CS5r. (jOSl: Oq^ i^^Sfc^/ Pacific Marine Environmental Laboratory Oceanic Internal Waves from Ship, Aircraft, and Spacecraft: A Report on the New York-to-Bermuda Remote Sensing Experiment U. S. DEPARTMENT OF COMMERCE NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION ENVIRONMENTAL RESEARCH LABORATORIES SEATTLE, WASHINGTON fC- OCEANIC INTERNAL WAVES FROM SHIP, AIRCRAFT, AND SPACECRAFT: A REPORT ON THE NEW YORK-TO-BERMUDA REMOTE SENSING EXPERIMENT John R. Proni John R. Apel* H. Michael Byrne* Ronald L. Sellers Fred C. Newman Ocean Remote Sensing Laboratory Atlantic Oceanographic and Meteorological Laboratories Environmental Research Laboratories National Oceanic and Atmospheric Administration Miami, Florida 33149 *Present address: Pacific Marine Environmental Laboratory Environmental Research Laboratories National Oceanic and Atmospheric Administration Seattle, Washington 98105 a. o o a to CONTENTS I. INTRODUCTION 1 A. Background 1 B. Purpose 2 C. Method of Attack 4 II. CRUISE CHRONOLOGY 7 III. INTERNAL-WAVE AND GULF STREAM OBSERVATIONS FROM SHIP AND SATELLITE 9 A. Landsat-1 Images for June 12 and 13, 1974 9 B. Ship Data for June 20, 1974 10 C. Observations of the Northwest Wall of the Gulf Stream from Ship and Satellites on June 29 and 30, 1974 10 D. Measurements of Oceanic Internal Waves, June 30, 1974 .... 12 1. Expendable Bathythermograph and STD Records 18 2. Interpacket Separations (Data of June 30, 1974) 19 E. Measurements of Oceanic Internal Waves on July 1, 1974 .... 21 1. The Triangle Maneuver 21 2. The On-Station Measurement 26 3. Correlation Between Surface Slicks and Internal Waves . . 27 F. Oceanic Internal-Wave Observations, July 17, 1974 29 IV. ANALYSIS OF INTERNAL-WAVE CHARACTERISTICS 32 A. Packet Characteristics 32 1. Amplitude and Wavelength Variations 32 2. A Model of Amplitude Variations 33 B. Front or Surge Interpretation 34 V. SUMMARY AND CONCLUSIONS 39 FIGURES 43 REFERENCES 104 APPENDICES 107 Digitized by the Internet Archive in 2012 with funding from LYRASIS IVIembers and Sloan Foundation http://archive.org/details/oceanicinternalwOOatla I. INTRODUCTION A. Background The Ocean Remote Sensing Laboratory (ORSL) has been studying internal waves using remote sensing techniques employing three different types of ob- servational platforms: ships, aircraft, and spacecraft (Apel et al . , 1975a; Proni and Apel, 1975; Apel et al . , 1975b). Internal waves and their manifes- tations have been observed using the following techniques: (1) Satellite mul tispectral scanning imagers (principally in the visible and near-infrared); (2) Radar--both coherent imaging radar and standard meteorological radar (all from aircraft); (3) Hand-held visible photography (from spacecraft, aircraft, and ship; (4) Ship-towed thermistors; (5) STD and XBT casts. The first internal wave observations made by ORSL were those obtained using Landsat-1 (formerly Erts-1) images. The internal waves manifested them- selves through their ocean surface signatures which were detectable using digitally or photographically enhanced images. Apel et al . (1975a, figs. 2 and 3) show very early unenhanced and enhanced Landsat-1 images in the area of the New York continental shelf. The standard Landsat-1 data presentation is not designed for the study of oceanographic effects. It was discovered early on that enhancement of contrast in the intensity range of the ocean image brings out subtle surface features, such as internal -wave signatures (Apel et al . , 1976). Apel et al . (1975a and 1976) present a discussion of some enhancement techniques investigated in this laboratory. These two refer- ences and the reference to Apel et al . (1975b) are appended to this report. The study of enhancement techniques is continuing at a special image-processing laboratory constructed at the Pacific Marine Environmental Laboratory, NOAA. Several facts emerged from the early Landsat-1 observations: (1) The internal waves were being detected as a consequence of their surface signature, which is apparently a modulation of the surface optical reflectivity. (2) The internal waves occurred in groups or packets. Along the U.S. continental shelf the surface signature of these packets consists of several oscillations with individual wavelength 200-1000 m, the total packet width being 3-5 km. (3) Along the U.S. continental shelf the packets are separated by distances that are on the order of either 15 or 30 km, which, together with the esti- mated speed of propagation, suggests either semidiurnal or diurnal origin. (4) As the group progresses up the continental shelf, there is evidence of a continued lengthening of the waves, especially at the front of the packet. (5) Of the four spectral bands on Landsat-1, bands 5 (0.6-0.7 ym) and 6 (0.7-0.8 m) have proven most useful in identifying internal wave signatures. After these early deductions were made concerning the existence and characteristics of internal waves, it became clear that a "sea-truth" experi- ment was necessary in order to prove the wave hypothesis. It was, after all, an assumption that the periodic features seen in Landsat-1 images were basi- cally surface slicks associated with an underlying field of internal waves, as described, for example, by Lafond (1962). (Examples of earlier ship and aircraft observations of internal waves are shown by Apel et al . , 1975a, fig. 1.) It was decided to undertake a research cruise to verify that indeed an internal - wave field underlay the satellite-observed signatures. B. Purpose The New York-to-Bermuda Remote Sensing Experiment (NYBERSEX) had as its prime objective the verification of the internal -wave interpretation of periodic features observed in Landsat-1 imagery. Specific observational conclusions from the Landsat-1 imagery were drawn regarding internal-wave characteristics. The characteristics to be verified by shipboard observations follow: (1) The presence of an oceanic internal-wave field underlying a satellite- observed periodic surface signature. (2) Change in surface-wave structure (e.g., the presence of "slicks") discernable from shipboard occurs with the passage of an internal - wave field. (3) The accuracy with which internal -wave characteristics were mirrored in the Landsat-1 observations. Characteristics of the internal waves include the following: (a) Internal waves occur in discreet, well-del ineated packets; (b) Wavelengths present in the packets are on the order of 200- 1000 m; (c) Lateral dimensions of a packet are on the order of 2-4 km in the direction of propagation and on the order of tens of kilo- meters or more in the direction along the crests (i.e., perpen- dicular to the direction of propagation and in the plane of the ocean's surface); (d) Interpacket separation is appropriate for either diurnal or semidiurnal excitation, or both; (e) Wavelength decreases with distance from the front to the back of the wave packet; (f) Orientation of constant phase surfaces--i .e. , a crest or trough-- is roughly parallel to the local isobaths (i.e., topographic orientation) . (4) In addition to these well-defined satellite data verification goals, additional basic research goals were formulated. These included the following: (a) Determination of the wave amplitude distribution underlying a group of satellite-detected slicks; (b) Determination of time and location of origin of the wave groups; (c) Determination of whether high-frequency internal waves exist as packets in the deep ocean and, if so, whether they generate a surface slick signature; (d) Determination of the basic physical characteristics of the water column, such as the vertical density profile, in order to deter- mine the speeds expected for internal waves. These same data will be used to establish the basic internal-wave mode structure present. Of course allowance was made for the possibility of unexpected or unantici- pated data or events, and the chief scientist was given the freedom to pursue events of significance at his discretion. C. Method of Attack The most obvious requirement for verification of satellite data is that a ship be present within the satellite viewing area (or scene) gathering sea- truth data while the satellite records its multi spectral scanner data. The near-polar orbit of Landsat-1 was such that, after completing 14-13/18 orbits ewery 24 hours, its viewing area shifted 1.42° to the west. This slow westward drift thus allowed a ship to proceed westward synchronously with the spacecraft and be in its field of view once each day. A description of the Landsat-1 orbits and swath widths is given in Section II. In order to present as many opportunities as possible for the sea-truth ship to be within the satellite viewing area, it was decided that the vessel should begin at Bermuda and proceed northwestward at such a rate as to remain within each subsequent day's Landsat-1 overpass viewing area. Because one of the major in-water instruments to be used for mapping the internal-wave field at depth was a high-frequency acoustic echo-sounder, it was decided that a sailing vessel would be useful because of lower noise en- vironment. A calculation indicated that 12- to 15-dB increase in signal-to- noise ratio at 20 kHz could be expected under sail as compared with being underway on engine; this increase subsequently proved to be 22 dB, as the data below show. While a sailing ship is somewhat subject to the vagaries of wind, an auxiliary-powered vessel can circumvent much of those troubles. Therefore, a 30-m auxiliary staysail schooner, the RV Westward , was chartered for a 6-week interval. The Westward appeared capable of meeting a schedule that began with a Landsat-1 overpass of the New York Bight on June 12, 1974, followed by two successive 8-day drifts from Bermuda to New York, each a part of the fundamental 18-day Landsat-1 repeat cycle. Two different towed devices were used to gather data on the subsurface internal-wave field: a thermistor and an acoustic echo-sounder. These instru- ments provided two continuous data records of the internal-wave field of quite different characteristics. The single thermistor utilized could, of course provide information only on the temperature at the depth at which it was situ- ated. The acoustic device provided data at different depths (in principle, at any depth at which a sufficient concentration of acoustic reflectors existed) and gave a direct measure of internal-wave amplitude, encounter frequency, and packet length. Non-towed instrumentation included expendable bathythermographs (XBT's) which recorded the water-column vertical temperature profile, salinity-temperature- depth (STD) records which enabled the construction of water-column density profiles, and water-surface temperatures. Hand-held cameras were also utilized. including a Hasselblad 70-mm remote-sensing camera, a 16-mm movie camera, and a 35-mm reflex camera. In addition to Landsat-1 and the Westward , a final platform included was a NASA U-2 which had been scheduled to operate on July 18 and 19, 1974, but because of heavy overcast weather conditions did not fly until July 21. The U-2 provided 70-mm, four-channel Vinton visible and near-infrared photography with filters identical to those on the Landsat-1 multispectral scanner and also five-channel multispectral scanner data from a Nimbus G prototype instru- ment. The planned objective was for the U-2 to obtain data from similar in- struments at aircraft (20,000 m) and satellite (900 km) altitudes simultaneously, and, at the same time, for the Westward to provide in-situ sea truth. This objective was not achieved because of the delay in the overflight mentioned above. II. CRUISE CHRONOLOGY 'The New York-to-Bermuda Remote Sensing Experiment began on June 11, 1974, and terminated on July 21, 1974. The cruise was divided into two legs: Leg 1 with John R. Apel as chief scientist and Leg 2 with John R. Proni as chief scientist. The track of the ship for Leg 1 is shown in figures 1-4. After departing Woods Hole on June 11, the vessel proceeded to the head of the Hudson Canyon and was present in the Landsat-1 viewing area on June 12 and 13. A failure of the main generator precluded gathering any acoustic or thermistor data, however, and forced the return of the Westward to Woods Hole. However, useful XBT and photographic data were obtained, and Landsat yielded imagery showing internal-wave slicks on June 12. On June 13, a strong northeast breeze eradicated surface signs of the waves at both deck height and satellite altitude. The vessel departed Woods Hole again on June 14 and proceeded to Bermuda, taking an XBT section along 71 °W in the process. After a 3-day layover at Bermuda, the Westward departed for New York. Plans called for the vessel to proceed along a northwesterly course at a speed that would maintain the ship within the Landsat-1 viewing area for a series of 8 consecutive days. Figure 2 shows the actual course of the Westward from Bermuda to New York from June 24 through July 1. The circles indicate the location of the ship at the time of the satellite overpass for each day. The series of sloping rectangles shows the satellite viewing area on the day indicated in the lower left-hand corner of each rectangle. The Westward encountered unavoidable delays due to weather and equipment problems that prevented it from keeping the schedule with the Landsat-1 overflights until near the end of Leg 1. The loss of syn- chronism resulted in ship data that were gathered nearly simultaneously or concurrently with Landsat-1 only on June 30 and July 1; tracks for these two days are shown in figures 3 and 4, respectively. An important series of shipboard measurements on an internal -wave packet took place during a maneuver executed between 1700 and 1950 UT on June 30. This consisted of three nearly parallel traverses through the same internal - wave packet that yielded data discussed in detail below. The location of these runs is indicated in figure 3. On July 1, 1974, a triangle maneuver and an on-station experiment on a wavepacket took place. The locations of these events are shown in figure 4. Data from these two days proved to be of considerable value in supporting the internal-wave interpretation. Leg 2 began with the departure of the RV Westward from Floyd Bennett Field on July 4, 1974. The vessel proceeded directly to Bermuda, gathered some acoustic data along the way, and arrived on July 9. Leg 2 was designed to be a repeat of Leg 1, with the Westward again attempting to be within the field of view of Landsat-1. Figure 5 shows the ship's course from Bermuda to New York from July 12 through July 19, 1974. Again, circles indicate the position of the ship at the time of the satellite overpass for each day, and the series of sloping rectangles gives the satellite viewing area on the day indicated in the lower left-hand corner of each rectangle. Note that the ship was within the satellite viewing area on July 17, 18, and 19. Unfortu- nately, cloud coverage was very high and most of the satellite pictures could not be used. The data gathered on July 17 have proven to be useful; they provide nearly simultaneous satellite and sea-truth data. The ship tracks for July 17, 18, and 19 are shown in detail in figure 6. A final stretch of data was gathered on July 20 and 21 as the ship went from New York to Woods Hole, and this track is also shown in figure 6. The arrival of the Westward at Woods Hole on July 22, 1974, completed the seagoing phase of NYBERSEX. III. INTERNAL-WAVE AND GULF STREAM OBSERVATIONS FROM SHIP AND SATELLITE A. Landsat-1 Images for June 12 and 13, 1974 On June 12, Westward was in the vicinity of the Hudson Canyon during the Landsat-1 overpass (see fig. 1) and observed weakly developed, somewhat inco- herent slick patterns; the satellite images for this date are shown in figures 7 and 8 and illustrate the same features. The ship was in the area shown in the Landsat image, figure 8. Little supporting data were obtained from the ship because of the generator failure mentioned earlier; however, the widths and intervals between slicks were timed for a few of the more coherent groups. A histogram for slick widths observed during passage through an in- ternal wave group is given in figure 9. Note that there is a definite upper limit of order 125 m observed for the slicks' widths. Such a limit is expected to exist if these slicks are indeed produced by an underlying internal -wave field. More will be said on this matter in Section IV. On June 13 no slicks were observed at the time of the Landsat-1 overpass because of the presence of a fresh, 7-8 m/s northeast breeze. The concurrent satellite image is shown in figure 10a and 10b, and the effects of the wind are clearly visible on the photographs. Only a small area south of eastern Long Island, a lee shore, shows any hint of internal-wave slicks. However, by approximately 2100 UT (1700 EOT) the wind had decreased to near-calm conditions and in 30-45 minutes periodic slicks began to reappear. By sunset, extraordinarily uniform bands were observed as the ship proceeded parallel to the south shore of Long Island on its return to Woods Hole for repairs. This observation, together with the more quantitative data presented below, suggests that internal waves on the shelf can generate a surface signa- ture within two or three periods of oscillation. 10 B. Ship Data for June 20, 1974 An interesting series of data was gathered approximately 100 nmi from Bermuda at 33°40'N, 65°20'W as the ship was proceeding toward that island at a speed of approximately 3.5 m/s. The acoustic echo-sounder revealed the presence of both short-wavelength and long-wavelength oscillations in the upper layers of the water column. The short-wavelength oscillations appeared to be centered near a depth range of about 20 m. The short oscillations occurred in narrow bursts, with a burst length of about 2000 m and a wavelength typi- cally on the order of 100-200 m. The acoustic record is shown in figure 11a. At 0530 GMT, a series of oscillations of extremely large amplitude was encountered, as shown in figure lib. These were the largest internal waves detected on the entire cruise, with a peak-to-trough range of approximately 35 m. Their wavelength was about 500 m, and they were concentrated in an iso- lated packet whose extent as projected along the ship path was nearly 7 km. Little activity (outside of the short-wavelength oscillations above) was ob- served either before or after this packet was encountered. Because these data were taken at night, it was not possible to determine if surface slicks accom- panied the packet. It may be noted that there is a distinct asymmetry between the widths of the "crests" and of the "troughs." This wave group is discussed at greater length in Section IV, where use is made of certain general features of the wavepackets to expand upon the interpretation here. C. Observations of the Northwest Wall of the Gulf Stream from Ship and Satellites on June 29 and 30, 1974 Between approximately 0900 and 1100 UT on June 29, RV Westward traversed the Gulf Stream front from southeast to northwest as evidenced by XBT, towed thermistor, and echo-sounder data; the location of the maximum geostrophic tilt was approximately 37°42'N, 69°50'W. Traces from XBT's Nos. 76-79 are n shown in figure 12 (d,b,c,d). These data are plotted as isotherm depth contours along the ship track in figure 13a. The 18°C isotherm at 200 m depth may be taken as the position of the Gulf Stream boundary; it was passed at about 0830 UT. Figure 13b illustrates the echo-sounder record for a portion of the same time period; the sloping acoustic reflecting layers are quite visible. If it is assumed that the reflectors are biological in origin and tend to accu- mulate along constant density surfaces at a given time of day, then the slopes of the reflecting layers may be taken to be the same as the slopes of the isopycnal surfaces at the Gulf Stream boundary. Using this interpretation for the interval between 1100 and 1150 UT, the acoustic data yield a cross- stream (normal) slope. (1^) ~- 9.4 K 10-3 , acoustic (s) where n = depth of an isopycnal and n = normal range. Over the same interval, the 18°C isotherm has a slope, - 5.3 X 10-3 _ temp Because the relation of temperature to salinity (thus of temperature to density) changes across the Stream, the slope derived from the acoustic echoes (assumed to come from isopycnal surfaces) is believed to represent a geostrophic tilt more closely than isotherms do. The corresponding speed of the baroclinic current is 150 cm/s at 100 m depth. This is in general accord with histori- cal speeds derived from oceanographic sections across the Stream near this location. At the Stream edge were many narrow surface slicks having some appearance of periodicity, along with short-wavelength, small -amplitude oscillations in 12 acoustic reflecting layers. It is not known how internal waves differ from those described above. Satellite data were not obtained over this portion of the Stream on June 29 because of cloud