Variability in the properties of Shelf Water in the Middle Atlantic Bight,

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. C1, 3014, doi: /2001jc001044, 2003 Variability in the properties of Shelf Water in the Middle Atlantic Bight, David G. Mountain Northeast Fisheries Science Center/NMFS, Woods Hole, Massachusetts, USA Received 3 July 2001; revised 23 April 2002; accepted 16 May 2002; published 25 January [1] The seasonal and interannual variability in the temperature, salinity, and volume of Shelf Water (SHW) in the Middle Atlantic Bight (MAB) is described for the period Large interannual variations in the volume, salinity, and, to a lesser extent, temperature of the SHW occurred that were coherent over multiple year time periods. The variations in volume and salinity originated through processes acting outside of the MAB and were advected into the region from the Gulf of Maine. On a decadal average the SHW observed in the 1990s was approximately 1 C warmer, 0.25 PSU fresher, and 1000 km 3 more abundant than during the period. The warming during the 1990s was largest in the southern part of the bight during the winter, when the SHW was more than 2 C warmer then during the earlier decade. INDEX TERMS: 4215 Oceanography: General: Climate and interannual variability (3309); 4219 Oceanography: General: Continental shelf processes; 4223 Oceanography: General: Descriptive and regional oceanography; 4546 Oceanography: Physical: Nearshore processes; KEYWORDS: Middle Atlantic Bight, shelf water, interannual variability Citation: Mountain, D. G., Variability in the properties of Shelf Water in the Middle Atlantic Bight, , J. Geophys. Res., 108(C1), 3014, doi: /2001jc001044, Introduction [2] The Middle Atlantic Bight (MAB) is the continental shelf region between Nantucket Shoals and Cape Hatteras (Figure 1). The primary water mass in the bight is Shelf Water (SHW). It is generally cooler and lower in salinity than the oceanic waters seaward of the shelf, commonly termed the Slope Water (SLW). The boundary between these two water masses occurs in a narrow transition region referred to as the shelf/slope front [see Linder and Gawarkiewicz, 1998]. [3] The SHW in the MAB is formed as a water mass in the Gulf of Maine. Cold, low-salinity water from the Scotian Shelf (SSW) enters the gulf in the surface layer around Cape Sable [Smith, 1983] and the warmer, more saline SLW enters the gulf at depth through the Northeast Channel [Ramp et al., 1985]. These two water masses mix as they move in the circulation around the gulf. From the western gulf the product of this mixing enters onto the northern side of Georges Bank to flow clockwise around the bank and then westward from the bank s southern flank past Nantucket Shoals into the MAB. Occasionally SSW flows directly across the Northeast Channel to mix directly with the waters on eastern Georges Bank [Bisagni et al., 1996]. Direct measurements of the westward transport past Nantucket Shoals have been described by Beardsley et al. [1985] and Ramp et al. [1988]. Once in the MAB the properties of the SHW are modified locally by seasonal heating and cooling, by local precipitation and river runoff, and by mixing with the offshore SLW. Fairbanks [1982], This paper is not subject to U.S. copyright. Published in 2003 by the American Geophysical Union. Chapman et al. [1986], and Chapman and Beardsley [1989] have shown that much of the freshwater component of the SHW originates to the north, as far as the Labrador Sea, and that the SHW in the MAB is part of a large scale coastal current system that extends from Labrador to Cape Hatteras. [4] SHW leaves the MAB through a number of processes. Some SHW traverses the length of the MAB and leaves the shelf near Cape Hatteras, where it then may flow eastward along the northern edge of the Gulf Stream [Ford et al., 1952; Kupferman and Garfield, 1977; Churchill et al., 1989, 1993]. Warm core Gulf Stream rings can entrain SHW when they impinge upon the edge of the shelf [Morgan and Bishop, 1977; Bisagni, 1983; Garfield and Evans, 1987]. Smaller eddies generated by instabilities in the shelf/slope front can entrain SHW [Ramp et al., 1983; Garvine et al., 1988]. Smaller scale mixing and exchange also occur between the SHW and SLW at the shelf/slope front. The Shelf Edge Exchange Processes experiments, SEEP I [Walsh et al., 1988] and SEEP II [Biscaye et al., 1994], did extensive studies of the cross frontal exchange in the MAB. While the transport of SHW into the MAB can be directly measured [e.g., Beardsley et al., 1985], the processes removing SHW from the MAB are much more difficult to measure and act along the entire length of the shelf. Quantitative estimates of the rate SHW is removed by the various processes listed above and of seasonal or interannual variations in those rates are not well documented. [5] The characteristic temperature and salinity properties of SHW have been described by Bigelow and Sears [1935], Wright and Parker [1976], and Beardsley et al. [1976]. Manning [1991] has shown that interannual variability in the SHW salinity is associated largely with variability in 14-1

2 14-2 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY Figure 1. The Middle Atlantic Bight and the five subregions used in this analysis. river runoff (Gulf of Maine rivers and the St. Lawrence River) and in local precipitation. Mountain and Manning [1994] and Mountain and Taylor [1998] have shown that the interannual variability in salinity is quite coherent from the MAB off Chesapeake Bay northward into the Gulf of Maine. Mountain [1991] described changes in the amount or volume of SHW in the MAB, showing large interannual variations that appeared associated with changes in rate of transport through the Gulf of Maine/Georges Bank system. [6] Smith et al. [2001] report recent measurements ( ) of the transport of SSW and SLW into the Gulf of Maine system. The annual mean SSW inflow around Cape Sable was approximately twice that measured by Smith [1983] (0.30 versus m 3 s 1 ), and the inflow of SLW through Northeast Channel at depths >75 m was half of that measured by Ramp et al. [1985] (0.14 versus m 3 s 1 ). While the total transport of the two sources was approximately the same in the late 1970s as in the mid-1990s, the proportion represented by the two sources was reversed. Smith et al. [2001] attribute the lower salinity observed in the gulf region during the 1990s compared to the earlier period to the greater contribution of low-salinity SSW. Smith et al. [2001] also measured the transport of SSW into the Gulf of Maine in the surface layer over Northeast Channel (<75 m) (annual mean = m 3 s 1 ). This component of the inflow to the gulf was not addressed in previous measurements. [7] Manning [1991], Mountain [1991], Mountain and Manning [1994], and Mountain and Taylor [1998] all used the MARMAP hydrographic data set as the basis for their analyses. That data set covered the period An additional decade of hydrographic data has been collected across the MAB as part of routine sampling operations by the National Marine Fisheries Service (NMFS). This report combines data over the period to consider decadal scale changes in the temperature, salinity, and volume of SHW in the MAB. 2. Data and Methods [8] Two hydrographic data sets are used in this analysis. The first set is from the NMFS MARMAP program, that covers the period and sampled the shelf from Cape Hatteras northward to Georges Bank and through the

