Cruise Report

R/V EDWIN LINK 9904

Georges Bank

14-28 April 1999

TABLE OF CONTENTS

Purpose of the Cruise   1

Cruise Narrative   1

Individual Reports   3

Appendix. I. Event Log   12

Appendix II. List of Personnel   20

Appendix III. List of Figures  21

Appendix IV. Drifter Log  23

Purpose of the Cruise

Cruise Narrative

Individual Reports:

Physical Oceanography

         Drifter Deployments

         Shipboard Sensors

         Acoustic Doppler Current Profiler

         Hydrography

         Satellite Imagery

Real-Time Modelling

Ichtho-Zooplankton

          MOCNESS Sampling

          Bongo-net Sampling

          Special Collections

         Video Plankton Recorder (MOCNESS mounted)

Micro-Zooplankton

Nutrients

Biochemistry

Appendix I. Event Log

Appendix II. Personnel List

Appendix III. List of Figures

Appendix IV. Drifter Log

Acknowledgements

We gratefully acknowledge the very able assistance provided by the officers and crew of the R/V EDWIN LINK, the University of Miami Ships Support Group, and student volunteers. This report was prepared by Lew Inzce,  Betsy Broughton,  Larry Buckley, Cisco Werner, K.Fisher,  Maureen Taylor, and Jim Manning. This cruise was sponsored by the National Science Foundation and the National Oceanic and Atmospheric Administration.

Purpose of the Cruise

• Test agreement between model drogue predictions and the actual drifter trajectories.
• Use the model to forecast the on/offshelf movement of the tidal front to help guide the sampling effort; compare results with drifter data (on/offshelf excursions) and calculations using tidal polar plots.
• Run the model to forecast particle trajectories through the study area and the evolution of physical fields over time as a test of concept (can this be done realistically in a cruise framework, can it help inform the cruise in real time) and later for comparison with processed data from the cruise and moorings.
• Sample larval fish and zooplankton vertical and horizontal distributions so they can be combined with data and models to understand and quantify exchange across the tidal front.
• Sample larval fish and their predators and prey for biological rates (e.g., growth, condition, mortality). 

Cruise Narrative

R/V Edwin Link arrived in Woods Hole Tuesday morning, April 13, after a stormy passage from Fort Pierce. Waves had damaged the bulwark gate at the starboard quarterdeck, requiring repairs before sailing. These were made April 14, the original day scheduled for departure. Sailing was delayed until April 15 to accomodate this need and others: the ADCP needed to be adjusted (Charlie Flagg flew up from BNL to do this), the MOCNESS computer needed fixing (completed by Peter Wiebe very late Wednesday night), and Chief Scientist Greg Lough was suffering from a severe flu. On Thursday morning, April 15, Greg was still too sick to come to the vessel, and Lew Incze agreed to serve as Chief Scientist for the cruise. Edwin Link sailed from the NMFS dock in Woods Hole at approximately 13:30 that afternoon. We stopped in Nantucket Sound to test several systems before continuing on to the Southeast Part, Georges Bank, to begin operations in the daylight hours of Friday, April 16.

The first two days of sampling (April 16-17) were spent occupying a bongo grid beginning 1/2 nm east of, and parallel to, the Schlitz mooring line (Fig. 1; mooring positions are given in Table 1). Station spacing was 4 nm across the tidal front (first four stations from shallow end) and then 5 nm out to depths of about 85 m. We occupied three transects spaced 7 nm apart, proceeding westward. On-board counts of cod and haddock larvae were made at sea using one side of the bongo sample and revealed an abundance of both species (Fig. 2). A fourth transect was made east of the initial grid. The location of the tidal front was identified as centered on about the 60 m isobath using SeaCat profiles from the bongo tows . As seen in  Fig. 3,  the across-shelf temperature gradients were weak, but consistent. A deep- (35 m) and shallow- drogued (15m) ARGOS drifter was deployed on each side of the tidal front on transect 4 before beginning our work just west of the mooring line.

