. Hence, the partitioning of the ocean must reflect the regional seasonal variation of the physical factors governing the dynamics of the mixed layer. Studies of seasonal patterns of variations of phytoplankton biomass therefore require the analysis of the information on both physical and biological processes of comparable spatial and temporal scale. The remotely sensed data collected since late 70's provide oceanographers with a large volume of information on the state of the surface of the World Ocean although these data are not 3

4 always accurate enough (it concerns especially the surface chlorophyll concentration derived from water color measurements). The most important remote sensing instruments are infrared sensors collecting data on Sea Surface Temperature (SST), altimeters obtaining anomalies of Sea Surface Height (SSH), and the sensors working within the visible light band. The information collected by the visible light sensors is especially valuable for biologists, because it characterizes the concentration of plant pigments in the most productive surface ocean layer. The surface chlorophyll concentration is, in its turn, an important quantity which can be used for predictions of the type of vertical distribution of chlorophyll (Morel and Berthon, 1989) and other properties of the pelagic community, including primary production rate and biomass of different trophic groups of zooplankton (Vinogradov et al., 1992, 1997, 1999). Recent development of new sources of remotely sensed bio-optical data (in particular SeaWiFS radiometer working since autumn 1997) enables the analysis of seasonal variations of phytoplankton biomass in different regions of the World Ocean. Previous studies of seasonal cycles of remotely sensed phytoplankton biomass in many regions of the World Ocean (Banse and English, 1994; Longhurst, 1995, and many others) were based on the data of CZCS (Coastal Zone Color Scanner) radiometer, collected during However, after launch of the OrbView-2 platform with SeaWiFS (Sea-viewing Wide Field-of-view Sensor) in August 1997 the amount of remotely sensed bio-optical data significantly increased. Nowadays we can combine the remotely sensed Chl data with simultaneously measured satellite data on physical environment (SST, SSH, meteorological information, etc.). The main goal of the present study is to analyze the principle features of seasonal variations of phytoplankton in the Newfoundland Region during a recent two year period on the basis of newly obtained remotely sensed data, to compare them with the obtained before climatilogical data, and to interpret the peculiarities of spatial and temporal variability of phytoplankton distribution in terms of the effects of hydrographic and meteorological fields (SST, ocean currents, wind). In Section 2 we describe the remotely sensed data we use for analysis. In Section 3 we give a brief review of the bathymetry of the Newfoundland Region and the pattern of currents. In Section 4 we describe the general patterns of seasonal variations of Chl in the region under study (Section 4.1) as well as the peculiarities of Chl and physical environment in (Section 4.2); in these sections we also make an attempt to explain the observed variations of 4

5 Chl with the peculiarities of water circulation. The seasonal variation of SST and Chl in four small sub-regions, which characterize the biogeographical provinces, are described in Section 4.3. In the Discussion (Section 5) we make an attempt to interprete the general pattern of seasonal variations of Chl using the simple model, which correlates water column stratification with phytoplankton. 2. REMOTELY SENSED DATA USED FOR ANALYSIS All the remotely sensed data used in this study were obtained via the Internet from their centers of processing and dissemination. The following data were used: MCSST Sea Surface Temperature (MCSST) data. The data collected by AVHRR (Advanced Very High Resolution Radiometers) aboard the NOAA-7, -9, -11 and -14 polar orbiting satellites and processed using the Multi-Channel Sea-Surface Temperature algorithm (McClain et al., 1985) in the Physical Oceanography Distributed Active Archive Center at the Jet Propulsion Laboratory (JPL PODAAC Product 016). The data have been collected since November 1981, averaged weekly and interpolated (without missing values) over global equalangle grids of spatial resolution of 2048/360 pixels (~18.5 km) per degree of longitude and latitude. The nominal accuracy is 0.3 C. We used only the data collected during night-time (descending pass) to avoid the short-period SST variations resulting from heating of thin surface layer. These data were used to produce weekly maps of SST distribution. Pathfinder Sea Surface Temperature. The data were processed within the scope of NOAA/NASA AVHRR Oceans Pathfinder Project in JPL. The sea surface temperature (SST) data were derived from the AVHRR Radiometers (see above) using the enhanced (as compared with MCSST) nonlinear algorithm (Walton, 1988). We used daily data for descending pass (night-time) on global equal-angle grids of 4096 pixels/360 degrees (~9.28 km resolution). The best SST data, i. e., only the highest quality pixel values were used. The versions interim V4.1 ( ) and V4.1 (1997) were used to derive absolute SST values for time series analysis. Climatically averaged Sea Surface Temperature. The data were prepared on the basis of the Pathfinder Sea Surface Temperature data in JPL. The details of the process of smoothing and averaging are described in (Casey and Cornillon, 1999). We used the pentad (five-day) climatologies for at ~9.28km resolution to calculate SST anomalies from Pathfinder SST values. 5