cover. However, on June 30, 1974, the region to the imme- diate southwest of the Gulf Stream crossing was imaged by scanners on both Landsat-1 and NOAA-4 satellites. In this region the Stream meander period is of order 40 days and the daily variation in position is roughly 10 km. Therefore, the satellite data on June 30 are thought to be representative of the Gulf Stream position on June 29 to within approximately 10 km. Figure 14 illustrates the Landsat-1 imagery (Image IDs 1707-14541 and 1707-14552) and figure 15, the NOAA-4 infrared image for June 30. It is clear that the thermal boundary of the Stream as observed in the IR image is accompanied by a change in the optical reflectivity that is apparent in the Landsat-1 image. The Stream is warmer and darker than the slope water to the northwest. In the Landsat-1 image there were also narrow, highly reflective filaments oriented parallel to the boundary that could be the surface slicks observed from the ship, but this is only conjecture. D. Measurements of Oceanic Internal Waves, June 30, 1974 The data presented in this section were gathered on June 30, 1974, as the Westward proceeded shoreward on a course of 332° starting at 39°49"N, 7r21.5'W at 1251 UT (fig. 1). This area is immediately northeast of Hudson Canyon. The data may be conveniently divided into three regions: Region I, in deep water, seaward of the continental shelf; Region II, on the shelf rise; Region III, on the continental shelf. Region I . In this region the thermistor was towed at a depth varying between 30 and 35 m; the ship's speed was approximately 2.3 m/s; the sea-surface 13 temperature was 21 °C. The temperature at the thermistor (33 m) was approximately 17°C (fig. 16). A definite strong scattering layer is present between 20-25 m depth. At 1315 UT the echo-sounder showed a deepening of the reflecting layer, and at 1320 UT the thermistor recorded an abrupt 2°C increase in temperature at 33 m depth. The crossing of an oceanic front was also marked at 1320 UT. The ship exited from this region at about 1340 UT. For reasons to be elucidat- ed below, this region is denoted as "packet A" and will be treated in a dis- cussion on interpacket separation (Section III.D.3). It is believed that the formation of internal-wave packets begins with the formation of fronts in Regions I and II. At 1358 UT, a dramatic deepening of an acoustic reflection located at a depth of 35 m occurred (fig. 17). By 1416 UT this layer had descended to a depth of 65 m. The rate of change of depth of the reflecting layer with hori- zontal range (i.e., dh/dr, where h = reflecting layer depth and r = horizontal distance) was 0.01. An XBT was taken at 1400 UT. This XBT is shown in figure 18 (a surface bucket temperature reading at 1400 UT yielded 21.0°C). There are three features of interest shown in this XBT trace: (1) a relatively well-mixed layer exists to a depth of 20 m; (2) a layer of cold water exists near a depth of 50 m (the temperature at a depth of 48 m is 14.4°C); and (3) a temperature inversion exists at a depth of 60 m (the temperature at 60 m is 17.1°C). Region II . This region was observed between 1400 UT and 1500 UT (fig. 17). An increase in acoustic reflected energy was observed between 1400 UT and 1500 UT. This was particularly clear in the upper 50 m of the water column; in fact, the entire water column was filled with material capable of causing acoustic reflections. It is not possible to say at this time what the material is; it could, for example, be biological or particulate. Speculation on the nature of the material leads to the suggestion that particulate material is 14 being stirred (perhaps from the bottom) and placed into the water column; some of this material makes its way up into the upper 50 m of the water column and enhances acoustic reflections at certain scattering layers. There is some question as to how much of the material makes its way into the shallow (i.e., upper 20 m of the water column) mixed layer; there are experimental and theo- retical reasons to expect this amount to be relatively small compared to that in the 20-50 m depth range. Admittedly, this interpretation is speculative and requires further experimentation for confirmation; nevertheless, it is consistent with all experimental observations made during NYBERSEX. Note finally the presence of many acoustic reflecting layers throughout the water columns (perhaps made visible by the presence of bottom material). The steepest part of bottom has a slope of .045. The corresponding temperature record (fig. 17) shows a general decrease in temperature at the depth of the thermistor (30 m). There were several encounters with low temperature ranges, e.g., at 1445 UT and at 1457 UT. An XBT was taken at 1500 UT and is shown in figure 18. Note the dramatic decrease in temperature at 48 m depth (10°C) from the temperature at 1400 UT (14.4°C). Region III . In this region well-defined internal -wave packets propagat- ing up the shelf were encountered. At 1530 UT, temperature fluctuations with high wave number were registered by the thermistor. The wavelengths of these fluctuations were, wery roughly, 150 m and less. At 1535 UT, at a depth of approximately 22 m in the acoustic record of figure 19, a horizontal patch of high acoustic return was observed. There were, as well, rapid oscillations in this patch; patch dimensions are 2250 m in the horizontal and 6 m in the vertical, giving a height-to-width ratio of 6/2250 ^ .0026. An XBT was taken at 1630 UT and is shown in figure 18. The two warm temperature spikes shown 15 between 1520 UT and 1525 UT in figure 19 are thought to be part of a developing internal -wave packet (Sections II.D.3 and IV. A). The large increase in tem- perature at 1640 UT is mirrored in the acoustic record by a downward dip in the acoustic reflecting layer (fig. 20). At this point the vessel came upon a well-defined internal wave, denoted by wavepacket B-1 , which was proceeding shoreward in the direction of progress of the Westward . A well-defined series of surface slicks was observed, alined roughly along bathymetric contours. The VJestward was proceeding \/ery closely parallel to the wave propagation direction. From 1700 UT to 1728 UT, a re- markable temperature and acoustic record of the internal -wave packet was made. As the ship progressed through the packet, larger and larger internal -wave amplitudes were recorded. Also, longer and longer wavelengths were encountered as the ship proceeded through the packet. (From data of July 1, 1974, it will be shown that lower and lower frequencies were encountered as well.) Peak-to-peak temperature excursions of 7° were registered by the thermis- tor at the depth of the V-fin (30 m). Peak-to-peak amplitudes of 10 m were registered by the acoustic recorder. By coupling information from the therm- istor and from XBT's, it is possible to make an amplitude estimate from tem- perature information alone. This has been done (Apel et al., 1975a) for part of the present data and will be discussed presently. XBT's taken at 1715 UT and 1728 UT are presented in figure 18. The cross-correlation function of the acoustic and temperature records for wavepacket B-1 (1700-1728 UT), as transformed to distance, has been com- puted and is shown in figure 21. Note that the peak correlation of 0.4 is observed near zero lag. There are several factors which prevent the corre- lation from being higher; two of these factors are as follows: (1) the acoustic reflecting layer and the thermistor were not always at the same depth (This effect will be more evident in data to be presented.); (2) the 16 acoustic record does not contain the high wave-number detail of the temperature record. (Note: Presently this appears to be a limit imposed by the rela- tively crude acoustic apparatus and recording devices used and is not an inherent limitation of the acoustic technique.) Ihe-_acoustic- temperature cross-spectrum is shown in figure 22. Note the prominent spectral peak at wavelengths of 450 m and 300 m. A peak is also present at very low wave numbers; the significance of this peak is discussed in subsequent paragraphs. However, it is clear that, due to the shortness of the record, such long wavelengths, amounting to one-fourth and one-half the length of the data stretch, are only approximately estimated by spectral analy- sis. Indeed, none of the spectra presented in this document have more than two degrees of freedom, so that no statistical reliability is realized. These "spectra" are simply Fourier transforms of the data. The wave number periodo- gram computed from the acoustic record is shown in figure 23. (The record length used runs from 1630 UT to 1728 UT.). The wave number periodogram com- puted from the corresponding temperature record is shown in figure 24. From the cross-correlation function, one is led to expect corresponding spectral peaks at a wave number of 2.2 cycles/km or a wavelength of roughly 450 km. Such a peak is present in both the temperature and acoustic records. Addi- tional corresponding peaks are present at roughly 300-320 m and at 600-610 m. Less agreement is obtained at smaller wave numbers. As indicated in the pre- vious paragraph, this disparity is believed to be due (at least partially) to the difference in depth between the thermistor and the acoustic reflecting layers. The ship and wavepacket were traveling in essentially the same direction. The internal waves thus appear to have a longer wavelength than they actually possess. In other words, the Fourier transforms in figures 23 and 24 are 17 "encounter" spectra which have been Doppler downshifted. To verify this interpretation the ship reversed course, this time proceeding antiparallel to the direction of propagation of the packet, which is now termed wavepacket B-2. This run took place between 1731 and 1805 UT (fig. 25). As can be seen by comparing figures 20 and 25, the corresponding acoustic and temperature traces have many features in common. Figure 26 shows an excellent correlation of 0.85 near zero lag. The acoustic temperature cross-spectrum for this record is shown in figure 27. Note the appearance of the spectral peak at a wave number of 2.9 cycles/km (i.e., A = 350 m) . The individual temperature and acoustic spectra are shown in figure 28. Here again, these records are "encounter" spectra, being Doppler upshifted. Several spectral peaks are indicated in figure 28; for example, peaks occur at wavelengths of 311-318 m, 435 m, and 710 m. It is possible to obtain an estimate of