3 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY 14-3 Gulf of Maine. A set of standard stations (approximately 90 in the MAB) was occupied on 49 survey cruises over the period, although not all stations were occupied on each survey. The hydrographic measurements were made by water bottles with reversing thermometers, until the last year of the program. Beginning in 1987, observations were made using an electronic conductivity/temperature/depth profiling instrument (CTD). However, after 1987 no routine hydrographic surveys in the MAB were made by NMFS until [9] The second data set used in this analysis consists of the routine observations made by NMFS between 1991 and Beginning in 1991, CTD measurements were routinely made on NMFS surveys conducted to collect information on the distribution and abundance of fish species in support of stock assessment activities. These surveys used a stratified random station design. Therefore, unlike the MARMAP observations, the data from the 1990s were not collected at standard, repeated station locations. From 1991 to 1999, hydrographic measurements were made on 70 surveys that covered at least major portions of the MAB. [10] Following Mountain [1991] (hereinafter referred to as M91) and Manning [1991], SHW is defined as the water with salinity less than 34 PSU. Generally on the shoreward side of the shelf/slope front the salinities are less than 33.5 and on the seaward side, above 34.5, indicating a distinct separation in properties between the SHW and SLW [Manning, 1991]. The appropriateness of using 34 PSU (or any constant value) to define SHW is considered in the Discussion section. [11] The average temperature, salinity, and volume of SHW have been determined for 5 regions of the MAB (Figure 1). While M91 divided the bight into 3 regions, using 5 regions allows the movement or advection of properties through the MAB to be identified more clearly. From northeast to southwest the five regions are SNE (southern New England), NYB1 (New York Bight 1), NYB2 (New York Bight 2), SS1 (Southern Shelf 1) and SS2 (Southern Shelf 2). Within each region, for each survey with sufficient station coverage, the volume of SHW was determined and the volume-weighted average temperature and salinity of the SHW was calculated. To do this the area of a region was apportioned among the stations occupied in that region, based on an inverse distance squared weighting. The thickness of SHW at a station multiplied by its apportioned area determined the volume of SHW represented by the station. The sum for all stations determined the total SHW volume in the region. For each station the average SHW temperature and salinity also were calculated. These average values were multiplied by the volume of SHW for the station. The sum of these products for all the stations in the region was divided by the total SHW volume in the region to determine the volume-weighted SHW temperature and salinity for the region. The average year day for the observations in the region was determined through a volume-weighted averaging, as well. Variations in the sampling locations on the different surveys resulted in estimates of total volume for a region varying by up to 10%. To avoid bias due to station coverage, the total volume in a region was also calculated and the average value determined. For each survey the calculated SHW volume was then normalized by the ratio of the mean volume for the whole region to the regional volume calculated for that survey. For stations near the edge of the shelf, only the upper 100 m of the water column was included in the calculations. SHW was never observed at depths deeper than 100 m. The number of stations contributing to a SHW value in a region ranged from 7 to 44, with an average of approximately 20. [12] On some surveys SHW extended seaward of the boxes in Figure 1 and seaward of the observations. To address these situations M91 included estimates of SHW seaward of the hydrographic sampling domain by using the surface manifestation of the shelf/slope front observed in satellite imagery. Because of uncertainties in making those estimates, that technique is not used here. The current analysis addresses only the waters inside the boxes in Figure 1. The implications for the analysis of SHW extending beyond the sampling domain are addressed in the Discussion section. [13] To determine seasonal and interannual variability in the water properties, annual cycles for the properties in each region were calculated following the method described by M91. The annual cycles were calculated using the MAR- MAP data and are used as reference for determining interannual variability throughout the whole period of observations: The MARMAP data set used a fixed set of station locations and had good spatial and seasonal coverage in all regions of the MAB. For each estimate of a SHW parameter value (temperature, salinity, and volume) in a region an anomaly value was determined by subtracting from the estimate the value of the MARMAP annual cycle for that parameter in that region on the corresponding year day. These anomaly values represent the interannual variability in the water properties. A MABwide anomaly was calculated for each instance when there were values in each region within a 30 day period. The volume anomaly was the sum of regional volume anomalies. The temperature and salinity anomalies were determined by the average of the regional temperature and salinity anomalies. Using the MARMAP data as the basis for the annual cycles and the anomaly calculations allows direct comparison with earlier analyses that were based on the MARMAP data set [e.g., M91, Manning, 1991; Mountain and Manning, 1994; Mountain and Taylor, 1998]. Since the anomalies are referenced to annual cycles calculated for the MARMAP period, the anomalies for the whole record ( ) do not integrate to zero. [14] Following Manning [1991], the distribution of volume by salinity (or the percent of volume observed in different ranges of salinity) was determined for each region sampled on each survey. Volumes were determined for each 0.5 PSU interval from The calculations were done in a manner similar to determining the volume of SHW, with the thickness of water column in each salinity interval being multiplied by the area represented by the station and summed for all stations in the region. Any volume at values less than 31 was added to the first salinity interval and for values greater than 36, to the last interval. [15] To investigate the cause of salinity variability, the monthly precipitation values at New York City (JFK airport) were used as a proxy for local fresh water input. The monthly mean for the reference period, , was determined and subtracted from the original values to derive an anomaly series for comparison with the SHW anomaly