6 TOPEX/POSEIDON Sea Surface Height. We used the gridded sea level anomaly heights, which have been measured by the TOPEX/POSEIDON satellite altimeter since The data were processed in the University of Texas at Austin Center for Space Research (UT/CSR). In this study the sea level anomaly is defined as the deviation of the sea surface away from the 4-year ( ) averaged mean surface. The altimeter observations were averaged into 1 degree by 1 degree grids and smoothed with a rectangular Gaussian squared filter over 1000 km in longitude and 400 km in latitude, with a roll-off of 400 km (Chambers et al., 1997). The topography has an estimated accuracy of cm. The data processing and error analysis is described in detail in (Tapley et al., 1994). We used these data to estimate anomalies of the main currents during For this purpose we calculated the slope (absolute value and aspect) values of SSH anomalies in each grid cell using the method of Terrain modeling. This method calculates the steepest slope and the direction of steepest slope (the gradient direction) at each grid node. The resulting values were mapped in the form of arrows of size proportional to SSH anomaly slope and directed 90 clockwise from the aspect of maximum slope (see Figures 2, 6 8). The resulting images indicate not the pattern of ocean currents, but the anomalies of the currents, i. e., the deviations from the multi-year averaged pattern. NCEP Wind data. We used zonal and meridional wind speed values at 10 m from SeaWiFS near-real time ancillary data set. The data were obtained from the National Center for Environmental Prediction (NCEP) and processed in GSFC SeaWiFS Data Processing Center. The data were interpolated to Equidistant Cylindrical images of 1-degree latitude/longitude resolution. The temporal coverage is from March 1997, temporal resolution 6 12 hours. The zonal and meridional wind speeds were monthly averaged to estimate the differences between winds dominating during the period under study ( ) and climatological winds (averaged over ) obtained from the same source. Coastal Zone Color Scanner (CZCS). The data were collected during by CZCS (Coastal Zone Color Scanner) radiometer (Hovis et al., 1980) on the Nimbus-7 satellite. It was collecting information from November 1978 to June The water color was measured at the bands 443, 520, 550, 670, and 750 nm. More than 60,000 images were processed at Goddard Space Flight Center (GSFC, NASA) and summarized as climatic monthly data averaged over the entire period of observations. These data were stored on CD's in the form of regular grids of 6

7 resolution, equivalent to ~18.5 km on equator. The structure of this archive is described in (Feldman et al., 1989). We used these monthly climatological data to analyze the general regularities of plant pigments concentration dynamics in the region under study. Sea-viewing Wide Field-of-view Sensor (SeaWiFS). The ocean color data ( ) used in this study were produced by the SeaWiFS Project at Goddard Space Flight Center. The data were obtained from the Goddard Distributed Active Archive Center under the auspices of NASA. Use of this data is in accord with the SeaWiFS Research Data Use Terms and Conditions Agreement. The OrbView-2 platform with SeaWiFS radiometer was launched August 1, Data acquisition commenced on September 20, SeaWiFS acquires approximately 15 pole-topole orbital swaths of data per day, and approximately 90% of the ocean surface is scanned every two days. The band center wavelengths (in nm) are 412 (violet), 443 (blue), 490 and 510 (blue-green), 555 (green), 670 (red), 765 and 865 (near IR). The collected data are regularly processed by GSFC within two weeks after measurements. They are available via the Internet for authorized SeaWiFS users. In this study the Level 3 Standard Mapped data were used (global grids of the resolution similar to Pathfinder SST). The values of chlorophyll a and CZCS-like pigment concentration were used. The algorithms of optical data processing and calculating plant pigment concentrations are given in (O'Reilly et al., 1998). We used for analysis both chlorophyll a and CZCS-like pigment concentrations (named according to DAAC terminology CHLO and CPIG). The monthly averaged maps of "phytoplankton" distribution were produced in terms of CZCS-like pigment concentration to enable comparing them with CZCS data obtained in The seasonal variations in the selected regions (see below) were analyzed in terms of chlorophyll a concentration. The shading scale on Figure 5 was selected according to the classification of the productivity zones of the World Ocean (Vinogradov et al., 1992, 1997). According to this classification, waters of surface CZCS-derived plant pigment concentrations were named: <0.15 mg m -3 oligotrophic, mg m -3 mesotrophic, mg m -3 eutrophic, and >1 mg m -3 hypertrophic waters; each "gradation" has the specific principle ecosystem characteristics (see Discussion). Data processing. The data grids were processed using a software (DDNPN) specifically designed for this purpose (Nezlin et al., 1999; Nezlin, 2000). DDNPN can be used on personal 7