Most of the remainder of the cruise (6+ d) was spent along the Schlitz mooring line to sample conditions (nutrients, larval fish, zooplankton, hydrography) near moorings on the two sides and inside the tidal front. The seven moorings extend approximately 8 nm across the front from ca. 50-75 m bottom depth (see Table 1 for mooring instrumentation). Real-time runs of the Dartmouth Circulation Model for Georges Bank were implemented on board using 3-day forecasts of boundary sea level and wind which were e-mailed to the ship by Rick Luettich (UNC). A daily crontab job also sent ETA model wind and heatflux forecast files in the form of an automated email from a NMFS SGI machine.

Our approach was to sample 1 day at the deep end of the mooring transect (this became our Cruise Station #30); one day in the tidal front (Cruise Station 33); one day inside the front (Cruise Station 34); then a repeat of Station 30 for 2 d and a repeat of Station 34 for 1 d. These were interspersed with some additional CTD work which accounts for the particular numbering above. All of the work conducted at a "site" retained the same station number even with subsequent visits. Sampling consisted of a 1m² MOCNESS, CTD, CTD/pump and nutrients, on various schedules. Twice after the mooring transect study began, we ran an along- track survey along the current mooring array using the ship's ADCP; surface fluorescence, T and S; and SeaCat (SBE19) hydrography.

It will be noted from the event log and from any plotting of the stations that the precise locations of the stations moved around somewhat. There were several reasons for this which should be understood. First, we were operating near the moorings and needed to keep a safe operating distance from them. This distance and direction from the mooring changed with the phase of the tide and with weather conditions. Second, each sampling suite involved a few hours of wire time. We relocated to station between samples when we were more than 1/2 nm away from the original deployment site, but to save time we did not insist on a precise relocation (it was not necessary within the context of this study). Finally, on our first occupation of the tidal front station (Cruise Station 33), we moved with the tide rather than staying with a fixed location near the mooring. On our second occupation of Station 33 we remained near the mooring and let the tide do the moving. This is further explained below.

On the final day of sampling we went to a crest station to get nutrient conditions well away from the tidal front, and to a flank station to sample nutrients, larval fish and zooplankton. Two drifters were left in place (vicinity of tidal front) so they could collect data as long as possible for comparison with the model (two had been removed the day before). Unfortunately, sustained winds of 30+ kts moved into our work area about 12h ahead of the forecast. This curtailed sampling at the flank station. We completed the CTD/nutrient cast, but none of the other activities. We fished the MOC but destroyed or damaged several nets due to worsening conditions (surging) during the tow, and we decided not to attempt a pump cast (seas were running 8-10' by the end of the MOC tow). Instead, we moved back on bank to recover the two remaining drifters and headed for Woods Hole in steep seas, 12-14' high, from the NW. We arrived at the WHOI dock at 10:00 H, Sunday, April 25. 

Model and Tidal Front Study

In our first occupation of the tidal front station (Station 33, April 19, nominally near CM4 on the Schlitz mooring line), we tried to follow the front as it moved on and off shelf with the tide. Our drifters, deployed a day and a half earlier, were still east ("upstream") of us, having made little progress westward. We used the model to generate a predicted tidal ellipse with "time stamps" on it  to help set up our field sampling schedule and locations. We did this at 15, 30 and 50 m depth. The model agreed with the drifter trajectories in timing, across-shelf excursion lengths, and its prediction of little along-shelf movement. We compared model predictions with hand calculations based on local polar plots of the currents taken from the chart and the time of high water at Pollock Rip Channel. Under the prevailing conditions the two methods were in close agreement, although the model results provide a more objective and precise target location and would incorporate forcing from boundary sea level and winds when those are strong enough to generate local perturbations of the tides. We did not follow the ellipse, but simply projected the times onto a central, across-shelf transect. Subsequent analysis of data from the ship and moorings should reveal how well we did. The across- shelf T and S gradients were not steep enough to tell us much immediately from the CTD data.

We used the model in our second occupation of Station 33 as well. Although we remained near CM4 throughout that day, time-stamps on the model ellipse helped us schedule sampling times so that we could sample various parts of the front as it moved on and off shelf.

The model proved helpful in both approaches.  