4 14-4 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY Figure 2. Annual cycle of SHW temperature for the different regions in Figure 1. The curves were determined from data for the MARMAP period ( ). The vertical bars represent ±1 standard deviation of the residuals for the original data from the fitted curves. series. Similar anomaly series were derived for monthly air temperature at New York City and at Norfolk, VA to consider the potential for atmospheric forcing of SHW temperature variability. 3. Results 3.1. Temperature [16] The annual cycles for temperature in the different regions during the MARMAP period (Figure 2) are asymmetric, with the warming period being longer than the cooling (approximately 220 days versus 145 days). The minimum temperatures in the regions are within about one degree and occur progressively later from south to north, with the minimum in SNE being three weeks after that in SS2. The maximum values have a similar shift in timing, but also exhibit a spatial pattern with the maximum in SS2 being about 4 C warmer than that in SNE. The seasonal range increased from 9 C in SNE to 13 C in SS2. [17] During the 1990s the temperature patterns in the northern two regions were quite similar to Figure 2. In the southern regions, however, the winter temperatures were considerably warmer than during the MARMAP period. The temperature anomalies during the 1990s in each region were averaged over the decade for three periods of the year (calendar days 1 125, , and ) (Figure 3). The spatial and temporal patterns of the average anomalies suggest a winter warming that increased toward the southern part of the bight. The middle part of the year was generally cooler and the latter part of the year was moderately warmer Salinity [18] As shown by Manning [1991] the SHW salinity during the MARMAP period exhibited interannual variability that was comparable in magnitude to the seasonal variability. Significant annual cycles were found only for NYB1 and NYB2, where the nearby Hudson River inflow causes a consistent seasonal decrease of the salinity to a minimum value in the summer. In the other regions the interannual variability overshadowed the seasonal variability and a significant annual cycle was not found. In these regions the mean salinity value was used as a reference for Figure 3. Average SHW temperature anomalies during the 1990s in each MAB region for three periods of the year. calculating anomalies. Unlike temperature, the salinity anomalies did not show a characteristic spatial or seasonal pattern and are discussed below on a MAB-wide basis Volume [19] A significant annual cycle in SHW volume was found for each region except SS2 (Figure 4). The cycles Figure 4. Annual cycle of SHW volume for the regions in Figure 1. The curves were detemined from data for the MARMAP period ( ). The times of maximum volume are indicated by the labeled vertical lines. The vertical bars represent ±1 standard deviation of the residuals for the original data from the fitted curves.

5 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY 14-5 Figure 5. The characteristic bottom salinity distribution for calendar day 30. show a progression in the timing of the maximum volume from north to south. The time of maximum volume shifts from day 109 in SNE to day 224 in SS1. This represents an propagation rate of about m s 1, similar to the result reported by M91. The ratio of the range to the mean in the annual cycles increases progressively from 39% to 45% to 64% to 80% from SNE to SS1. The large relative amplitude of the cycles, particularly on the southern part of the area, implies that the boundary between the SHW and SLW undergoes significant, seasonal onshore-offshore movement. The characteristic bottom salinity distribution on calendar day 30 (Figure 5), when NYB2 and SS1 are near their annual minimum in volume, was determined from annual cycles for individual MARMAP stations [see Mountain and Taylor, 1998]. The 34 PSU contour is over the midpart of the shelf in these regions. The surface 34 PSU contour would be in approximately the same location, except south of New England where it would be closer to the shelf edge. By contrast, at the time of maximum volume (approximately calendar day 210) the SHW extends across the width of the shelf. The 34 contour at the bottom would be near the shelf edge and at the surface, seaward of the edge MAB Anomalies [20] The MAB-wide property anomalies were derived by combining the anomalies in the different regions (Figure 6). Each anomaly series is characterized by variability that is coherent on a multiple year timescale. The magnitude of the volume variability is large, with an overall range of about 4000 km 3, comparable to the mean volume. The salinity variability, as also shown by Manning [1991], is large, with a range of over 1 PSU. The volume and salinity anomalies exhibit a significant inverse relationship (R 2 = 0.51). The temperature anomalies are generally ±1 C during the MAR- MAP period and 0 2 C during the 1990s, and are small in relation to the amplitudes of the annual temperature cycles (Figure 2). Compared to the MARMAP period, the SHW in the MAB during the 1990s was about 1 C warmer, 0.25 PSU fresher, and about 1000 km 3 greater in volume. [21] The large range in SHW volume evident in Figure 6 is particularly striking. A few examples illustrate the spatial changes in SHW distribution associated with the large volume anomalies. The bottom salinity distribution in February 1993 (Figure 7a) indicates that almost no water with salinity >34 was on the shelf. In contrast during February 1995 (Figure 7b) SHW was limited to the inner half of the shelf. The salinity distribution on a section across the shelf from near the mouth of Delaware Bay in March 1982 shows SHW extending to the shelf edge (Figure 8a). In March 1980, on the same section, the SHW boundary was about 70 km closer to the coast (Figure 8b) Volume by Salinity [22] The percent of the total volume of water represented by each 0.5 PSU increment of salinity was calculated for all regions covered on each survey. To determine the characteristic distribution of volume by salinity class for a region, the surveys were separated first into those from highsalinity and low-salinity periods evident in Figure 6. Averaging the distributions for all of the surveys in a region would result in a distribution that was spread across a wide salinity range and that was not representative of the conditions at any one time. Given the large amplitude of the annual cycle in SHW volume (Figure 4), the surveys also Figure 6. MAB-wide anomalies in SHW volume, salinity, and temperature.