8 computers in Windows environment and enables the user to work with files containing SST, SSH, CZCS, SeaWiFS data. The data of SST (Figures 4, 6 8) and Chl (Figures 5 8) were smoothed using a cosinefilter of 50 km radius. The purpose of smoothing was both to remove small-scale variability and to fill small gaps between the grid points containing valid data. The statistical distribution of surface pigment concentration is evidently asymmetric (Banse and English, 1994); therefore, the Chl values were log-transformed before smoothing and exponentially transformed after smoothing. These procedures are convenient in processing ecological characteristics; the arithmetic means calculated from log-transformed biomass values are closer to modal values. Four small regions (2 degrees latitude x 3 degrees longitude) were selected for timeseries analysis of seasonal variations of SST and Chl (Figure 1, Table 1). These regions were considered to be typical to the main natural zones, located from north to south (Longhurst, 1995). Region 1 was selected in the zone of influence of the Labrador Current (the Arctic Province according to classification of Sathyendranath et al., 1995), Region 4 being in the zone of influence of the Gulf Stream (the Gulf Stream Province). Regions 2 and 3 were selected in the zone of the Grand Banks (the North-West Atlantic Province from the Coastal-Boundary Domain) to emphasize the differences between the shallow zone of the Grand Bank itself and the deeper area of the Flemish Pass. The arithmetic mean SST values were calculated for each region from each daily image. Because the seasonal variations of SST are extremely pronounced we operated with SST anomalies rather than SST absolute values. The SST anomalies were estimated by subtracting climatic means from the actual SST values. The climatic means were estimated from pentad climatologies for for each 5-day period and applied to the middle (3-rd) day of each 5-day period. The climatic SST values for other days were estimated using the method of linear interpolation. For the analysis of the chlorophyll concentrations, we calculated the medians instead of arithmetic means. This non-parametric statistic is preferable in this case due to sharply asymmetrical distribution of surface chlorophyll values (Banse and English, 1994). To emphasize the patterns of seasonal variations of both SST anomalies and surface chlorophyll concentrations the time series data were smoothed twice using moving average. First, the smoothing window of 30 points was selected (to reveal the general pattern of seasonal 8

9 variation); then the smoothing window was equal to 10 points to produce the resulting curve. Chlorophyll values were log-transformed before smoothing and exponentially transformed after smoothing. 3. REVIEW OF PRINCIPLE FEATURES OF BATHYMETRY AND SURFACE CIRCULATION IN THE NEWFOUNDLAND REGION The shelf occupies a significant part of the region under study (Figure 1). The wide (>200 km) and deep (from 200 to 500 m) Labrador Shelf occupies the northern part of the region. The Grand Newfoundland Banks (depth <100 m) are in the center and shelf of Nova Scotia Peninsula is located in the southwest part of the region. The Grand Bank is separated from the Flemish Cap bank by the relatively narrow and deep (>1000 m) Flemish Pass. Steep continental slope surrounds the shelf regions. The general pattern of water circulation is evident from the map in Figure 2 drawn on the basis of sea surface topography calculated from annual climatological data (Levitus, 1982) and referenced to 2000 m. The arrows indicate the direction and intensity of the main baroclinic flows. The size of each arrow is proportional to the slope of SSH. This pattern corresponds well to the general features of water currents described in (Petrie and Anderson, 1983; Krauss, 1986; Greenberg and Petrie, 1988; Smith and Schwing, 1991; Mertz et al., 1993; Lazier and Wright, 1993; Sheng and Thompson, 1996; Narayanan et al., 1996), although the map in Figure 2 is of rather coarse temporal and spatial resolution. The Labrador Current flows to the southeast over the continental shelves and slopes of Labrador and Newfoundland. Its total water transport is about 4 Sv (1 Sv=10 6 m 3 s -1 ) (Lazier and Wright, 1993). At Hamilton Bank (~52ºN) on the southern Labrador Shelf the current appears as two streams or branches. A small inshore stream carries ~15% of the total amount of water transported by the Labrador Current; some part of it penetrates through the Strait of Belle Isle into the Gulf of St. Lawrence. The intensity of this inflow is on the average 0.13 Sv increasing occasionally by 0.6 Sv (Petrie et al., 1988). The main stream over the upper continental slope carries about 85% of total water transport (Lazier and Wright, 1993). A strong hydrographic front is coincident with the main branch of the Labrador Current. This front separates the cold low-salinity Baffin Bay waters over the continental shelf and warmer and more saline waters of the open ocean. On the northern edge of the Grand Banks the Labrador 9