 Individual Reports

Physical Oceanography (J. Manning & K. Fisher)

Drifter Deployments

Four GPS/ARGOS drifter deployments were made: a surface and deep drogue on either side of the weak tidal front. A list of times and positions of each deployment is given in the Appendix IV Drifter Log. Temperature probes were installed on each drifter. The shallower drifter (13m) on the mixed side of the front lost its drogue soon after deployment and had to be recovered after only a few tidal cycles. Its surface canister holding the electronics moved rapidly to the east (Fig.4).  For a complete description of the drifter and the attached drogue see earlier cruise reports such as SJ9503. The shallow units on this cluster of deployment were the Brightwater Model 104AVs and the deeper were the newer Brightwater Model 115s. The difference is that the newer models do not have the VHF antennae and the drogues are 1mx10m rather than 1.6mx6m.  The drag ratio of the deep to surface is designed to be 40:1 so that there should be no difference in their water following characteristics.  The three remaining drifters were advected for at least a few days and, in one case, for over a week. These deployments (Fig. 5) together with Ted Durbins deployments off the ENDEAVOR, provided a record of velocity  (Fig.6) to be assimilated into the circulation model.   All processed data are posted on the GLOBEC homepage under data|processed|1999|nmfs_drfit.

Shipboard Sensors

Alongtrack data from the R/V Edwin Link's  underway system were used to monitor fluorescence, temperature, and salinity over the course of the survey grid, and while on station.  Apart from transits to the Bank to and from Woods Hole, temperature and salinity were remarkably consistent, with very weak frontal structure throughout the cruise (Fig.7, aprat105.jpg).  By contrast, the fluorescence fronts were striking.  Several wind events occurred over the course of the cruise, which is apparent in both the wind velocity and barometric plots (Fig.8, apratmp105.jpg). Air temperatures hovered between 6 and 8 degrees for most of the cruise.  The short wave radiation sensor seems to have been giving meaningless data streams, and needs to be calibrated.  Finally, the structure of temperature and salinity with respect to the tidal velocities experienced by the ship show the influence of the slight fronts encountered for each tidal cycle during the cruise. (Figs 9a, 9b, and 9c: aprtide6.jpg& aprtide12.jpg & aprtide18.jpg) While on station, these fronts were occasionally advected past the ship, as during the ninth tidal cycle.  Where tidal velocity ellipses seem irregular, the ship was moving, and hence changes in the shape of salinity and temperature ellipses in these plots are mainly due to the ship changing position. Note that the tidal velocities (Fig.10, aprtidv105.jpg)used in these plots are derived from Candela's estimates based on previous GLOBEC ADCP data. 

Acoustic Doppler Current Profiler

It is very unlikely that any shipboard ADCP data will result from this cruise. While Charlie Flagg's DAS acquisition routine was operating prior to sailing, there were episodic loses of the heading and GPS input stream as soon as the ship was underway.  Subsequent to the cruise it was determined that the syncro-to-digital convert card in the instrument's deck box was gradually failing.  The acquisition routines require a consistent and accurate series of GYRO heading and lat/lon strings to accurately calculate the relative velocity.  After discovering that the DAS routine was not getting the navigational information, we switched the acquisition to the older DAS routine but that was not configured for bottom tracking. At the time of this writing, an attempt to resurrect any usable ADCP data from this cruise seems unlikely and not worth the effort.   In fact, given all the maneuvering we were doing with the ship,  the 3-4  drifter time series of velocity  from different areas of the study region were probably better quality than that of the shipboard ADCP.  Given our location in the vicinity of the Schlitz mooring line, we potentially have a good record of the current structure  from the stable platforms of bottom-mounted tripods.

Hydrography

The SEABIRD Model 19 CTD data was processed a few days after our return by Maureen Taylor (NMFS). Several horizontal distribution figures were made with SURFER software including a station map (Fig.1) and temperature/salinity (Fig. 11) as well as the anomalies (Fig. 12) relative to MARMAP.   Cross-sectional plots for the bongo-grid transects 1, 2, 3, and 4 in Figs. 13a-d were redone with MATLAB.  The SURFER package provides good gridding options but limited plot export quality. As seen by the cross-bank transects there was very little structure.  Transects 1 and 4 did depict the slightly cold/fresh water lens on the off bank edge , but otherwise the transects were fairly typical. Later in the cruise smaller cross-bank transects were conducted along the mooring line as in Figure 13e.  Forty-six Model 19 casts were conducted.