the true dominant wavelength of the internal -wave packet under consideration. By "true wavelength" is meant the non-Doppler-shifted wavelength. It is easy to show (assuming no mean currents) that X-r = 2 T x^ + x_ ' where x^ = the non-Doppler-shifted wavelength, X^ = the up-Doppler-shifted wavelength, X_ = the down-Doppler-shifted wavelength. x^ and x_ are taken from the peak of the respective acoustic temperature cross- Fourier transform. From figure 22, x^ = 450 m and from figure 27, X = 350 m; hence, X-j- - 400 m. Commencing at 1800 UT (fig. 29) the ship began a turn which placed its direction of travel parallel to the propagation direction of the packet. The ship had completed its turn by 1810 UT. From 1810 UT to 1845 UT, the ship 18 once again passed through the packet, now termed wavepacket B-3. The acoustic temperature cross-correlation function for this pass is shown in figure 30. Note that a peak correlation greater than 0.5 is obtained near zero lag. The cross-spectrum for this pass is shown in figure 22 together with the cross- spectrum for the first pass. Transforms for the acoustic and temperature data for this packet are shown in figures 31a and 31b, respectively. From 1840 UT to 1950 UT the RV Westward held station and executed two STD casts. The data from these casts are discussed in the following paragraphs. Two additional segments of simultaneous acoustic and temperature data are shown in figures 32 and 33. An internal -wave packet (from 2155 UT to 2207 UT) is shown in figure 33. In addition, two large zones of warm water, 2215 UT to 2235 UT and 2245 UT to 2300 UT, are shown. Data-taking ceased at 2300 UT. 1 . Expendable Bathythermograph and STD Records The XBT records gathered on June 30, 1974, are shown in figure 18, One of the most noticeable features in this sequence of XBT's is the sharp decrease in temperature at 50 m depth, occurring at the edge of the continental shelf. On the continental shelf there is a temperature minimum (at about 50 m) with a warming as the bottom is approached. Figure 34 portrays a typical density profile (which has been smoothed) observed on the continental shelf during NYBERSEX. Also portrayed in figure 34 is the corresponding Brunt Vaisala profile. The corresponding mode struc- ture is shown in figure 35. The dispersion curves for the first five modes are shown in figure 36. Finally, the phase and group speeds for the first five modes are shown in figure 37. 19 2. Interpacket Separations (Data of June 30, 1974) From the data shown in figures 16 through 33 for June 30, 1974, it is possible to construct a table of interpacket separations. The leading edge of the first packet for that day, packet A, appears at 1345 UT. This packet is not well-formed. In the analysis section it will be argued that the packet shown in figure 16 is in the first stage of development. The second packet (B-1 , B-2, and B-3), makes its appearance with its leading edge encountered at about 1730 UT. This packet is well-formed and will be discussed at some length in the analysis section. The next possible packet occurs at about 2020 UT. This packet (if indeed it is a packet) is weakly developed as an oscillatory structure and is not considered to be an internal-wave group. The next well -formed packet occurs at about 2213 UT and is denoted by packet C. From these three packets table 1 may be constructed. The column entitled "Corrected Interpacket Separation" is computed using the following rationale. The corrected interpacket separations are the distances between packets A, B, and C from figure 3 as corrected for propagation during the ship transit. A wavepacket propagation speed of 0.5 m/s was assumed in order to derive the corrected separations (cf. fig. 37). Corrected separations between A-B and B-C are 25.8 and 25.5 km, respectively, and are to be compared with an expected separation of 22.4 km, assuming semidiurnal generation. An assumed propaga- tion speed of 0.56 m/s would further bring these numbers into excellent coin- cidence: 25.0 and 24.5 km compared with a semidiurnal separation of 25.0 km. Thus the assumed propagation speed need only be in error by 10/^ in order to reconcile the results--an amount that nonlinear effects can easily introduce. Therefore it is clear that the internal-wave packets appear on a semidiurnal and not a diurnal basis. 20 Table 1. Interpacket separations constructed from data of June 30, 1974. Aooarent Corrected Packet ^::r..::. n.-^;:::... mterpacket mterpacket No Time of Time Encounter Difference (UT) (min) Separation , Separation ^(. ^ (assumed group ^^^' speed 0.5 m/s) 1340 1725 2210 220 32.4 25.8 290 34.2 25.5 Predicted interpacket separation, assuming semidiurnal generation and group speed of 0.5 m/s: 22.4 km. 21 E. Measurements of Oceanic Internal Waves on July 1, 1974 There are three prime events registered in the data gathered on July 1, 1974. These are the following: 1. The triangle maneuver; 2. The on-station maneuver; 3. The slick-acoustic measurement. The triangle maneuver is shown in figure 4. The northwest corner of the triangle is located at 39°51'N, 72°38'W. This maneuver is defined to commence at 1525 UT, with a course heading of 115°T. Data were gathered, however, prior to the commencement of the triangle. Two important sections of the pre-triangle data are shown in figures 38 and 39. In section 1 of the data (fig. 38), temperature fluctuations can be ob- served. The possible existence of a poorly formed internal-wave packet is shown between 1320 UT and 1350 UT in that figure. Section 2 (fig. 39) con- tains data obtained between 1430 UT and 1520 UT. This section of data was obtained while running a course antiparallel to the course run during Leg 1 of the triangle. Two clear temperature spikes are observed in these data; one at 1448 UT and one at 1457 UT. These spikes may be part of a poorly de- veloped internal-wave packet. In the discussion of the data gathered during Leg 1 of the triangle, it will be pointed out that these same features may possibly have been observed again later. 1 . The Triangle Maneuver The data gathered in the triangle maneuver established the following facts: (a) The direction of alinement of the crests and troughs of the under- lying internal wave is the same as the alinement of surface slicks, 22 (b) The underlying high-frequency internal-wave field is clearly anisotropic, with dominant wave energy propagating in a direction normal to the crests and troughs as indicated from surface slicks. (c) Characteristic features of the underlying internal wavepacket (e.g., a down-going sharp onset and wave amplitude) are approximately pre- served over the horizontal (i.e., along a slick) direction. This observation carries the implication that high horizontal (i.e., along slicks) coherencies may be expected. (d) The large-amplitude features, generally (but not always) found pre- dominantly toward the leading edge of an internal wavepacket appear to retain their general characteristics as a function of horizontal separation "better" than smaller amplitude, high-frequency oscillations toward the rear of the packet. Lea_i Leg 1 of the triangle maneuver is shown in figure 4 and was defined to run from 1525 UT to 1602 UT. The data from Leg 1 (labeled section 3) is shown in figure 40. Note the large temperature spike beginning at 1538 UT with maxi- mum at 1539 UT. Several spikes follow in the temperature record, and there is a gradual warming trend at the thermistor depth. The time 1539 UT was inter- preted by the authors as being the time of onset of an internal wavepacket propagating up the continental shelf. A surface slick was in evidence at or near this time. This particular wavepacket does not appear to be well -developed, as can be seen by comparing this packet with the packet data obtained on June 30, 1974. It is possible that the wavepacket observed in Leg 1 (commencing at 1539 UT) is the same wavepacket (fig. 38) observed before the triangle maneu- vers and discussed in the preceding paragraphs. In particular, the temperature spike at 1539 UT in Leg 1 corresponds to the temperature spike occurring at 23 1457 UT. In Leg 1 the ship was proceeding against the general direction of propagation of the packet; in the pretriangle maneuver, the ship was proceeding in the general direction of propagation of the packet. There appears to be substantially more "oscillatory" structure to the packet in the second (paral- lel ) pass through. Periodograms and acoustic temperature cross-correlation functions have been computed for Leg 1. These are shown in figure 41. These plots indicate a dominant wavelength of about 350 m. A peak correlation of about 0.7 is observed near zero lag. Leg 2 Leg 2 of the triangle is shown in figure 42 and is defined to run from 1602 UT to 1629 UT. The course heading was 25°T, i.e., generally up the continental shelf. A clearly defined wavepacket is evident in the data, and the packet extends from the beginning of Leg 2 to 1620 UT. Additional oscilla- tions are visible commencing at 1627 UT; however, the authors do not believe that these oscillations are part of the first wavepacket which terminated at 1620 UT. The packet in this leg was \jery similar in form to those observed on June 30, 1974. It is believed that the wavepackets observed in the pre- triangle maneuvers of Leg 1 and in Leg 2 are probably one and the same. In Leg 2, the packet was overtaken by the ship. The spatial extent and orienta- tion of the wavepacket as deduced from these observations are shown in figure 43. Periodograms and the acoustic temperature cross-correlation for these data have been calculated and are shown in figure 44. As was pointed out in the preceding paragraph, the wavepacket seen in Leg 2 is relatively "well -formed," using the data of June 30 as a yardstick. The question may now be asked: Why is the wavepacket in Leg 2 so "well-formed" 24 compared to the pre-triangle packet and the packet in Leg 1? (By 'Well -formed" is meant the persistence of several clearly defined oscillations beyond the first one or two temperature spikes.) There are some theoretical arguments that bear upon this question; they will be discussed in the analysis section. It can logically be argued that since the data in Leg 1 and the pre-triangle packet data are separated by a minimum of 41 minutes (1538 UT-1457 UT) , the packet in Leg 1 represents the time-evolved form of the pre-triangle packet. However, it is difficult for this argument to account for the difference in packet form from Leg 1 to Leg 2. The difference in packet form would surely imply a rapid change. This is par- ticularly true for the form of the latter half of the packet, since the ship did not get "out" of the packet in going from Leg 1 to Leg 2. There is an alternative explanation, however, for the difference in packet form observed between Leg 1 and Leg 2. It may be hypothesized that the ship's track in Leg 1 (as shown in fig. 1) just barely encountered an "edge" of the wavepacket. That is, it may be that the lateral extent (i.e., at right angles to the wavepacket's direction of propagation) of the latter part of the wave- packet is not as large as the lateral extent of the leading portion of the packet. This is, of course, consistent with satellite observations. In Leg 2, the ship passed more nearly in the "center" of the wavepacket. As is discussed in the analysis section, there are theoretical reasons to expect the oscilla- tions in the latter portion of the wavepacket to be weak (or small) compared to those in the leading edge. There are also reasons to expect that the spatial coherency of these oscillations would be small compared to the lead portion of the wavepacket. Such a lack of spatial coherency would lead to dissimilar waveforms (apparently, at least, for the distances involved in Leg 1 and Leg 2) . 