10 Current splits into three branches. The inshore (western) branch goes through the Avalon Channel and turns westward following the coastline around the Avalon Peninsula. Then it penetrates into the Gulf of St. Lawrence through the Cabot Strait along the Newfoundland coast. The countercurrent flows along the northern shore of Nova Scotia. The intensity of both currents vary seasonally within the limits of Sv with minimum inward flow in June and maximum in August, depending on freshwater runoff (El-Sabh, 1977). The main (middle) branch of the Labrador Current follows the eastern edge of the shelf through Flemish Pass, and the eastern branch flows to the north of Flemish Cap, both carrying ~2 Sv (Petrie and Anderson, 1983). The Gulf Stream flowing from the south branches at the Newfoundland Rise, the northern branch forming the North Atlantic Current. Its water transport is estimated to be 35 Sv (Krauss, 1986). The North Atlantic Current reaches its maximum speed of about 1 m s -1 at Flemish Cap. There it separates cold and fresh Labrador water from the warm and saline Atlantic water of the Gulf Stream extension area. A vigorous eddy field is observed to southeast of the frontal jet. Superimposed on this field is a mean drift toward the east of 5 10 cm s -1. The North Atlantic Current flows along the continental slope about the 4000-m isobath (Horne and Petrie, 1988). It is worth mentioning that the positions of even the main currents can vary in space and time. The position of the Gulf Stream varies from north to south by hundreds of kilometers, forming meanders and rings (Hall and Fofonoff, 1993). Smoothing and averaging of the multiyear data used to draw a map shown in Figure 2 resulted in underestimation of actual SSH heterogeneity. This explains the fact that the intensity of Labrador Current near Newfoundland appears to be greater than the intensity of the Gulf Stream in the southern part of the region. 4. RESULTS 4.1. General Patterns of Seasonal Variations of Surface Chlorophyll Concentration and Physical Environment The important factor determining the hydrological and biological features of the region under study is the westerly winds dominating in North Atlantic all year round (Figure 3). These winds are most pronounced in winter. They weaken in spring, while in summer their direction changes to southwesterly. In autumn westerly winds again dominate, the wind speed gradually increasing to a maximum in January February. 10

11 Figure 4 illustrates the seasonal variations of SST derived from climatologies (Casey and Cornillon, 1999). The pentad (five-day) images for winter (mid-january), spring (mid-april), summer (mid-july), and fall (mid-october) are given. The frontal zone between the Gulf Stream and the Labrador Current evidently manifests itself in terms of the sharpest SST gradient. Its position well corresponds to the general structure of surface circulation in the region (Figure 2). The tongue of cold water of Labrador Current directed to southeast is particularly evident during spring, summer and autumn. The warm Gulf Stream waters flow almost eastward and then turn to north-northeast, bounding the Newfoundland Rise. These two sharply different water masses are separated by the frontal zone, which may be associated with the 8 12 C isotherms in winter and spring and C in summer and autumn. It is worth mentioning that even in the averaged over long period (15 years) climatological maps the position of the main frontal zones can be clearly seen, without any significant seasonal variations. The general pattern of seasonal variations of CZCS-derived surface plant pigment concentration is illustrated in Figure 5. The climatically averaged (November 1978 June 1986) data reveal that the surface Chl concentration was rather low in January while patches of high concentrations (indicating phytoplankton blooms) were scattered along the trace of the North Atlantic Current. As for the Gulf of St. Lawrence, only the western and southern parts were covered with high Chl concentration (in the mouth of the St. Lawrence River high remote sensed Chl values can result from Gelbstoffe rather than chlorophyll concentrations). We can speculate that the coastal branch of the Labrador Current, which penetrates the Gulf of St. Lawrence through the Strait of Belle Isle, transported water poor in chlorophyll into the Gulf. In February March high pigment concentration remained in the southern and western parts of the Gulf of St. Lawrence while the bloom was observed along the shelf of Nova Scotia. In April the main part of the region under study was occupied by high pigment concentration. The waters rich in chlorophyll occupied the area between the Gulf Stream and the cold northern branch of the Labrador Current. From May to June we observe the zone of high pigment concentrations shifting northward, into the zone of influence of waters transported from the Labrador Sea. In July August this zone became less pronounced. The pigment concentration over The Grand Banks was low, with similar values observed during June August along the southern and western coast of Newfoundland. We can assume that these poor in chlorophyll waters were 11

12 transported to the Gulf of St. Lawrence through the Cabot Strait by the current mentioned above. In September Chl concentration in the northern part of the zone of interaction between the northern branch of the Labrador Current and the North Atlantic Current increased. In September October the entire area of the Gulf of St. Lawrence was covered with high Chl concentrations. In October the entire Scotian Shelf and the northern part of the Newfoundland Shelf were covered by waters rich in Chl. Similar pattern, but with lower concentrations, occurred in November. In December high concentrations were observed to the south of the banks and extended from Cabot Strait across the continental shelf and slope and beyond the 1000 m isobath Distribution of Surface Chlorophyll Concentration and Physical Environment in The analysis of monthly averaged images of SeaWiFS data processed using the algorithm similar to the one used for the CZCS data, revealed the general similarity between the patterns observed in and in A few discrepancies between the data obtained by two radiometers during these time-distant periods are, however, worth mentioning. The most evident feature was that Chl concentration during winter spring seasons of was in general higher as compared with the period of Some differences were observed between two years of SeaWiFS observations. The Chl concentration during the beginning of 1999 (January March) was evidently higher than in the beginning of 1998, the latter being significantly higher as compared with the period of It is interesting that the pattern of Chl distribution during both summer (June August) seasons of 1998 and 1999, and climatic summer months ( ) were similar including minor details. The increased Chl concentrations during winter spring seasons of seem to result from peculiarities of the hydrometeorological situation in the Northwestern Atlantic. Figures 6 8 illustrate the patterns of distribution of physical and biological properties during fall (October 1998), winter (February 1999), and spring (April 1999). The patterns of wind, ocean currents anomalies, SST and Chl in autumn winter spring seasons of and were similar, as far as the differences between these two years and the climatically averaged data (see Figures 2 5) are concerned. Below we analyze the patterns observed during 12