The 38 SEABIRD Model 911 cast were processed with an "endcast.bat" routine similar to those used on the Broadscale Surveys, to convert the raw to bin-averaged split cast files. Bottle files were generated as well. While the downcast data was used in all cases, the near-surface values (<5m) were often obtained from the upcast values. This occurred whenever the instrument was not properly equilibrated prior to lowering. As in the case of Model 19 data, very little structure was depicted at any site. Individual profiles of temperature/salinity (Figure 14a cast 1-36) and fluorescence/sigmat (Figure 14b cast 1-36 ) were plotted. The last cast (#38 not included in Figure 14) captured an entirely different water mass at the flank station. The top 20m of the water column contained a cold (5.7)/fresh(32.0PSU) layer. All processed data are posted on the GLOBEC homepage under data|processed|1999|nmfs_ctd.

Satellite  Imagery

Satellite  imagery before and after (Figs. 15) the cruise indicates slight warming on the shelf during the period of the cruise but that slight change is on the same order of the day/night and day/day variations.  The second picture also indicates the beginnings of the tidal front gradient with warmer water being shoalward of the colder mid-shelf water.  There is no indication of stratification buildup.  The big picture including the offbank region (available on the UMASSD web archive) associated with before and after the cruise indicates no significant effects of offshore intrusions in our region of study. 

Real-Time Modelling (C.Werner)

The real-time modeling component of the Project ``Real Time Data Assimilation" (D. Lynch, PI) was tested on the EL9904 cruise. For a web-based description of the project see the http://www-nml.dartmouth.edu/circmods/RTDA/ site. The on-board modeling team was composed of Craig Lewis (Darmtouth College); Jim Manning (NMFS); and Brian Blanton and Cisco Werner (UNC-CH). Shorebased meteorological forecasts provided by Rick Luettich (UNC-CH). Shorebased consultation and troubleshooting provided by Dan Lynch, Chris Naimie (Dartmouth College) and Dennis McGillicuddy (WHOI). Results of the second step in the modeling component are contained in reports of the EL9905 and the EN323-4 cruises.

On the EL9904 cruise, one of the main goals was to test the various components of the Forecast (FCAST) system, including: on-board computing, e-mail/land-link and on-board networking for webpage broadcasting. The comprehensive cruise report is available in postscript format from the http://www-nml.dartmouth.edu/realtime/public_html/EL9904/ site. Key aspects are highlighted next.

• 15 April: model set-up tested with initial condition hydrography for the FCASTs provided by the CMAST broadscale hydrography that ended 8 April 1999. No ADCP data were acquired during the entire cruise period. Computers prepared for computation were an SGI O2 (Dartmouth College) and an IBM laptop Thinkpad (UNC-CH). Webpages were created and links between the shipboard computers established.16 April: first e-mail received from land containing buoy heat flux data, hindcast and forecast winds and boundary elevations for the 14 April to 18 April period.17 April: first FCAST without shipboard ADCP data.18-19 April: no or partial e-mail contact with land allowing for additional webpage set-up and model testing, and runs of the Pseudo-quoddy and drogue3ddt codes.20 April: using Argos drifter data, our first FCAST with current velocity data assimilation was issued. After this experiment, drifter data was used for assimilation.21-22 April: testing of model FCAST parameters with available data sets. It was concluded after several tests (on board and on land) that the inverse solution with 10 Fourier modes, and with a condition number (RCOND) of 0.1 was optimal for the type and quantity of data available.23 April: last FCAST including assimilation of data from 5 drifters was issued.

Results from the 23 April FCAST follow.