25 Leg 3 Leg 3 is defined as extending from 1630 UT to 1715 UT. The ship's course while executing Leg 3 was 250°T. The temperature and acoustic records for this leg are shown in figure 45. Groups of fluctuations (which may be wave- packets) occur from about 1627 UT to 1640 UT and from about 1647 UT to 1710 UT. The fluctuations occurring from 1647 UT to 1710 UT are possibly associated with the wavepacket observed on the earlier legs. If one assumes that the signal occurring at 1647 UT is the leading edge of the wavepacket observed on Leg 2 and shown in figure 42, then one can esti- mate the wavepacket propagation speed. On the basis of the information gathered in Legs 1 and 2, one can determine the orientation of the wavepacket propaga- tion sector to within +11°. By adding the data from Leg 3, assuming that the signal occurring at 1647 UT does indeed originate with the leading edge of the wavepacket, one can estimate the wavepacket 's direction of propagation even more closely. This leads to a northerly direction of propagation (shown in fig. 43) and a wavepacket travel speed of about 0.6 m/s. This speed of travel , if assumed to be a speed of propagation (in still water), is higher than expected from the computations of phase and group velocity shown in figure 37. However, a wery slight current of roughly 0.05 m/s in the general direc- tion of wavepacket propagation may be present (see the on-station discussion), and such a current could reduce the required propagation speed to a compatible value of about 0.55 m/s. A sequence of photos was made of the ocean's surface from shipboard. Figure 46 shows slicks associated with the wavepacket of the triangle maneuver. Visual and photographic observations showed the surface slicks to be propagating in a northerly direction, in confirmation of results of analysis of acoustic and temperature data. 26 2. The On-Station Measurement The on-station measurement is defined to occur from 1733 UT to 2030 UT (figs. 45-48). The Westward's location at the start of the on-station experi- ment was 39°49.8'N, 72°29.0'W. Some interesting features appear in the acoustic and temperature records prior to 1733 UT, however. Note from figure 45 that a white band of little or no acoustic return begins at 1714 UT. Just prior to 1733 UT, large excursions in both the acoustic and temperature records can be observed, particularly the excursion at 1731 UT. These excursions are be- lieved to be associated with the onset of a wavepacket whose time history is shown in figures 47 and 48. This wavepacket made its way past the Westward and occupied essentially all the on-station acoustic record. The V-fin, having dropped to a depth of 45 m upon the ship's taking station, was readjusted to be at a depth of ap- proximately 28 m. This caused the thermistor data to portray only an incom- plete record of the temperature oscillations. Figure 45 shows a second large downward excursion of the acoustic record occurring at about 1735 UT. Note the flattened crests and sharp troughs. In this leading portion of a wave- packet the width of the trough was at most 3 minutes, while the crest width was roughly 7 minutes. The amplitude of the second downgoing oscillation was 20 m, fully one-third of the total depth. This is clearly a highly non- linear condition. As time passed, wave amplitude diminished. The change of frequency with time (the observance of which was one of the prime goals of NYBERSEX) is demonstrated nicely by the spectra in figures 49 and 50. Note particularly the acoustic periodogram (fig. 49, upper left) for the earlier time interval. The peak at 3 cycles/hr is twice the size of the peak at 4.5 cycles/hr. In the equivalent periodogram for the later data interval, the peak at higher frequency (5 cycles/hr) is about twice as high as the peak 27 at 3 rycles/hr. Energy clearly shifts from lower to higher frequencies from front to rear in the wavepacket. This substantiates satellite observations that show wavelength increasing with progression inward from the front of a wavepacket. The overall spectral energy is less in the later than in the earlier interval, indicating diminishing wave amplitudes with time. The presence of a steady current does not affect the conclusion reached above, i.e., the frequency of the waves in the wavepacket increases with distance from the front edge of the packet. The relative amount of energy at the two frequencies will remain the same in the presence of a steady cur- rent. What can change, however, is the absolute value of the frequency meas- ured. This is a well -recognized problem in the measurement of internal -wave spectra (Garrett and Munk, 1972). Indeed, another facet of this basic problem is the distinction of internal waves from relatively stable structures being advected by the current field. In order to rule out the effects of currents, one must know what currents are present; i.e., some current measurements are required. In NYBERSEX, no current measurements were made because of financial and logistical problems. Some information on currents may be obtained, however, from a study of the ship's drift during the on-station maneuver. The times of several fixes obtained on-station are given in figure 4. The error in any given fix is about 0.8 km. From one fix to the next, position changes at a rate that is nearly constant both in magnitude and direction. A straight line fit to the positions leads to a drift speed of 0.11 m/s in a direction 60° west of north (300°T). 3. Correlation Between Surface Slicks and Internal Waves We now turn our attention to the question of the location on an internal- wave cycle of a surface slick. Indeed, one of the central Questions in our 28 investigations is how to define the process or processes by which an internal - wave field manifests itself on the ocean's surface. The answer to this ques- tion proved to be yery difficult to obtain in the field. Slick regions devoid of short capillary waves were indeed observed, but when viewed from a ship at close range became very difficult to identify. From the data gathered on July 17, 1974, presented in the following sec- tion of this article, it is clear that a ruffled region may also simultaneously be associated with internal waves. This possibility is discussed in the fol- lowing section. From shipboard, no definite evidence on ruffled regions was obtained, even visually. However, cat's-paw-like structures were seen on occa- sion and did appear to be affected by the presence of internal waves. One attempt to correlate visual sightings of slick regions with acoustic and temperature recordings of internal waves was made on July 1 , from 2050 UT to 2125 UT. The acoustic and temperature data from this period are shown in figure 51. Also visible in this figure are light vertical lines in the acous- tic record. These lines denote positions at which slicks were encountered by the ship. Start slick and end slick positions are marked by these lines. (Additional lines marking times are also present.) A table has been constructed relating acoustic and temperature internal- wave records with visual shipboard slick observations for the data gathered between 2050 UT and 2125 UT. A slick width interval (as determined by the time spread between the start slick and end slick positions) was observed to encompass a trough or a crest. The results of these observations are pre- sented in table 2. As can be seen from table 2, the observed surface slicks are associated with troughs rather than crests. To be more precise, the observed slicks appear to be more closely associated with that portion of an internal -wave 29 which is rising from a trough. In other words, the surface slicks appear to be "surfing" on the internal waves. However, it must be borne in mind that this result is extremely tentative and is based on a very small data base. More fieldwork is clearly needed. Table 2. Visual slicks and acoustic and thermal waves. Slick No. No. of Crests No. of Troughs 1 1 2 2 1 3 1 4 1 F. Oceanic Internal-Have Observations, July 17, 1974 July 17, 1974, provided one of the few clear days for satellite imagery during NYBERSEX. A portion of the data has already been published (Apel et al . , 1975b). We begin our discussion with figure 52 which is a standard Landsat-1 image (134 km x 184 km) which has been photographically enhanced. Figure 53 is a line drawing presenting several features of interest in this image. Notice the large internal-wave field marked A in figure 53. These waves are judged to be shoreward propagating waves according to two simple tests: (1) the wave-front curvature, and (2) the arrangement of the internal wave wavelengths within the packet (i.e., the longest waves are at the front of the packet). This group of internal waves (marked A in fig. 53) clearly illustrates the presence of successive zones of intense reflectivity-low reflectivity- background reflectivity associated with the internal waves. This phenomenon was mentioned briefly earlier in this report. Figure 52 is a positive print 30 so that the order of intensities is indeed background-high-low-background. Such a sequence is apparently associated with each wavelength within packet A. The zone of intense reflectivity suggests "ruffled" water while the zone of low reflectivity suggests "calm" water. To be more precise, the zone of in- tense reflectivity suggests the existence of diffuse reflection occurring from short wavelength (e.g., capillary range) surface waves. The zone of low reflectivity indicates the absence of such waves. Zones of low reflectivity are expected to correspond to the "slick" lines previously discussed in this article. Now notice the group of internal waves marked B in figure 53. Using the two simple criteria mentioned earlier in this section (i.e., curvature and wavelength arrangement), this group of waves is judged to be propagating away from the shore. Whether these waves are generated by reflection of in- coming wave groups, by topographical interaction of outgoing tides, or by other means is not presently known. A striking feature of internal -wave group B is that the strong intensity variation background-high-low-background appar- ent in wave group A is lacking. Such an arrangement of reflectivities may be present, but if so it is far less discernible, particularly the high re- flectivity portion. This particular Landsat-1 scene (fig. 52) contained such interesting features that the scenes were reprocessed using the Image Analysis Laboratory facilities developed at Atlantic Oceanographic and Meteorological Laboratories (AOML), partially under ARPA sponsorship. This process (basically contrast stretching) resulted in the series of excellent images portrayed in figure 54. Note the clarity with which wave groups A and B (figs 52 and 53) are illus- trated. Note the appearance of additional outgoing and ingoing wave groups north of these groups. Figure 54, in fact, contains outgoing and ingoing 31 wave groups that are passing through each other. The fact that the ingoing and outgoing wave groups appear at essentially the same position suggests that the different reflectivity effect noted earlier does not arise because of sun angle illumination effects. It is interesting to note that the integrity of the internal-wave surface signature is preserved through the internal-wave interaction. Such an integrity- preserving interaction (i.e., elastic scattering) may even be expected if the leading internal waves are solitons. Such expectations are not readily obvious for the corresponding surface signatures, however (even though such expecta- tions may nevertheless be fulfilled). Other features of interest are sand waves, visible in the Nantucket Shoal area, and a pattern of extremely long wavelength (~ 2 km) surface waves, visi- ble on the original computer-enhanced images. 32 IV. ANALYSIS OF INTERNAL-WAVE CHARACTERISTICS A. Packet Characteristics 1 . Amplitude and Wavelength Variations An internal-wave packet on the New York continental shelf is a dynamic entity. This may be clearly noted by comparing figures 20 and 29, which por- tray the same internal -wave packet. All wave packets observed from the RV Westward during NYBERSEX have begun with a sharp down-going oscillation. In addition, a general deepening of isopycnals and isotherms occurs with the onset of a wavepacket. Prior to the undertaking of NYBERSEX, Landsat-1 (formerly ERTS) data indicated (Apel et al . , 1975a) that the longest wavelengths within a given internal -wave packet were at the front of the packet. The wavelength gradually decreased with increasing distance from the front of the wavepacket. It was not clear from the satellite imagery whether it was also true that the largest wave amplitudes were at the front of a wavepacket. It was, however, antici- pated that the wavelength arrangement could be understood by a combination of nonlinear and dispersive effects. First, a clarification of the word "wavelength" is required. The authors of this report believe that nonlinear effects are present, at least in the leading portion of a wavepacket. As is discussed in the following section, it is possible that the first few leading isotherm or isopycnal down-going excursions may be associated with solitary waves or soli tons. In such a case, the term "inter-slick distance" is perhaps better than wavelength. In the linear case, inter-slick distance and wavelength coincide. Some degree of judgment must be exercised in determining both the onset of a wavepacket and wavepacket amplitudes. Consider the first wavepacket to be discussed--the wavepacket in figure 20. The leading perturbation occurring at 1725 UT is 33 somewhat separated in space (by about 1 km) from the rest of the packet. However, this leading perturbation has been included as being part of a wave- packet. The wavepacket, in addition to being made up of short-wavelength (or short-period) internal waves is characterized by a sudden deepening of isotherms or isopycnals. This depression gradually decreases as the packet passes. The authors have measured all amplitudes from the peak of a disturb- ance to an estimated minimum lying at the depth of overall depression. In figure 55 are plotted the experimental data of inter-slick separation versus time (measured from the leading oscillation in the wavepacket). Note the general tendency toward shorter wavelengths (i.e., shorter encounter periods or wavelengths) as one proceeds farther into any given packet. This confirms the satellite observations mentioned earlier. The decrease is by no means monotonic, but a general decreasing trend is present. A second set of data is presented in figure 56. This figure presents a graph of wave amplitude versus distance (or time) into the wavepacket. As in figure 55, the initial oscillation is taken as the time origin. No esti- mate of this amplitude distribution was made from satellite data prior to NYBERSEX. As a general rule, larger amplitudes occur near the leading edge. Note that the largest-amplitude perturbation does not lead in any of the three cases shown in figure 56. 2. A Model of Amplitude Variations The authors of this article have used various acoustic and satellite observations to generate a model for an internal -wave packet. A schematic diagram that summarizes the salient characteristics is shown in figure 57. This figure defines several quantities used in an empirical mathematical model for the amplitude of the packet, equation (1): 34 N(r) = N^(R^{x,y)) W(z) f^ e'^/^^ . (1) {Sin [K^(l +f^) x-f]} Here, N (R (x,y)) describes the peak amplitude and the curvature, R , of the front of the packet; W(z) is the solution to the linear eigenvalue problem for the density profile existing. Three scale lengths are defined: ii is the rise-length of the first perturbation of a wavepacket; £2 is the overall length of a packet; and £3 is the distance over which the wavelength shortens by a factor of 2. Typical values for these lengths from NYBERSEX (on the New York continental shelf) are: for £1, 100 m; for £2* 3-5 km; and, for £3, 2-3 km. It is possible to consider the definition of an additional scale length, £i+, which is the crest-length (as measured along the crest). This additional possible length is made possible because of the large surface area coverage of a satellite image. However, such a length depends heavily on the distance of the crest from the point of origin (roughly circular spreading) and other factors, such as water-mass boundaries, topography, etc. From satellite data, it is clear that crest-length dimensions can be tens of kilometers. B. Front or Surge Interpretation The tidal origin of internal-wave packets is shown by data acquired and displayed within this article, in particular, the acoustic-temperature data of June 30, 1974, and the satellite data of July 17, 1974. The question now is: Through what mechanism does the tide convert a portion of its energy to internal waves? Unfortunately, no current data were gathered during NYBERSEX due to financial and equipment constraints. This lack of data prohibits the authors from stating whether the gradient Richardson number Ri , is greater than or less than 1/4. As is well known (Woods, 1968; Lee and Beardsley, 1974), 35 when Ri approaches 1/4 it is possible to obtain internal waves through a type of shear instability. Several authors (Barnes, 1973; Long, 1955; Prinsenberg and Rattray, 1975) have discussed processes by which tidal energy may be partially converted into internal waves at the continental slope. Lee and Beardsley (1974) have recently shown that tides can form an internal front and generate internal waves through topographic interaction even when Ri is larger than 1/4. These authors propose that internal -wave groups are formed in a three-stage process: (1) the formation of an initial (warm) front by partial blocking of the strati- fied flow by a topographic obstacle; (2) the nonlinear steepening of this internal front; (3) the formation of internal waves due to the interplay of dispersion and nonl inearity. The data from NYBERSEX, in particular the data from June 30, display features similar to those proposed in the Lee and Beardsley theory. The first stage of the L-B process is the formation of a front. Such a front is depicted in figure 16. The location of this front is correct for semidiurnal genera- tion sequence (see Section III.D.2). The bottom depth at 1300 UT is greater than 300 m. The front passes a 2°C temperature rise at 1500 m. The next appearance of a wave group occurs at 1525 UT. Two wave peaks appear to be present. A well-developed wave group occurs at 1725 UT (fig. 20). This is stage three in the L-B process. Consider the wavepacket shown in figure 20: if it is assumed that the leading wave is a solitary wave, then it is possible to estimate the range to the formation point of the internal-wave packet. A solitary wave moving on a density interface at depth ^2 above the bottom is expected to travel at speed c where c = g'(h2 + a))'^ , (2) where a is the wave amplitude and g' is the reduced gravity. In a two-layer model , 36 g' = .g , (3) P2 where q^, p^ = the density of the lower and upper layers, respectively. The values a = 13 and \\2 = 50x10^ cm gives c = 0.8 m/s . (4) The long-wavelength, linear approximation for internal-wave speed is C = g'hi . (5) In our case the thickness of the upper layer is 25-30 m, and c' = 0.50 m/s . (6) The wavepacket shown in figure 20 will thus have traveled a distance, d, where d = F^. • c' . (7) and L is the spatial extent of the packet. For the packet shown in figure 20, L = 4.5 km. Thus, d = ^-Q 3Q^^ ^ X 0.8 ^ 12 km . (8) The distance, 12 km, should be roughly the distance from the point of observation of the wave group (shown in fig. 20) to the "point" of origin of the wave group. The use of the word origin in the previous sentence requires amplification. In that sentence, the word "origin point" refers to that point where both the nonlinear solitary waves and the linear waves evolved. The origin of the front from which the waves are thought to have evolved is farther east (on the continental slope). The distance from the point of observation of the wave group to the point of the front shown in figure 16 is roughly 31 km. This is consistent with the view that the observed wave group did evolve from a front generated at the continental slope. The calculated dis- tance, -12 km, also agrees very roughly (within 100%) with the distance be- tween packets A and B on table 1. This suggests that this figure portrays the area in which a front evolves into a train of internal waves. 