13 the cold period (October, February, and April) of only, implying that in the period of its general features were similar. The regular pattern of dominant westerly winds was observed in October of These winds are usually more pronounced in the northern part of the Newfoundland Region and less pronounced in the southern part (Figure 3). The pattern of wind distribution was reversed in October 1998: westerly winds were more pronounced in the southern part of the region and weaker in the northern part. This unusual pattern of wind forcing inevitably influenced the pattern of water circulation: in October 1998 the Labrador Current was weaker as compared with the mean pattern. Taking into account that the maximum speed of the Labrador Current is typically observed in October (Lazier and Wright, 1993), and the fact that the pattern illustrated in Figure 6 indicates the deviations from the general (rather than seasonal) mean pattern, we can conclude that in October 1998 the strength of the Labrador Current was much weaker than usual. This conclusion is supported by the comparison of the climatically averaged SST distribution (Figure 4) and the SST distribution in October 1998 (Figure 6). Usually the tongue of cold waters of the Labrador Current is clearly indicated by the 8 C isotherm penetrating to the zone of Flemish Pass at 47 N, 46 W (Figure 4). In October 1998 the 8 C isotherm was shifted by 2 to the north and by 3 to the west (Figure 6). Also the isotherm 10 C was significantly (about 4 ) shifted westward. The positions of other isotherms (12 C and higher) were similar to the climatically averaged pattern. The pattern of Chl distribution in October 1998 (Figure 6) also differs from the climatically averaged pattern (Figure 5). The waters with high Chl values (hypertrophic and eutrophic waters) occupied the larger area including the Flemish Pass and the entire area of the Grand Banks. In winter (February 1999, Figure 7) the wind field differed from the climatically averaged pattern (Figure 3) over the entire Newfoundland region. In the southwestern part the winds were strong and directed offshore. Over the Gulf Stream frontal zone the winds were weaker than usual. Strong southwesterly winds occurred in the northeastern part of the region under study. This unusual wind pattern influenced the water circulation mostly in the southern and eastern parts of the region. The SSH anomalies indicate that the Gulf Stream speed was weaker than usual. In the eastern part of the North Atlantic Frontal Zone two pronounced hydrological structures appeared: the anticyclonic gyre with the center at approximately 44 N, 45 W, and the cyclonic gyre with the center at 48 N, 41 W. Both gyres seem to result from the 13

14 strengthened eastern branch of the Labrador Current flowing along the northern boundary of the Grand Banks. It is worth mentioning that the anticyclonic gyre in the rectangle with boundaries of N, W seems to be a permanent feature of large-scale circulation in this region (Mann, 1967). The pattern of SST distribution in February 1999 (Figure 7) also differed from the climatic one (Figure 4). Firstly, SST was evidently higher than usual in the zone of the Labrador Current in February The position of the 0 C isotherm was shifted to the north and to the west. Nevertheless, the location of the 4 C isotherm was similar to the location of the climatic one due to the presence of the relatively stable Gulf Stream front. The tongue indicating the main branch of the Labrador Current directed southward along 50 W was as sharp as usual. Comparing this pattern with SSH anomalies we conclude that the Labrador Current in February 1999 was not weaker as compared with climatic average. The significantly higher SST near Newfoundland resulted from air temperatures, which were 4 C higher than usual in February 1999 (Table 2). The Gulf Stream Frontal Zone was meandering more than usual in February This fact seems to be a consequence of the lower speed of the Gulf Stream resulting from atypical northern offshore winds in the region of Nova Scotia and further to south-west (see Figure 7), which should cause water transport directed to southwest. The strengthened eastern branch of the Labrador Current caused the tongue of cold waters indicated by the meander of the isotherm 4 C to appear north from the Flemish Cap. The values of Chl in February 1999 (Figure 7) were higher than climatic means (Figure 5) over Scotian Shelf and to the south of the Grand Banks. The Chl pattern followed the configuration of the Gulf Stream Frontal Zone in detail. It is interesting that the outer edge of the anticyclonic gyre at 44 N, 45 W indicated by SSH anomalies, is associated with the zone of high Chl concentration. In April 1999 the wind field was similar to the pattern observed in February The absolute values of wind speed to the south of 45 N, however, were higher as compared with values for February March April is usually the season of a minimum of the Labrador Current (Lazier and Wright, 1993). We can expect therefore the water transport anomalies (Figure 8) to be directed to the northwest (opposite to the main current). Although the transport anomalies do directed this way, their relatively low intensity indicates that the Labrador Current was not weaker, but stronger than usual. This conclusion is supported by the pattern of SST distribution in April 1999 (Figure 8). At climatically averaged maps (Figure 4) the 0 C isotherm 14