6 14-6 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY Figure 7. Bottom salinity distribution during (a) February 1993 and (b) February were sorted to focus on the times of maximum and minimum SHW volume in a region, times when the boundary between the SHW and SLW would be either near the shelf edge or near the middle of the shelf. The surveys were selected if they occurred within 45 days of the time of maximum Shelf Water volume for the region or within 45 days of the time of minimum volume. For SS2, where no significant annual cycle in volume was found, the times of maximum and minimum volume were estimated by extrapolating the progression of the annual cycle evident in Figure 4 southwestward into the region. [23] Four sets of average volume by salinity distributions were calculated, for maximum and minimum volume during high- and low-salinity periods. The curves for two of these sets are presented (Figure 9). As noted in reference to Figure 6, when the SHW salinity was low (high), the SHW volume generally was high (low). So the distributions for the annual maximum of volume in the low-salinity periods (Figure 9a) represent the conditions during the maximum extent of Shelf Water observed in each region - the maximum volume case. Likewise, the distributions for the annual minimum of volume in the high-salinity periods (Figure 9b) represent the minimum extent of Shelf Water observed: the minimum volume case. For the maximum volume case (Figure 9a), the distributions are narrow, with most Shelf Water in SNE contained in the salinity range of 32.0 to As the pulse of maximum volume moves southwestward from SNE to SS2, the peak of the distribution remains at about the same salinity, although the amplitude gradually decreases as a separate, low-salinity peak increases due input from the Hudson River and Chesapeake and Delaware Bays. The comparable curves (not shown) for the high-salinity periods (at the annual maximum in volume) are similar to Figure 9a, although shifted to higher salinity by about 0.5 PSU. [24] The volume by salinity distributions in the lowvolume case (Figure 9b) are quite different. In SNE a broad distribution extends from 33 to 35 PSU. The two maximums at either end of the distribution represent the SHW on the inner shelf and the SLW on the outer shelf. Moving down the shelf through the other regions, the distribution narrows to a single, broad peak at PSU, and a clear separation between SHW and SLW does not exist. The comparable curves (not shown) for the low-salinity periods (at the annual minimum in volume) retain an identifiable peak in SHW volume between PSU from SNE to SS2, although not as distinctly as in Figure 9a. Therefore, in the four cases considered, SHW was evident as a distinct Figure 8. Salinity distribution on sections across the shelf from the mouth of Delaware Bay during (a) March 1980 and (b) March 1982.

7 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY 14-7 SHW. The situations represented by Figure 9b were not excluded from the analyses presented since that would have required a rather arbitrary selection of data to be used or excluded. Their inclusion in the analyses, however, likely does contribute to variability or noise in the results and may have made more difficult identifying relationships concerning SHW in the MAB. Figure 9. The average distributions of water volume by salinity class in the different regions for (a) the time of annual maximum SHW volume during low-salinity periods and (b) the time of annual minimum SHW volume during high-salinity periods. The heavier solid and dashed lines represent the curves for SNE and SS2, respectively. peak in volume at salinities < 34 in all situations except in the extreme minimum volume case (minimum volume/high salinity, Figure 9b), particularly in the southern regions of the bight. 4. Discussion [25] The analyses presented here assume that a salinity value of 34 can be used as an upper bound for identifying SHW. The volume by salinity distributions (e.g., Figure 9) support this assumption. In nearly all situations a peak in volume, identifying the SHW mass, is evident at salinities below 34 and volumes for salinity values at or just above 34 are small compared to those for the peak of the SHW. Only in the extreme minimum volume case (minimum volume/ high salinity, Figure 9b) is the SHW mass not distinct at values below 34. Therefore, the results and conclusions presented here are believed to be accurate representations of the SHW characteristics and not to be artifacts of using 34 PSU as a convenient upper bound for SHW. The results in Figure 9b indicate that under certain conditions, particularly in the southern regions of the bight, SHW does not exist as a distinct water mass that can be clearly separated from SLW, regardless of the choice for an upper salinity bound for the 4.1. Annual Cycles [26] The regional annual cycles for SHW temperature (Figure 2) reflect the coastal air temperature patterns with summer maximums increasing from north to south. However, unlike the water temperature cycles (Figure 2), the coastal air temperature cycles have symmetric warming and cooling periods and do not exhibit a shift from south to north in the time of the annual minimum and maximum values. The flattening of the SHW temperature curves (or a decrease in the rate of warming) during the summer in NYB2 and SS1 likely is due the southwestward flow of SHW through the MAB. With the along shelf temperature gradient, the southwestward flow represents a local advective cooling that partially balances the local surface heating. Houghton et al. [1982] showed that below the thermocline in the cold pool on the outer shelf this advective cooling can cause decreasing temperatures during the summer in some locations. [27] Significant annual cycles for SHW salinity were calculated only for NYB1 and NYB2, where a summer salinity minimum results from the inflow through the Hudson River system. The absence of significant annual cycle in the southern regions of the shelf is surprising, given the