• BPE vs Observed. The comparison of the observed position on 23 April 1999 of the two drifters and the best-prior-estimate (BPE) with the release point on 17 April ( i.e., with no assimilation of drifter data) is shown in Figure 16. The agreement between observed locations (the green dots) and the model location is remarkable considering the 6-day length of the run. Note how the model has captured the difference in the direction of drift between the drifter drogued at 35m (drifter 89; top panel) and that one drogued at 13 m (drifter 393; bottom panel).FCAST vs Observed. The comparison of the position of the FCAST model drifter trajectories at the end of the run (26 April 0000Z) for the original release point and time (9-day run; top panels) versus the release reinitialized at the observed location on 23 April (3-day runs; bottom panels) is shown in Figures  17a and  17b. Drifter 89's forecast ending position for the release at the original site (top) appears to be slightly shoalward of the ending position of the forecast from the reinitialized position (bottom). For drifter 93, the ending position for the 9-day run (top panel) is ``downstream" of the 3-day reinitialized run.

Summary

Testing of the FCAST system was successful including aspects of hardware, software and links to land. Experiences and protocols developed during the EL9904 cruise were implemented in the May-June (EL9905 and EN323-4) cruises. The meteorological forcing during EL9904 was relatively weak, and thus the good agreement between the observed and modeled drifter locations verifies the model's ability to capture the circulation forced by tides, hydrography and imposed heat flux during this time period. Additional experiments during EL9905 and EN323-4 provide additional insights into the FCAST capabilities with variable meteorological forcing.

Two other cruise reports with more extensive followup on this project are posted at  http://science.whoi.edu/users/mcgillic/globec/EN323-4/html/en323-4.html and    http://www.wh.whoi.edu/~jmanning/cruise/el9905.html, respectively.  

Ichtho-Zooplankton Studies

Bongo-net Survey (G.Lough, M.Kiladis, E.Broughton)

Twenty-nine bongo tows were made with a 61-cm frame fitted with 333 and 505 mesh nets using standard MARMAP procedures; i.e., double-oblique from surface to within 5 m of the bottom. A SeaBird CTD (Model 19) was attached to the towing wire above the bongo to monitor sampling depth in real time and to record temperature and salinity. The 505 net sample was sorted at sea to provide counts of cod and haddock eggs and larvae. Larvae removed from the bongo-net samples were individually frozen in liquid nitrogen for biochemical analysis ashore. 

MOCNESS Sampling (G.Lough, E.Broughton, M.Kiladis)

The 1-m² MOCNESS with nine 333 mesh nets was used to sample larval fish and larger zooplankton. A total of 25 hauls were taken. Sensors on the 1-m² MOCNESS included downwelling light, fluorometry, depth, temperature and salinity. A Video Plankton Recorder (VPR) also was attached to the MOCNESS frame to record fine-scale zooplankton during the tow. The high magnification camera was set to a field of view of 2.5 x 3.0 mm and the low magnification camera captured a 2.0 x 2.5 cm area.The tow profile for the MOCNESS was nominally 10-m strata within 5 m of the bottom; extra nets were used for special collections. The 1-m² MOCNESS nets typically sampled for 5 minutes to filter about 250 m³ of water.

The MOCNESS sampling strategy was to make four tows every 24 hours at 0600h, 1200h, 1800h, and 2400h.The plankton from the first down profile would be preserved in formalin for gut content analysis. Two nets were samples from 0-20m and 20m-the bottom to be used for biochemical specimens and special samples.

Table 1a and 1b documents the repository of samples from MOCNESS and BONGO nets, respectively.  

Special Collections (E.Broughton, L.Buckley)

Samples for biochemical and age analysis were taken from three 333 mesh, 61-cm bongo nets, twenty nine 505, 61-cm bongo nets, and twenty five 333 mesh 1-m² MOCNESS hauls. All samples were rinsed from the nets using minimal seawater pressure and transferred to buckets containing ice packs. Plankton from nets that were not to be sorted was preserved immediately using 4% buffered formaldehyde in seawater. Plankton samples sorted for fish or invertebrates were picked in seawater filled translucent sorting trays on ice covered light tables. Every effort was made to keep samples cold during processing to delay decomposition. Samples taken for biochemisty (Buckley) were video taped for later measurement using a Zeiss Stemi SV 6 stereomicroscope equipt with a MTI CDD72 high resolution black and white video camera then individually frozen in liquid nitrogen. Larval fish taken for otolith analysis (Burns, Townsend) were preserved in 85% EtOH. 