15 never penetrates to the south of 50 N. In contrast, in April 1999 similar isotherm reached Avalon Peninsula (Figure 8). The same effect was observed for the 2 C and 4 C isotherms: they were further to the south and to the east. The meander of the 4 C isotherm at 50 W at Figure 8 is worth mentioning. It also indicates the tongue of cold water transported by the central branch of the Labrador Current. On the climatically averaged map (Figure 4) it is absent. The mean air temperature in April 1999 was near the value typical to a recent 7-year period (Table 2). However, in March it was significantly higher (+4 C). Hence, the lower SST could not be explained by lower heat flux from the atmosphere. Hence, the Labrador Current in April 1999 seems to be stronger than usual. The Chl distribution in April 1999 indicates the phytoplankton bloom over almost the entire region under study. The hypertrophic waters occupy the Labrador Shelf, the Gulf of St. Lawrence, the Scotian Shelf, the Grand Banks, and the area to the north of the Gulf Stream Frontal Zone, following its configuration Seasonal Variation of SST and Chl in Different Sub-Regions Sea Surface Temperature (SST). Figure 9 illustrates the seasonal variations of SST anomalies in the selected small sub-regions (2 degrees of latitude x 3 degrees of longitude) from August 1997 to July In the Region 1 (Labrador Current) the SST in spring season of 1999 (March April) was noticeably (~1.5 2ºC) lower than its climatic value. On the contrast, during summer 1998 (June August) SST evidently differed from the climatic values (+2ºC). In the Region 2 (the Grand Bank) and Region 3 (Flemish Pass) the SST anomaly was positive during almost the entire period of time. This difference was most significant (+3ºC) during summer (June August). Over the Tail of the Bank (Region 4) the SST differed from climatic values only during spring season (April May) of 1999: the SST anomaly being negative (-3ºC). In most of the regions the sharp negative SST anomalies occurred from time to time during both warm and cold periods; most often they occurred in December. The SST anomalies of about -6ºC occurred often. In the most southern region (Tail of the Bank) the negative SST anomalies exceeded 10ºC twice during the period of observation. These anomalies seem to be manifestations of the jets of cold water transported by Labrador Current to the south. They were most pronounced in the 15

16 frontal zone between the warm waters of the Gulf Stream and the cold waters of the Labrador Current (Region 4). Surface Chlorophyll Concentration (Chl). In all of the sub-regions the pattern of seasonal variations didn t differ greatly during two years of observations (From September 1997 to July 1999, see Figure 10). In the Region 1 (Labrador Current) the chlorophyll concentration increased from January until June, then it decreased, and from July to December remained stable. In the Regions 2 and 3 (the Grand Bank and the Flemish Pass) the spring maximum in March April was pronounced. It was followed by decrease in chlorophyll concentration with minimum in summer. Then Chl gradually increased by November. In the Region 3 (Flemish Pass) the autumn maximum (much less than the spring one) was observed, in the Region 2 (the Grand Bank) it was almost absent. In the southern Region 4 (Tail of the Bank) the surface chlorophyll concentration was minimal in summer and then increased during the cold season (from November till April). The seasonal variations during 1997, 1998, and 1999 didn t differ evidently. The spring maximum in 1999, however, was more pronounced in the Regions 2 and DISCUSSION In this paper we compare CZCS ( ) and SeaWiFS observations collected during the period from autumn 1997 to autumn However, numerous disadvantages of CZCS data as compared with SeaWiFS have to be kept in mind. Although the CZCS was operative for 7.5 years, most of the data were collected during first three years, and the accuracy of the latter data was worse due to the sensitivity deterioration of the instrument. Furthermore, the spatial coverage of CZCS was also much less than the spatial coverage of SeaWiFS; the region of interest is frequently cloudy; therefore, the CZCS composites are based on insufficient number of data, especially during winter months and in the northern part of the study area. That is why we derived Chl seasonal cycles from SeaWiFS not CZCS data General patterns of Chl seasonal variation The seasonal cycles of Chl observed in in small sub-regions (Figure 10) correspond well to patterns of seasonal variations of phytoplankton typical to different climatic zones of the World Ocean (see Cushing, 1975; Raymont, 1980). The most detailed summary of these cycles 16