large, seasonal freshwater inflow from the Delaware and Chesapeake Bays. In these regions the large interannual variability in SHW salinity apparently overshadows the seasonal signal. With a finite number of observations (40) scattered over high- and low-salinity periods during the ten year MARMAP sampling, annual cycles in salinity are not evident in the data sets. [28] The annual cycles for SHW volume (Figure 4) have characteristics similar to those described by M91. The amplitudes of the cycles are significant portions of the mean volume. The cycle s phase advances through the MAB from the northeast to the southwest at the approximate rate of the integrated mean flow ( m s 1 ) [Beardsley et al., 1976], indicating that the variation in volume advects as part of the mean flow on the shelf. M91 concluded that the annual cycle in SHW volume originates from the seasonal patterns in the inflows of Scotian Shelf water and Slope Water to the Gulf of Maine. The encroachment of the SHW/SLW boundary onto the midshelf during the wintertime minimum in SHW volume (Figure 5) also is evident in the results of Linder and Gawarkiewicz [1998] for a region off New Jersey, approximately comparable to region NYB2 in this analysis. In addition, the summary of surface fronts evident in AVHRR satellite imagery in the region by Ullman and Cornillon [1999] suggests a characteristic midshelf thermal front during the winter, particularly in the region east of Delaware Bay. The lack of an annual cycle for SHW volume in SS2 is believed due to the large interannual variability overshadowing the annual cycle and to the lack of a distinct SHW mass in some situations (e.g., Figure 9b).

8 14-8 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY [29] The volume by salinity distributions (Figures 9a 9b) document the change in the SHW mass as it moves southwestward through the MAB. In the high volume case (Figure 9a), when the boundary is near the edge of the shelf, the core salinity of the water mass does not change significantly, suggesting relatively little cross frontal exchange. In the minimum volume case (Figure 9b), when the boundary (i.e., the 34 PSU isohaline) is near the midshelf, the SHW and SLW become less distinct and merge as the water moves southwest through the bight, suggesting considerably more cross frontal mixing Interannual Variability [30] The temperature and salinity variability in the MAB during the MARMAP period has been shown to be similar to that observed in the Gulf of Maine and on Georges Bank [Mountain and Manning, 1994; Mountain and Taylor, 1998]. The salinity variability on northwestern Georges Bank (Figure 10), calculated from the same data sets and methods used here, is similar to the variability in the MAB (Figure 6) throughout the period of the observations. In a regression analysis the variability on Georges Bank could explain 61% of the salinity variability in the MAB. In a similar comparison for temperature the Georges Bank variability could explain 42% of that observed in the MAB. Therefore for the whole period the majority of the SHW salinity variability and, to a lesser extent, of the temperature variability occurred on a regional scale and not independently within the MAB. [31] The variability in water properties likely was caused by a combination of local atmospheric processes (e.g., changes in precipitation/runoff for salinity and in surface heating/cooling for temperature) and advection. The atmospheric variability generally is coherent over large length scales and may have similarly influenced the waters in both the MAB and the Gulf of Maine, contributing to the regional nature of the variability identified above. For example, the R 2 values between the anomalies at Boston and New York City (JFK airport), integrated over six months, are 0.41 for precipitation and 0.64 for air temperature. Given this regional nature of the (local) atmospheric processes, the advective changes of most interest are changes in the properties of the major inflows to the Gulf of Maine/Georges Bank/MAB coastal system: the inflows of SSW around Cape Sable and of SLW through the Northeast Channel. [32] Manning [1991] showed that the salinity variability during the MARMAP period could be explained largely by a combination of local precipitation and river inputs to the coastal system. However, these local freshwater sources are not as successful in explaining the variability over the whole record considered here since there was not an increase in precipitation or discharge that could account for the general decrease in salinity during the 1990s. For example, the R 2 between the precipitation variability measured at JFK airport summed over six month and the salinity variability is 0.47 during the MARMAP period and 0.27 during the 1990s, with the calculation done separately for each period. For the record as a whole, however, the value is only 0.22, suggesting that variations in precipitation did not cause the general decrease in salinity observed during the 1990s relative to the MARMAP period. The total precipitation at Figure 10. Salinity anomaly for the surface layer (0 30 m) on northwestern Georges Bank. JFK airport during the MARMAP period and during the 1990s was the same to within 0.5%. In quantitative terms, the range of the precipitation variation, if applied to a 60 m water column, could cause salinity changes of ± PSU. Therefore, while changes in local freshwater input may have made a significant contribution to the variability in SHW salinity, it was not a factor in the general freshening trend for the MAB SHW during the 1990s. [33] Comparisons of the monthly air temperature anomalies for New York City (JFK Airport) with the temperature anomaly series for the two northern regions of the bight, SNE and NYB1, yield R 2 values of 0.40 for each region. Similar comparisons of the temperature anomalies in the three southern regions with the air temperature anomalies for Norfolk, VA yield R 2 values of Therefore a significant portion of the interannual variability in SHW temperature, particularly