Table 1a. MOCNESS sample log w/number of specimens acquired.

	Net 0	Net 1	Net 2	Net 3	Net 4	Net 5	Net 6	Net 7	Net 8
Number of Jars	0	25	24	25	29	27	23	0	0
Cod-Buckley	19					3	98	360	325
Haddock-Buckley	13					1	97	310	302
Cod-Burns								86	73
Haddock-Burns							6	142	167
Cod-Townsend								96	51

Table 1b. BONGO sample log w/number of specimens acquired.

	333 net	505 net
Number of Jars	29	29
Cod-Buckley	41	478
Haddock-Buckley	36	420

Video Plankton Recorder (MOCNESS mounted) (G.Lough, Betsy Broughton)

The Video Plankton Recorder, an underwater imaging video microscope, was mounted above the net opening on the 1-m² MOCNESS. This particular system was held in four underwater housings and consisted of two Hi-8 Video Camcorder interfaced with a Tattletale Computer Software, a low (5.6x) and a high (72x) magnification cameras, a strobe, and a 24V-8amp Gel battery pack. Operation was independent of the MOCNESS. Recordings were later dubbed to SVHS tape format together with time code. Recordings were made for all five 1-m². All in-focus images will be identified to the lowest taxon possible. 

Micro-Zooplankton

Pump Sampling for Small Zooplankton

CTD/pump casts followed the MOC1 tows so that we can examine the prey fields for larval fish and obtain complementary zooplankton distributions with smaller mesh size (40 m vs. the 150 m of the MOC) and finer vertical resolution (5 m in the upper 40 m, 10 m intervals below that).

Information on depth distributions will be used in conjunction with models and other GLOBEC findings relative to larval feeding and growth and cross-frontal mixing of various species and stages. We sampled 14 profiles with a total of 123 samples using a gas-driven diaphragm pump on deck. Generally, 3 profiles were obtained each day of sampling: once each in the forenoon, mid afternoon, and after sunset. Sample volumes were measured to the nearest deciliter using a recording in-line flowmeter. Samples averaged 30 l per depth which was subsampled from a larger flow of 0.3 m³/min for an average of 2.3 minutes. Samples were formalin-preserved.

Table 2. CTD/pump casts EL9904

Sampling was done at cruise station 30 (deep end of Schlitz mooring line) on two separate days and at the shallow end of the same line one day. Two days were spent in the tidal front, although this was difficult to define precisely due to a lack of stratification (see physical description, this report). On the first day we used the Dartmouth model to give us a tidal ellipse and we followed the prediction on and off bank according to the times of day that we were sampling. This prediction fit closely with a polar coordinate prediction using the local published tidal vectors and time of tide at Pollock Rip Shoal, and with the ellipses being made by the drifters which were to our east. The second time we occupied a fixed station and let the tide pass beneath us.   

Nutrients  

Nutrient samples were taken for Dave Townsend at the second occupation of the frontal site, cruise station 33, when we remained at a fixed location and let the tide pass beneath us. Five profiles were taken which included the top and bottom of the tidal excursion. Additional profles were taken at cruise station 30 (mixed site, shallow end of Schlitz mooring line) and at a flank station (cruise station 36, in 86 m water). Seven profiles, 44 samples frozen.

Biochemistry (L. Buckley, J. Burns)

As previously described in the special collections section, a total of 2,597 cod and haddock larvae were collected for biochemical analysis from the bongo-net survey hauls and extra net profiles of the 1-m² MOCNESS hauls. Species distribution was almost even: 51% cod, 49% haddock. The larvae will be analyzed for their RNA, DNA, and protein content and the data used to determine the growth rate and nutritional condition of the individual fish. A comparison will be made of fish taken from the different sites and at discrete depths. A subsample of 107 larvae will be shipped to Dr. Mike St. John at the Danish Institute for Fisheries Research for lipid analysis.  

Appendix. I. Event Log

Appendix III. List of Figures

Figure captions for EL9904 cruise report with plot filename and initials of illustrator.