17 based mainly on the remotely sensed data was given in (Longhurst, 1995), where 8 principle types of phytoplankton seasonal cycles were described. Four of them are associated with the open ocean (Polar, Mid-latitudinal, Subtropical, and Tropical zones); other four include coastal regions. The most important of these patterns occur in the Newfoundland region (Polar, Midlatitudinal, and Subtropical zones of open ocean, and the zone of mid-latitudinal continental shelf). In simplified form these types of seasonal cycles were also described in (Campbell and Aarup, 1992; Banse and English, 1994). These cycles can be interpreted in terms of general factors, which control the rate of phytoplankton growth. In simplified form, these variations can be explained by the model of physical control of algal growth introduced by Sverdrup (1953). This model connects the stability of upper water layer with two main factors limiting the rate of primary production: light and nutrients. On the one hand, the stability of water column within the euphotic layer results in that phytoplankton cells remain in this layer (with illumination sufficient for phytoplankton growth), and are not transported to deep dark layers. On the other hand, the pronounced pycnocline hinders upward penetration of nutrients to euphotic layer. The first kind of limitation is typical to winter seasons in high and temperate latitudes (e. g., the Arctic and the North Atlantic Drift Provinces); the second one is typical to tropical and subtropical zones (e. g., the Gulf Stream Province) and also to the summer period in mid-latitudinal zone. Certainly, there are other factors, which control the dynamics of phytoplankton, the most important being grazing of phytoplankton by zooplankton (Heinrich, 1962). However, we limit our discussion to only those physical effects controlling Chl seasonal dynamics for which the remotely sensed data are available, although other factors may also be significant. The remotely sensed surface chlorophyll concentration is linked to other basic parameters of pelagic ecosystem. This fact allows one to classify epipelagic communities based on the data on the surface chlorophyll concentration (Vinogradov et al., 1992, 1997). The linear correlation, however, can only be used effectively for identification of global-term spatial (geographic) distributions of ocean waters "rich" and "poor" in terms of productivity. The same straightforward approach based on linear correlation must be applied very carefully (if applied at all) when temporal variability within each productive zone is concerned especially in temperate and high latitudes (Vinogradov et al., 1999). The reason is that high latitudes are inhabited by 17

18 communities with evident seasonal cycle, the correlations between phytoplankton (chlorophyll concentration) and biomasses of bacteria, zooplankton, etc. being far from linear. The seasonal variation of Chl observed in the zone of influence of the Labrador Current (Region 1 in Figure 10) manifests the pattern typical to Arctic regions. The general features of this cycle were described in application to different high-latitudinal regions, including the North Atlantic (Colebrook, 1982). The productivity there is light-limited and not nutrient-limited. As a result, the phytoplankton seasonal cycle is nearly symmetric about the local irradiance maximum, the latter occurring in June July. The stability of water column in summer is induced by shallow halocline resulting from melting ice (Smith, 1991). The opposite situation is observed in the most southern Region 4 (Figure 10, Tail of the Bank), which is under the influence of the Gulf Stream warm waters. The seasonal cycle there was smooth. Higher Chl values were observed during cold season and lower values were observed during summer period. This seasonal cycle is nutrient-limited rather than light-limited. The winter increase in productivity is forced by the response of pycnocline stability to seasonality in wind stress, regularly occurring in the zone of the Gulf Stream (Longhurst, 1995, see also Figure 3). It is worth mentioning that Winn et al. (1995) explained the winter increase of surface Chl concentration in subtropical oligotrophic zone by photoadaptation of phytoplankton in response to decreased average mixed-layer light intensity rather than a change in phytoplankton biomass. This mechanism can be important in oligotrophic zone; however, the Chl values in the region we study are much higher. Hence, we prefer the explanation of winter increase of Chl concentration with enhanced nutrient supply to the euphotic zone. The patterns of Chl variations observed in Regions 2 and 3 (The Grand Bank and Flemish Pass) include both types of limitation. In winter the Chl concentration is rather low due to light limitation resulting from the deep excursion of mixed layer while in summer the phytoplankton production being limited by lack of nutrients resulting from the stability of water column. This model is typical both to the temperate latitudinal zone of open ocean, especially to the North Atlantic (see Longhurst, 1995), and to mid-latitude continental shelves (Riley, 1947; and many others). This model includes either one pronounced spring maximum resulting from formation of seasonal thermocline, or two maximums in spring and autumn. The latter maximum is explained either by breakdown of stratification which enhances nutrients supply from the layer below the summer pycnocline, or by decrease of grazing pressure on 18

19 phytoplankton by herbivores (mainly Calanus finmarchicus) which descend to greater depth to overwinter (Colebrook, 1982). Over the Flemish Pass (Region 3) the autumn Chl maximum was more pronounced, as compared with The Grand Bank (Region 2, Figure 10). The reason for this behavior seems to be in the fact that the Flemish Pass is much deeper (>1000 m) than the shallow zone of The Grand Bank (<100 m). Two effects related to the water depth are worth mentioning: firstly, the herbivores perform both diel and ontogenetic vertical migrations in deep North Atlantic regions rather than over the shelves (Hays, 1996; Planque et al., 1997); secondly, the hydrological conditions in shallow regions are more complex than in the deep ocean regions. It is worth mentioning that Longhurst (1995) was unable to reveal the pattern of seasonal variations of Chl in mid-latitude coastal regions on the basis of CZCS data. SeaWiFS data collected during the last two years ( ) illustrate this pattern in detail Chlorophyll blooms in Below we discuss the differences between CZCS-derived multi-year ( ) monthly averaged seasonal patterns of phytoplankton and the data observed in and make an attempt to explain these differences in terms of hydrological situation. However, we have to keep in mind the drawbacks of CZCS observations as compared with SeaWiFS. The chlorophyll concentration during spring blooms of 1998 and 1999 was higher, as compared with the data observed in The intensity of phytoplankton bloom during autumn (Figure 6) of 1998, winter (Figure 7), and spring (Figure 8) of 1999 was higher than usual (Figure 5). In October 1999 Chl concentration was especially high in the northern part of the Newfoundland Region, in February 1999 the bloom occurred in the Gulf Stream frontal zone, and in April 1999 almost the entire Newfoundland Region was occupied by hypertrophic waters (Chl >1 mg/m 3 ). The remotely sensed data we have analyzed allow us to consider the following three main effects which influence the phytoplankton development: the intensity of the Labrador Current and the Gulf Stream, the wind forcing, and the heat flux through ocean surface influencing stratification of the upper water layer. The SSH and SST data confirm that in October 1998 the intensity of Labrador Current was lower than usual, in February 1999 it was about the climatic mean, and in April 1999 it was significantly stronger. The latter is especially 19