in the southern part of the bight, appears related to variability in the local atmospheric heating. However, given the relatively small magnitude of the temperature variability compared to the amplitude of the annual cycle, a detailed surface heat flux analysis would be required to adequately test this hypothesis. [34] Smith et al. [2001] measured the properties of the major inflows to the Gulf of Maine and showed that the salinity variability around the Gulf of Maine and on Georges Bank (i.e., Figure 10) during the mid-1990s was coherent and due, in large part, to changes in the relative proportions of SSW and SLW flowing into the Gulf of Maine. They also showed [Smith et al., 2001, Figure 10] that the temperature/salinity properties at middepth in the Gulf exhibited a close relationship, falling along a mixing line between SSW and SLW. The slope of the mixing line is about 2.6 C/PSU. To consider these advective changes as a source for the MAB SHW variations, the regression of the SHW temperature anomalies on the salinity anomalies was determined in each region in the MAB, for both the MARMAP and the 1990s periods. Winter observations, when local cooling and convection could alter and dominate the temperature/salinity relationship, were omitted from the analysis. The R 2 values were small (<0.22) for all cases except for SNE (R 2 = 0.66) and NYB1 (R 2 = 0.51) during the 1990s. The temperature/salinity anomaly points for SNE and NYB1 fall along a line closely parallel to the mixing line in the Gulf of Maine (Figure 11). Therefore, a major source of the temperature and salinity variability in the

9 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY 14-9 Table 1. Average Air Temperature Anomalies During the 1990s Referenced to the MARMAP Period ( ) at New York City (JFK Airport) and Norfolk, Virginia, for Three Periods of the Year Calendar Days NYC 0.6 C 0.2 C 0.1 C Norfolk 0.6 C 0.8 C 0.7 C Figure 11. SHW temperature/salinity anomalies for SNE (solid circles) and NYB1 (open circles) for calendar days (excluding winter) during the 1990s. The dashed line indicates the slope of the mixing line between SHW and SLW observed in the Gulf of Maine [Smith et al., 2001, Figure 10]. northern MAB during the 1990s is believed to be changes in the mixing ratio of SSW and SLW flowing into the Gulf of Maine. The SNE and NYB1 points are consistently warmer then the mixing line by about 1 C. This increase in temperature likely was associated with the local surface heat flux, as suggested by the air temperature regressions discussed above. As the SHW flowed on to the southern parts of the bight, the tight relationship between the temperature and salinity anomalies likely broke down under the accumulated influence of the local processes described above. [35] Smith et al. [2001] also attribute the general decrease in salinity in the Gulf of Maine and on Georges Bank during the mid-1990s to a greater inflow of the low-salinity SSW into the gulf around Cape Sable and a lesser inflow of higher-salinity SLW at depth through Northeast Channel. On a decadal scale the advection of these changes into the MAB is believed the cause of the lower SHW salinity during the 1990s evident in Figure 6. From the slope of the mixing line between SSW and SLW (Figure 11), the advective change causing the decadal decrease in salinity also would have caused an accompanying temperature decrease of about 0.7 C. That the SHW temperature actually increased during the 1990s (Figure 6) suggests that the local atmospheric heating had to be sufficient to overcome the advective cooling and still result in the observed increase in SHW temperature. [36] The decadal average temperature anomalies during the 1990s (Figure 3) show a seasonal pattern marked by warmer winter values, particularly in the southern regions and cooler summers. Decadal average air temperature anomalies at NYC and Norfolk (Table 1) show a somewhat similar pattern, with positive anomalies in the winter and negative in the summer. The air temperature anomalies, however, do not exhibit a larger winter warming to the south or the moderate warming in the fall throughout the region as seen in the SHW anomalies (Figure 3). Also, the air temperature anomalies are smaller in magnitude than the regional water temperature changes. As noted above, while the statistical comparisons are suggestive, a detailed heat flux analysis would be needed to adequately determine the contribution of local heating to the decadal change in SHW temperature. Another source for the decadal temperature change could be an advective heat flux associated with a small change in the decadal mean flow rate through the MAB, particularly in summer when there is a strong northsouth gradient in the SHW temperature. However, no adequate flow information exists to investigate this potential source of variability. [37] M91 concludes that the interannual variability in SHW volume was associated with changes in the transport of water through the Gulf of Maine system. Specifically, the low volume observed in 1985 and the high volume observed in 1987 were related to changes in transport inferred from hydrographic measurements. Smith et al. [2001] report on direct measurements of the inflows to the Gulf of Maine from 1993 to For the latter measurements monthly transport anomalies were calculated (P. C. Smith, personal communication, 2000) for the combined inflows around Cape Sable and through Northeast Channel. A monthly, accumulated volume anomaly time series was calculated by integrating for each month the transport anomalies over the previous six months. This volume anomaly series is similar to the observed MAB SHW volume anomalies (Figure 12) in both the temporal pattern and the amplitude of the anomalies. Therefore, as suggested by M91, the interannual variability in the volume of SHW in the MAB is believed to result largely from changes in the net inflow of water to the Figure 12. Comparison of SHW volume anomalies (bars) with integrated anomalies of transport into the Gulf of Maine (open circles).