37 A major experimental observation made during NYBERSEX was that every internal-wave group observed began with a depression of isopycnals, isotherms, and acoustic reflectors. It should also be mentioned that the open-ocean internal-wave group shown in figure lib also began with a depression. This observation is in accord with shallow-water internal -wave theory as it is now understood. The most elementary equation governing the propagation of finite-amplitude long-wavelength internal waves in shallow water is that of Korteweg and de Vries (1895). Using the work of Berezin and Karpman (1967), Hunkins and Fliegel (1973) pointed out that, for a wide variety of conditions, solitary waves may be expected to evolve from an initial perturbation if i^c > > Zhi^c' , where i is a characteristic length of the initial perturbation, and c is the speed of the initial disturbance. In NYBERSEX, estimates of £, hi, and c' were obtained. A direct measure of c was not obtained. The amplitude of the initial front was measured, how- ever. Hunkins and Fliegel assume that c and c' are roughly the same. This is probably also true of the NYBERSEX data. One argument for this is as fol- lows: The internal-wave groups are separated by distances appropriate for semidiurnal tidal generation assuming a wavepacket propagation speed of c = 0.5 m/s. This includes the location of the front. In order for all the wave energy "contained" within the front to advance to the next semidiurnal posi- tion (which has been identified as the "origin" of the linear and solitary waves), the front must have moved with a speed of roughly 0.5 m/s. It is therefore assumed that c and c' are on the same order of magnitude. It is then only necessary that £2 > > h-2 38 At the most, h^ = 30 m. From the acoustic data of figure 20 and a ship speed =2.3 m/s , 1 ~ 2-5 km; hence, 2 -i- = 4 X 10-^ . Thus the condition for the excitation of solitary waves is satisfied. Let us now return to the question of the isopycnal depression caused by the onset of the internal-wave packet. Hunkins and Fliegel (1973) indicate that, on the basis of energy considerations, only the depression of isotherms allows loss of energy across the packet front. The elevation of isotherms corresponds to the impossible case of energy gain by the packet. From the work of Benjamin (1967) and of Lee and Beardsley (1974), it also follows that internal solitary waves must be troughs. 39 V. SUMMARY AND CONCLUSIONS As stated in Section I.B, the prime purpose of the New York-to-Bermuda Remote Sensing Experiment was the verification of the internal-wave interpre- tation of periodic features observed in Landsat-1 imagery. Equally important was the verification of certain observational conclusions about internal waves drawn from Landsat-1. It was hoped that it would be possible to obtain several Landsat-1 cloud- free images in which the RV Westward would be located, thereby facilitating comparison of satellite and shipboard measurements. This proved wery diffi- cult to do because of the clouds and generally poor weather conditions. The goal of having the vessel Westward in the "middle" of an internal wavepacket during a cloud-free Landsat-1 overpass was not realized. However, a situa- tion did occur on July 17, 1974, in which the Westward was within a Landsat-1 image, though not at the time of overpass, and within an internal wave group. This situation is amply discussed by Apel et al . (1976; attached as an appen- dix to this report) . Let us consider next how accurately the NYBERSEX observations of internal - wave field characteristics verified the characteristics predicted from Landsat-1 data examined prior to the undertaking of NYBERSEX. The characteristics predicted in Section I.B, item (2) and items (3)(a) through (3)(f) were fully confirmed. Goal (1) (Section I.B) was partially attained, as indicated in the above paragraph. Let us consider next the basic research goals (item (4), Section I.B). Before discussing the satisfaction of the research goals themselves, it is worth examining the state of knowledge in the general oceanographic community of the internal-wave field on the U.S. continental shelf at the time NYBERSEX was undertaken. In June 1974 it was well known that internal waves were 40 generated during peaks or turning points in tidal flows (Shand, 1953). By 1969 it was also certainly known that, for the Straits of Georgia at least, distinct internal-wave groups occurred with periods much shorter than tidal (e.g., 30 min) (Gargett and Hughes, 1972). In 1971, Halpern published the results of experiments he carried out in shallow water in Massachusetts Bay. These results again indicated the production by the tide of internal -wave groups made up of short-period internal waves. Halpern (1971) noticed, as had Shand (1953) and LaFond (1962) before him, that long surface slicks were associated with these groups. Halpern 's data showed the rapid onset of the internal -wave groups with a sudden rise in temperature at the recording therm- istor occurring within 10 min. In January of 1974, Lee and Beardsley published an analysis of finite-amplitude internal waves that compared numerical compu- tations of waveform and its development to observations made in the laboratory and in the ocean. According to their interpretation, wavepackets form when the internal front (formed when the tidal current sweeps over a sill) steepens to the point of instability. This, then, was the state of knowledge as de- rived from the published literature in June 1974. Certain questions can now be posed that were essentially unanswered in 1974. These questions include the following: (1) Are internal -wave groups being produced at the continental shelf? The answer to this question in June 1974 was, "Probably yes." (2) If internal-wave groups are being produced at or on the continental shelf, in what numbers are they being produced? The answer to this question in June 1974 was, "Unknown." (3) If internal-wave groups are produced, what is the amplitude and wave- length distribution within a packet? The answer in June 1974 was, "If packets exist, they may be of the form observed by Gargett and Hughes (1972), Halpern 41 (1971), and Lee and Beardsley (1974); i.e., largest amplitudes and largest periods in front and nonlinear." (4) What are the horizontal dimensions (i.e., along a crest) of a wave group likely to be? The answer to this question in 1974 was, "Not enough data at present for a complete response." (5) How does an internal -wave group generate a surface signature and thus become detectable using satellite techniques? The answer would be, "According to LaFond (1962), surface slicks are regions where capillary waves are damped by oily film concentrated at the convergence of horizontal currents induced at the surface by internal waves. According to recent work by Gargett and Hughes (1972), the internal-wave currents interact with surface waves, alternately damping and enhancing them to produce slick and rough bands." (6) Do internal-wave groups appear in the deep ocean and, if so, what is their structure likely to be? The answer to this question is, "Internal waves in the deep ocean have been observed for many years; however, whether these waves exist in the form of nonlinear groups is unknown. In view of the data presented here, it is now fair to say that the occur- rence of nonlinear internal -wave groups at the edge of the continental shelf and shoreward is common along the east coast of the United States. The ampli- tude distribution and wavelength distribution is such that the largest wave- lengths and longest periods are in the leading portion of the wavepacket. This same arrangement was observed by Gargett and Hughes (1972) and by Hal pern (1971). Evidence has been found (data of June 30, 1974) to support the "front- formation" theory of Lee and Beardsley (1974). The horizontal length scale of the packets is large, easily being tens of kilometers. Internal-wave packets do indeed occur in the deep ocean, as evidenced by the acoustic data in Section III.B. The data in Section III.B plus the data from GATE presented 42 by Ostapoff et al . (1975) and Proni et al . (in preparation), indicate a similarity of form between deep-ocean internal -wave packets and continental shelf wavepackets. Finally, the evidence in this document supports the following conjecture; Individual internal waves within a wave group can possess two surface signa- tures: the first is a ruffled area, replete with capillary waves; and the second is a slick area, devoid of capillary waves. One may also surmise that both the leading nonlinear waves within a packet (solitary waves) and the trailing linear waves can generate the above-mentioned dual surface signature, If one were interested in studying the interaction of nonlinear internal soli- tary waves, an excellent place to do it would be about 90 km north-northeast of Cape Cod. 43 O) o o o O O o o S- OO +-> O) . CD ^D CD O O) •.