Figure 1. Bongo haul station positions (stations.ps, MK).

Figure 2. Cod and haddock larvae (#/100m3) catch for April 1999 bongo hauls (codhad.ps,MK).

Figure 3. Cross-section of contoured temperature (sec1.ps,JM).

Figure 4. Trajectory on the drifter (093) which lost its drogue (a093a19.ps, JM).

Figure 5. Trajectories of our four drifters and Durbin's 3 drifters all deployed in April 1999 (alldrftwd.ps, JM).

Figure 6. Velocity time series generated by five drifters fro model assimilation (apr23m3d.ps, JM).

Figure 7. Hull mounted sensor data and anemometer data recorded on EDWIN LINK (aprat105.ps, KF).

Figure 8. Meteorological data recorded on the EDWIN LINK April 1999 (apratmp105.ps, KF).

Figure 9a. Structure of temperature and salinity data from hull-mounted sensors with respect to tidal velocities. Tidal velocities are estimates as computed from Candela empirical relations of phase &amp; amplitude. (aprtide6.ps, KF).

Figure 9b. Structure of temperature and salinity data from hull-mounted sensors with respect to tidal velocities. Tidal velocities are estimates as computed from Candela empirical relations of phase &amp; amplitude. (aprtide12.ps , KF).

Figure 9c. Structure of temperature and salinity data from hull-mounted sensors with respect to tidal velocities. Tidal velocities are estimates as computed from Candela empirical relations of phase &amp; amplitude. (aprtide18.ps, KF).

Figure 10. Tidal Velocities over the period of the cruise as estimated by Candela method (aprtiv105.ps , KF).

Figure 11. Seabird Model 19 temperature and salinity distribution (edlkts.ps, MT).

Figure 12. Seabird Model 19 temperature and salinity anomaly (relative to MARMAP) distribution (elkan.ps,MT)

Figure 13a. Cross-sectional plots of Seabird Model 19 data along bongo transect #1 (xs1tsdjm.ps, JM).

Figure 13b. Cross-sectional plots of Seabird Model 19 data along bongo transect #2 (xs2tsdjm.ps, JM).

Figure 13c. Cross-sectional plots of Seabird Model 19 data along bongo transect #3 (xs3tsdjm.ps, JM).

Figure 13d. Cross-sectional plots of Seabird Model 19 data along bongo transect #4 (xs4tsdjm.ps, JM).

Figure 13e. Cross-sectional plots of Seabird Model 19 data along the mooring line (xs5tsdjm.ps, JM).

Figure 14a. Profiles of Seabird Model 911 temperature and salinity cast 1-36. All profiles are plotted to the same scale. Note the range of temperature is 6.0 to 6.5 degrees. The range of salinity is 32.5 to 32.6. (pro136.ps, JM).

Figure 14b. Profiles of Seabird Model 911 fluorescence and sigmat cast 1-36. All profiles are plotted to the same scale. Highest values of fluorescence occurred at cast 13,14,15, and 36. Sigmat values varied by little more than 0.1 sigmat-t unit ( pro136fd.ps, JM).

Figure 15. Satellite imagery before (left panel) and after (right panel) the cruise plotted from raw CMAST ".gbs" files (Apr8_28_1999.ps, JM).

Figure 16. Comparison of BPE model drifter locations (full drifter paths) and observed locations after 6 days (dots). Isobath in meters (Drifters_BPE.eps, CW).

Figure 17a Comparison of FCAST model drifter \#89 location after 9 days on 26 April 0000Z using original release site (top panel) and a 3-day run reinitialized to observed position on 23 April (bottom panel) (full drifter paths) and observed locations after 6 days (dots). Isobath in meters (Argos89_FCAST_orig_latest.eps, CW).

Figure 17b. Comparison of FCAST model drifter \#393 location after 9 days on 26 April 0000Z using original release site (top panel) and a 3-day run reinitialized to observed position on 23 April (bottom panel) (full drifter paths) and observed locations after 6 days (dots). Isobath in meters (Argos393_FCAST_orig_latest.eps, CW) .

  Appendix IV. Drifter Log