10 14-10 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY Gulf of Maine system. Changes in the volume of SHW reflect changes in the balance between the rates of inflow to and outflow from the MAB. If the interannual variability in SHW volume matches the inflow variability, as in Figure 12, then the rate of outflow from the MAB must have exhibited little interannual variability. [38] The large positive volume anomalies for the Gulf of Maine inflows (Figure 12), particularly over the last half of the record, were caused primarily by increased inflow of low-salinity SSW in the surface layers [Smith et al., 2001, Figure 6]. Therefore the period of increased transport was associated with lower salinities in the Gulf of Maine and on Georges Bank (Figure 10 and Smith et al. [2001, Figure 8]). The inverse relationship between the MAB volume and salinity anomalies (Figure 6) is believed due to this same characteristic of the variability in the Gulf of Maine inflows. [39] The SHW extended beyond the boundaries of the sampling domain on some surveys, and in these cases the total volume of SHW was not determined accurately. Many of the MAB-wide estimates of SHW volume from the 1990s approach the total volume of the MAB (6500 km 3 ) and likely are underestimates of the true volume of SHW. The approximate 1000 km 3 increase in the mean SHW volume from the MARMAP period to the 1990s represents a lower bound on the actual increase that occurred. This sampling limitation would not be expected to have a significant effect on the SHW temperature or salinity estimates. [40] The cause of the large increase in mean SHW volume in the 1990s is not known. The estimated mean transport into the Gulf of Maine system around Cape Sable and at depth through Northeast Channel during [Smith et al., 2001] is remarkable similar to the combined earlier estimates of Smith [1983] and Ramp et al. [1985] (0.44 versus m 3 s 1 ). However, no earlier measurements exist of the transport into the gulf through the surface layer over Northeast Channel to compare with the m 3 s 1 estimate of Smith et al. [2001], and a decadal comparison of the total transport of the coastal current system is not possible. [41] The boundary between the SHW and SLW water masses generally is characterized as occurring in a sharp frontal region near the edge of the shelf, such that the boundary, the front, and the shelf edge all co-occur. With the large variability in SHW volume, the water mass boundary has been shown to occur over a wide range of the shelf (e.g., Figures 7 and 8). A series of models have investigated the formation of a coastal density front like the shelf/slope front in the MAB [e.g., Gawarkiewicz and Chapman, 1992; Chapman and Lentz, 1994; Chapman, 2000]. The models generally show the front to occur at an isobath that is determined by characteristics of the flow and not necessarily tied to the shelf break. The wide range of locations for the front in the observations presented here offers a range of conditions to which modeled frontal locations could be compared. [42] One counterintuitive aspect of the temperature variability in the MAB is illustrated by calculating the temperature for all of the water in the bight (within the boxes in Figure 1). While the SHW was about 1 C warmer during the 1990s compared to the MARMAP period (Figure 6), the average water temperature in the bight as a whole was about 0.2 C cooler in the 1990s. The overall cooling resulted from the larger volume of SHW in the bight during the 1990s, which was colder than the SLW it replaced. The common question, Are the waters warmer or colder? does not always have a simple answer, and needs to be considered separately on a water mass and a regional basis. 5. Conclusions [43] The analyses presented here indicate the following. 1. From 1977 to 1999 the Shelf Water in the MAB exhibited large interannual variability in its temperature, salinity, and volume. 2. The variability in Shelf Water volume and salinity, at least during the 1990s, was due, in large part, to changes in the rates of transport of Scotian Shelf Water and Slope Water into the Gulf of Maine system. 3. On a decadal average, the Shelf Water during the 1990s was about 1 C warmer, 0.25 PSU fresher and 1000 km 3 more abundant than during the MARMAP period. The lower salinity resulted from greater Scotian Shelf Water inflow to the Gulf of Maine. The temperature increase is believed due to increased local atmospheric heat flux, although a quantitative heat flux analysis will be needed to confirm this. The cause of the decadal increase in volume is not known. 4. With the large changes in Shelf Water volume the boundary between Shelf Water and Slope Water occurred over a wide range of locations across the shelf. [44] Acknowledgments. A debt of gratitude is owed to M. Taylor who coordinated the collection and directed the processing of most of the CTD data used in this analysis. C. Bascunan and T. Holzwarth also contributed significantly to the collection and processing of this data set. P. Smith kindly generated and provided the transport anomaly data used to develop Figure 12. Two anonymous reviewers are thanked for providing valuable criticisms and suggestions that improved the quality and clarity of this manuscript. References Beardsley, R. C., W. C. Boicourt, and D. H. Hansen, Physical oceanography of the Middle Atlantic Bight, in Middle Atlantic Continental Shelf and New York Bight, vol. 2, edited by M. G. Gross, pp , Am. Soc. of Limnol. and Oceanogr., Allen, Lawrence, Kans., Beardsley, R. C., D. C. Chapman, K. H. Brink, S. R. Ramp, and R. Schlitz, The Nantucket shoals flux experiment (NSFE79), part I, A basic description of the current and temperature variability, J. Phys. Oceanogr., 15, , Bigelow, H. B., and M. Sears, Studies of the waters of the continental shelf, Cape Cod to Chesapeake Bay II: Salinity, Pap. Phys. Oceanogr. Meteorol., 4, 94 pp., Bisagni, J. J., Lagrangian current measurements within the eastern margin of a warm-core Gulf Stream ring, J. Phys. Oceanogr., 13, , Bisagni, J. J., R. C. Beardsley, C. M. Ruhsam, J. P. Manning, and W. J. Williams, Historical and recent evidence of Scotian Shelf Water on southern Georges Bank, Deep Sea Res., Part II, 43, , Biscaye, P. E., C. N. Flagg, and P. G. Falkowski, The Shelf Edge Exchange Processes experiment, SEEP II: An introduction to hypotheses, results and conclusions, Deep Sea Res., Part II, 41, , Chapman, D. C., Boundary layer control of buoyant coastal currents and the establishment of a shelfbreak front, J. Phys. Oceanogr., 30, , Chapman, D. C., and R. C. Beardsley, On the origin of Shelf Water in the Middle Atlantic Bight, J. Phys. Oceanogr., 19, , Chapman, D. C., and S. J. Lentz, Trapping of a coastal density front by the bottom boundary layer, J. Phys. Oceanogr., 24, , Chapman, D. C., J. A. Barth, R. C. Beardsley, and R. G. Fairbanks, On the continuity of mean flow between the Scotian Shelf and the Middle Atlantic Bight, J. Phys. Oceanogr., 16, , Churchill, J. H., P. C. Cornillon, and P. Hamilton, Velocity and hydrographic structure of subsurface Shelf Water at the Gulf Stream s edge, J. Geophys. Res., 94, 10,719 10,800, 1989.