20 evident from Figure 9 (Regions 1 and 4) where the negative SST anomaly in April 1999 was extremely high. Clearly the transport of phytoplankton with currents cannot be neglected. The current intensity alone, however, cannot explain the increased Chl concentration. The Labrador water contains high Chl values in summer and low in autumn-winter-spring, when higher Chl values occurred. In contrast, the Labrador Current transports high phytoplankton biomass in summer. During summer seasons of 1998 and 1999 the Chl distribution was similar to patterns observed in (Figure 5). The oscillations of the intensity of the Labrador Current therefore do not correspond to Chl variations. Hence, it was the combination of wind forcing and heat flux that resulted in the enhanced phytoplankton bloom during the period autumn 1998 spring In October 1998 the mean air temperature did not differ evidently from the values observed during other years (Table 2). At the same time the westerly winds in the northern part of the Newfoundland Region (Figure 6) were much weaker than the climatic mean values (Figure 3). Since it seems to be incorrect to compare the absolute values of wind averaged during periods of different duration, we analyze the general pattern of wind over the Newfoundland Region. In October the wind is usually stronger in the northern part and weaker in the southern part (Figure 3). In October 1998 the pattern was reverse. We can speculate that the weaker winds resulted in lower level of autumn mixing. In the zone of influence of Labrador Current in October the phytoplankton growth is light-limited. Thus, the lower wind mixing (i. e., the increased water column stability) resulted in unusual phytoplankton bloom in the northern part of the region we study. In February 1999 we also observed weaker winds in some parts of the region (Figure 7). The weakest westerly winds appeared to be over the Gulf Stream Frontal Zone. Taking into account the extremely warm late winter and spring of 1999 (+4 C in February, see Table 2), we can conclude that the increased heat flux from the atmosphere coupled with weakened wind resulted in earlier formation of a seasonal thermocline and enhanced stratification of water column. This process seems to be most intense in the Gulf Stream Frontal Zone where the warm waters of the Gulf Stream cover cold waters of the Labrador Current (see Figure 2 in Vinogradov et al., 1998). This concept is confirmed by similarity between the patterns of Chl distribution observed in February 1999 (Figure 7) and those observed in April (Figure 5). Both hydrophysical and hydrobiological states in the Gulf Stream Frontal Zone in February 1999 therefore reflect the typical situation of spring bloom. The vernal phytoplankton 20

21 bloom occurred in 1999 (and 1998) in April as usual. This bloom was much stronger than the climatic mean (compare Figure 5 and Figure 8). The main reason again seems to be the atypical pattern of wind forcing together with warmer air temperature (see Table 2). The wind pattern in April was peculiar: the westerly winds were strong in the southern part of the region and weak in the north. We can speculate that the weak winds favored the formation of seasonal thermocline to the north of the Gulf Stream Frontal Zone. To the south of the Gulf Stream Frontal Zone strong westerly winds disturbed the stratification of the upper mixed layer (the effect similar to autumn mixing), enabling transport of nutrients to the euphotic zone. Both reverse processes resulted in similar effect: enhanced development of phytoplankton. As a result, the spring bloom over the Newfoundland Region was much stronger than usual. We have to keep in mind, that the region under study is subject to strong advective influence of nutrient supply (Williams and Follows, 1998). This mechanism could explain the increase of Chl in April 1999 as a result of strengthening of the Labrador Current. However, it cannot explain the pattern observed in October 1998 (slow Labrador current and high Chl concentration in some parts of the region). Therefore, the remote sensed data provide sufficient information on ocean productivity and physical environment to relate them using the simple verbal model of light and nutrient limitation of phytoplankton growth. The ocean regions of sharp climatic contrasts (e. g., the Newfoundland region) seem to be a convenient area for this kind of analysis. Acknowledgements We thank the centers of remotely sensed data processing and dissemination: Physical Oceanography Distributed Active Archive Center at the Jet Propulsion Laboratory PODAAC JPL), Austin Center for Space Research in the University of Texas, and Distributed Active Archive Center in Goddard Space Flight Center. N. P. Nezlin and A. G. Kostianoy appreciate the hospitality and financial support during their visit to the Memorial University of Newfoundland where this work was initiated. The research reported in this paper has been partly supported by the Natural Sciences and Engineering Research Council of Canada under Grant