11 MOUNTAIN: MIDDLE ATLANTIC BIGHT SHELF WATER VARIABILITY Churchill, J. H., E. R. Lavine, D. N. Connors, and P. Cornillon, Mixing of shelf, slope and Gulf Stream water over the continental slope of the Middle Atlantic Bight, Deep Sea Res., Part I, 40, , Fairbanks, R. G., The origin of continental shelf and slope water in the New York Bight and Gulf of Maine: Evidence from H 2 18 O 2 /H 2 16 O 2 ratio measurements, J. Geophys. Res., 87, , Ford, W. L., J. R. Longard, and R. E. Banks, On the nature, occurrence and origin of cold low salinity water along the edge of the Gulf Stream, J. Mar. Res., 11, , Garfield, N., and D. L. Evans, Shelf Water entrainment by Gulf Stream warm-core rings, J. Geophys. Res., 92, 13,003 13,012, Garvine, R., K.-C. Wong, G. G. Gawarkiewicz, and R. K. McCarthy, The morphology of shelfbreak eddies, J. Geophys. Res., 93, 15,593 15,607, Gawarkiewicz, G., and D. C. Chapman, The role of stratification in the formation and maintenance of shelfbreak fronts, J. Phys. Oceanogr., 22, , Houghton, R. W., R. Schlitz, R. C. Beardsley, B. Butman, and J. L. Chamberlin, Middle Atlantic cold pool: Evolution of the temperature structure during summer 1979, J. Phys. Oceanogr., 12, , Kupferman, S. L., and N. Garfield, Transport of low-salinity water at the slope water-gulf Stream boundary, J. Geophys. Res., 82, , Linder, C. A., and G. Gawarkiewicz, A climatology of the shelfbreak front in the Middle Atlantic Bight, J. Geophys. Res., 103, 18,405 18,423, Manning, J., Middle Atlantic Bight salinity: Interannual variability, Cont. Shelf Res., 11, , Morgan, C. W., and J. M. Bishop, An example of Gulf Stream eddy-induced water exchange in the Mid-Atlantic Bight, J. Phys. Oceanogr., 7, , Mountain, D. G., The volume of Shelf Water in the Middle Atlantic Bight: Seasonal and interannual variability, , Cont. Shelf Res., 11, , Mountain, D. G., and J. P. Manning, Seasonal and interannual variability in the properties of the surface waters of the Gulf of Maine, Cont. Shelf Res., 14, , Mountain, D. G., and M. H. Taylor, Spatial coherence of interannual variability in water properties on the U.S. northeast shelf, J. Geophys. Res., 103, , Ramp, S., R. C. Beardsley, and R. Legeckis, An observation of frontal wave development on a shelf-slope/warm core ring front near the shelf break south of New England, J. Phys. Oceanogr., 13, , Ramp, R. S., R. S. Schlitz, and W. R. Wright, The deep flow through the Northeast Channel, Gulf of Maine, J. Phys. Oceanogr., 15, , Ramp, S. R., W. S. Brown, and R. C. Beardsley, The Nantucket Shoals flux experiment, 3, The alongshelf transport of volume, heat, salt and nitrogen, J. Geophys. Res., 93, 14,039 14,054, Smith, P. C., The mean and seasonal circulation off southwest Nova Scotia, J. Phys. Oceanogr., 13, , Smith, P. C., R. W. Houghton, R. G. Fairbanks, and D. G. Mountain, Interannual variability of boundary fluxes and water mass properties in the Gulf of Maine and on Georges Bank: , Deep Sea Res., Part II, 48, 37 70, Ullman, D. S., and P. C. Cornillon, Satellite-derived sea surface temperature fronts on the continental shelf of the northeast U.S. coast, J. Geophys. Res., 104, 23,459 23,478, Walsh, J. J., P. E. Biscaye, and G. T. Csanady, The Shelf Edge Exchange Processes (SEEP) I experiment: Hypotheses and highlights, Cont. Shelf Res., 8, , Wright, R. W., and C. E. Parker, Volumetric temperate/salinity census for the Middle Atlantic Bight, Limnol. Oceanogr., 21, , D. G. Mountain, Northeast Fisheries Science Center/NMFS, Woods Hole, MA , USA. (david.mountain@noaa.gov)