- -l-> -C -C s- o c O 4-> OJ >- -^ S "O 4-) OJ 3 Z E O) O) CU O) .c CO S +-> E (T3 M- O -M O S- fO It- XJ (U O) CD l/l -M O) •>- •-- _J 13 I— S- I— CD O OJ C +-> •I- OJ ro s- x: CO -o o ■o (O 2 OJ E OJ en S- Oi I— c: -c: ■■- +-> fO • c: XJ JD c O E ■a E l/l OJ OJ -r- QQ •1- OJ -r- O fO Q. +-> CO X 4- LU r— oo c 00 OJ fC I— OJ C 5 -t-J -1- CU O 4-) Q. S- E C -r- ZJ OJ O -C cno; o CO 44 CO O) O O) o I— T3 -a O) C7)JZ -M OJ 4-> > O E ro O -a OJ E •■- OJ S- I— S- OJ I— (t3 +-> E TCS E O CO -T— s- S M- CU O) -E •■- Cn+-> > •r- M- CU O O +-> cn •<- (V I— cr- ■o 03 a- S CO O) -C (D +J -E +-> o -a d) E r- o (O ■*-> CO 03 u cu o CO 1 — S- =3 OJ o ^ o +-> OJ • +-> OJ 03 u O) -r- s- -o CJ>T- 4-> 03 • CO O) (U CD -E C 4-> fO +J CO U CD O S- (/) -E CJ CO 03 CD OJ C7)M- E O 03 +-> S- U CU CU E i- S- O CD U E •I- T3 Q. E O 03 1 — JE CO r +-> 4- i+- o ^'^' ,00^ 7I°00 30 40° 00 39° 30 Figure 3. Course sailed by the RV Westward on June 30, 1974 46 ^ r-- CTi • 1 — Q} +-> rt fO 1 ■a >, CO •-0 +-) c c o o -a ■(-> S- :3 fO o 5 +-> -o (/) O) CD • 1 — -^ s_ s- > , = X) <^ O) T3 > -a E cu CO o 4- cn c o en ■o m S -M (/I OJ >> ■o I/) OJ «/) s- o cu CD 48 O O o O O o CO O O o o o o O • o o dJ ro j^ 1^ +-> >^ ja -o O) 1 — "r- « (/) (U (/) s- 3 o o b o «^ o "sT QJ f^ :3 C75 49 Figure 7. Landsat-1 image for June 12, 1974. The eastern tip of Long Island is^visible in this image. The coordinates of the center of the imane are 40°25'N and 72°15'W. Some internal-wave activity is indicated in the lower central portion of the image. The bands to the right of center, with dark and light bands of nearly equal width, are probably clouds: internal grav- ity waves in the atmosphere. 50 S- 3 <^ M- O -o O) S- O) O) o 14- cu CJ) I +-> to ■o c CO OJ 51 O < a: o < LU Q X I- ^ 9 c?5 o 2 o tr UJ X O UJ CO O ^ CO :z UJ I- z: o o CO ID (M ro ro in ro \r) o in Z3 o o J- CD in 0) 3 1 c in CVJ •f— "'"' C (O CD 3 o ^ ^^. O -E o CO a: cn 03 UJ to fO H Q. in UJ f^ S 3 ■o > o in O -M "O •^ 3 in o •r- CJ «♦- o E 2 O +J (/) SN011VAd3SaO dO d3aiAinN 0^ 52 Figure 10a. Landsat-l image for June 13, 1974, showing the New York Bight area The coordinates of the image center are 40°25'N, 73°3rw. 53 Figure 10b. Landsat-1 image for June 13, 1974, immediately south of the Landsat-1 image shown in figure 10a. The coordinates of the image center are 39°01'N, 74°06'W. 54 UNIVERSAL TIME Figure 11. a. Observations of short internal waves observed on June 20, 1974. Ship's location: 33°40'N, 65°20'W. b. A localized group of large-amplitude (as large as 35 m) internal waves observed on June 20, 1974. The crests of these waves approach within 12 m of the ocean's surface. These waves are very similar to those reported by Ostapoff et al . (1975) and Proni et al . (in press). 55 10 -"^ TEMPERATURE {°C) 0800 UT. XBT#76 12(a) 10 20 TEMPERATURE (°C) 1000 UT. XBT#77 12(b) J 30 10 20 TEMPERATURE CO 1125 UT. XBT#79 12(c) 10 20 TEMPERATURE (°C) 1325 UT. XBT#79 12(d) J 30 Figure 12a-d. Temperature as a function of depth according to XBT's taken along the ship track. 56 250 0800 13a SURFACE I- ^ 50 U O 100 150 200 1000 1200 UNIVERSAL TIME 12.8 25.6 DISTANCE (km) T=23°C 1400 38.2 PPP ■1 ■ • ^ 1 _J 13b 1100 UNIVERSAL TIME 1150 Figure 13. a. Isotherms along the ship track according to the XBT data of figure 12, b. Acoustic echo for part of the same period. 57 Figure 14. Landsat-1 image of Gulf Stream, June 30, 1974, 58 'r^frared Tmage for j ^J"e 30, ]974. 59 80 75 70W LEGEND G.S. Gulf Strea^ Sharp thermal gradient Less distinct r>^^^ Clouds thermal front Figure 15b. Analysis based on NOAA-4 infrared images of Gulf Stream for June 28 to July 1, 1974, with position of Landsat-1 image (fig. 14) superposed. 60 M I I I I I I I I I 1 — I — I — I — I — I — I — r o n. o 2 o "^ H UJ O Z UJ 3 CO ^ J I I I L m Hid3a (Do) (*) °'l 3dniVa3dW31 c 1 S !^ -J < CO (C UJ > z 3 T3 +-> 03 fO S- O) T3 Q. s- E ro O) ■5 -M 1 — o •1 — JD to o q: fO (O CD 61 I I I I I I I I I I I I I I I I I I I I I I M I I ' ' I ' I I ' M I I I I I I I I I I ! I I I I 1 — I — r T — I — I — r J L o 55 i lO l_ C E 1 ^ ro 2 3: P S < CO q: iij > z 3 (UJ) Hld3a (Oo)(0°^i 3yniva3dW3i O) 5 -^-' O (O 4-> E s- o Ln +-> ^o 1/1 •r- O E ■•-> s- OJ 1- ^ => -M CX) E U-) O CO S- 1— 14- • +J >> OJ (O tn s_ i- =3 E Ol +J c: fO o <^- =5 O -o s- U C CD (O (O >^ fO ■o r^ O) Q. -M O cn U -M C O) — -' 'r- ^— -M M- O U OJ ^ OJ S- o <— O) M- 4- O) o O S- •r- *^ -M M- -o co O o 13 •r— O cn S- o c: OJ <=C •-- Q. c: OJ C71 • o. C ^-^ CD • 1— O) 0) s- Q.-0 ^ O -a r— OJ 00 ■*-> #* ^-- O O) ■z. to »— 4 e .iT u o +-> c •.- Q- • 1 — CTi OJ r^ ro h- 1 — 1 rD o OJ oo yo s_ r^- 13 4J 'a- cn ro r— 62 O CO (S»313N) Hld3a O O- Q o lo m ^ «-i TEMPERATURE CO 10" 15" ■^ ^ /' to HI CO >- Z O to Q o X < O O (SU313M) Hld3a ■n y S o o LU LU X < O v» GQ X (SU313M) Hld3a o 8 If) ■»-i < z 111 z z o (Sd313IN) Hld3a g I "\ \^ I UJ (Eo < UJ Q. Zo UJ a 1- ^1 /" \ / O UJ a: o 1- "" < ac UJ Q. UJ 5 1- ^ \ /■ \ r -!-> Q. dJ T3 E O IT) %- -a £ c •1— CO OJ M x: o ■M O >?^ T3 C (O c o *^ (U h- s =D 4J Q- < GO CD < o z LU a. (S»313IN HldBQ) ^ O O O o tf) m -H :^ (SU313N) Hld3a (Sd313iN) Hld3a O 8 TEMPERATURE CO 15" 20° ^ J - UJo a. m 3 ^ 1- '< a: UJ l°» UJ 1- \ \ /~ u o 3 "- < q: UJ UJ 1- ^ kf- (Sa3i3W) Hid3a 2 § o m -M to (U OJ +-> x: o +J o CO to (O T3 S- OJ O 4-> O O OJ ^ S- X 00 s- 3 C7^ 63 I I I I I I I I I I I O o 00 1 — I — r ' ' ' I I I I ' ' ' ' I o in o o T — I — I — r J L o fO (0 in CM s ID c lO S v4 CVJ •"" '"^ ""• '^ »-i ,^ »-• «-i ,_, (uj) Hid3a (o«) (0 0^1 3dniva3dW3i llJ < CO q: UJ > f I -o O) S- Q. S- O) 4-> c *■" (U s- O . CM CO un OJ 1 — > to -M s 03 ^— (/I fO OJ c: -i.i S- • 1— OJ C-M 10 c •r~ e S- CD (t3 C 2 1— o. OJ o -C r— 1— a; • ■a <+- to I — fO CD en OJ en CD 64 I I I I I I '-'^r„ i*'** I I I I I I V-/ f\i ^^4 •— * v« ^r-t *— • ^f-« •^ •■^ ^"-^ 8 K O CM lO o o lO o in CD (3 O n3 S- OJ O) -C CL-M E O) -^ ■(-> o o O) +-> +J s_ O O) ^ > o • XI CT> S- 1— rO 2 OJ -M S- U) ZS (U CT>3 CO CM o o o o . S- fO fO •I— CT> D1 +-> C =5 C -r- O O +-> S- o a -c CU +J cC i— M-T3 . S- I— <+- 4- fO 1 — O to OJ .C Q--0 CO •■- c 73 03 I— I x: I— I +J E O o Q. (UJ)Hld3a 0£ (Oo) (0 "u 3aniva3dW3i en QJ O) O > Qc "^ (O VO s r— 1 • 1— o +-> fO oo CO c S- 3 CU c Ol Q. •r— 65 O {D lO ^ ^ro CVJ — d d C ^ d o -+- ^*^^ — {- LJIQ 5 < UJ 2uj §s ujo: trill (T CD 81 O LJ UJ S a:uj a: ujo: Q-O UJ I O I- o o < o (U i- 3 4-> (0 &. (U Q. E (U +J "O c c Q. +-> CD O) cn s 03 o ■ — -M <4- S- O o +-> C to o • 1— E +-> S- o cu c .c rs 4-> 4- >. fO J2i CO ■o 03 S- c o o o •1 — - o CO z Ul o -J < h- o UJ Q. CO X= 3980 m RECORD LENGTH = 7960m 1.3 1.9 2.5 3.1 WAVENUMBER (CYCLES/ KILOMETER) SPECTRAL DENSITY FOR WAVE PACKET ONE FROM ACOUSTIC DATA (SHIP PROCEEDING IN THE DIRECTION OF WAVEPACKET PROPAGATION) Figure 23. Spectrum for acoustic data 68 a: UJ h- CM UJ CO S UJ o UJ _J q: y: iD UJ \ Ci UJ -I o > o CO z UJ o < I- o UJ a. CO 4.5 r 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 -< X = 4524m V-FIN RECORD LENGTH 9049 m X = 43lm X = 250m X=300m .55 I.I 1.65 2.2 2.76 3.3 3.85 WAVENUMBER (CYCLES/KILOMETER) GRAPH OF SPECTRAL DENSITY FOR WAVEPACKET ONE FROM TEMPERATURE DATA 4.4 Figure 24. Spectrum for thermistor data 69 W E X Q. UJ O O o a: a: UJ Q. UJ I- 50 100 18" 17" 16" 15" 14» 13" 12" ir 10* 9» 8* BOTTOM T r T r SECTION FIVE JUNE 30. 1974 j- d. j_ A. J_ i 1740 1745 1750 1755 1800 UNIVERSAL TIME, t(min) Figure 25. Acoustic and thermistor data taken as Westward met the internal -wave packet B-2 head on and sailed through it in the direc- tion opposite to that of previous encounter (fig. 20). 70 CROSS -CORRELATION FUNCTION -20 -16 - -L-OJ LAG NUMBER (ONE LAG = 45m) ACOUSTIC -TEMPERATURE CROSS -CORRELATION FUNCTION FOR WAVEPACKET TWO Figure 26. Cross-correlation between depth of acoustic echo and temperature at towed thermistor recorded during reverse run through the internal-wave packet. 71 8r- (O >- < t m T4 CM HI -J Q. 5 2 2.9 5.8 8.7 WAVENUMBER (CYCLE/ KILOMETER) ACOUSTIC - TEMPERATURE AMPLITUDE CROSS - SPECTRUM FOR WAVEPACKET TWO Figure 27. Power spectrum for cross-correlation function of figure 26. 72 in lO II I ZH Z I > o q: o o Ui DC CM d313tN0niM/310A0 zO S33y93a) AllSN3a "lVyiD3dS 5e _l o m Q ro cc ^ O o UJ a: UJ A t- ^ (J) liJ ih 2 O _j N- ^ '*•■ "^ CO UJ go P q: ro >- t- <.> U) cvi cr UJ z UJ Q CJ 2 < — UJ tr > h- <■ o i UJ O Q. ro -M ro T3 O) i- 13 . 4-> +-> ro (D S- .^ > ro ■D S E ro -e en ro 13 -M O ro S- -D -E 4-> u •r— E +-> 13 CO S- :3 o o O) M- s- ro C7) s- E ■*-> • u S- O) 3 0.-0 00 00 CNJ < CD -JUJ c •1— LJUJ 3 oro: T3 crx OJ Oh- o +-> 1 1- 17^ COUJ E Oo o CO q: r<. a:< •n- Ul ^ oa. cu GQ UJ -E s II iij> T3 3 CD < 'r- 4-> J -ll. o CO u 1 3 t/1 1/1 o o o <_) < o OJ en 75 CN CO UJ UJ q: LlI o cc UJ UJ UJ -I o >- o >- z UJ o < H O Ul 0. CO X=2785m RECORD LENGTH 5570 m X=428m ^\ =326m 1 i-^ .9 1.8 2.7 3.6 4.5 54 6.3 72 WAVE NUMBER (CYCLES/KILOMETER) SPECTRAL DENSITY PLOT FOR WAVE PACKET THREE FROM TEMPERATURE DATA Figure 31a. Power spectrum for temperature data, third pass through wavepacket. CO a: \- 501- UJ I- 100 UJ O 150 24 OH 22 UJ IM 1- Ld 20 (/) ^ u: O L_ -J 18 r UJ ^ 2 UJ _J () 16 >- 14 o > 12 H CO 10 UJ o 8 -J < q: 6 H O UJ 4 Q. CO 2 76 TEMPERATURE (X) 10 15 —[ 1 r 20 I I I I /////// I J. I J L T 1 1 1 1 1 1 T" RECORD LENGTH = 3600m. X=450m X=300m 123456789 10 11 WAVE NUMBER (CYCLES/KILOMETER) SPECTRAL DENSITY PLOT FOR WAVE PACKET THREE FROM ACOUSTIC DATA. Figure 31b. Power spectrum for acoustic data, third pass through wavepacket. 77 o O h UJ 1^ 18° IT** I6« 1 5° 14- I3" 12" II" 10* 9" 8"h SECTION SEVEN JUNE 30.1974 BOTTOM m 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 UNIVERSAL TIME, t(min) 2100 Figure 32. Acoustic and thermistor data taken on the continental shelf. 78 fO E S- O) . +-> . — > c r- •<- ZD c O CD o -i^ CO ro (XI +-> s- o O) +-> > o Ln ^ c OJ fO C\J CDTS c c • 1— fO 2 O h- -E ZD to m 4- ro 1 — CO 0) (XI -E to o +-> ^^ fO ID +J I — c C\J O) CM c — ' • r^ +-> s- c O) o +-> o fO 5 aj -E E 4-> s- (O C s o it- c o cu .^ to fd cu +j c: o ns N +-> fO -o -a C (O S- o ^ — ^ +-> 1— to ro *l— E r-- s_ o O) (XI .c CXI +-> o -a +-> E fO LD LO o c — •1 — (XJ -l-> ■ ■ to 13 +-> o QJ o -^ <: O (O Q. CO O) en > > CJ •r- 10 o OJ T3 >-«^ E (U -E O iiJ -M rj -1 LU E 40 - SCA Q O S- 3 <4- 1- -o OJ ' _J CL o IE < Z3 O Si LU > to CU -a o _| \-z E og < O) _J > UJ M- O cr to CM 1 81 -c: •4-> UJ S- _i o (4- o < o U) (/) c •r— O 4-> OJ c 1 o (U ^ •1— LlJ _ E o — . CM UJ Ln < 6 >- 5 o UJ o UJ MODE I MODE 2 MODE 3 MODE 4 MODE 5 I 2 3 4 5 6 WAVENUMBER CY/KM Figure 36. Dispersion curves relating frequency to wave number for the first five modes with the density profile of figure 34. 82 .5- o LJ if) A- o o _J LU > .3 o CL o tt: 2 < H o LlI < X 1 - 1 1 I" 1 1 N — - \ ^ MODE 1 - c Cg \ \ \ \ \ \ \ \ ^ - MODE 2 ■"" ^ — ^ MODE 3 MODE 4 MODE 5 ■""" " 1 1 1 1 .— L- I 2 3 4 5 WAVENUMBER CY/KM Figure 37. Phase speed (solid) and group speed (dashed) for the first five modes with the density profile of figure 34. 83 O O ro lO CD i^H O) O) o OJ S- § 2 CL CO +J c: •"- iS r^ 1 — lO CTi ^-t *" o "2: o lO r — .^^ » 13 O Q- iT CD lO OJ > ro > fO O) 1 w* C I— E e S- o o) a> ^ 1— -M G CT) C E -r- #' % •r- CU c S- 1— 1 -M JD lO •r— to ^■^ O) CO ro •• J^ CO 4-> O O. UJ O) S s- o O -M o ro 1- ^- ro jd c: ^-« J o < a cl (/) O) CO (T to UJ +-> S- CJ > o ro n3 u w^ z •*-> 3 n3 +-> ■O 03 -E o OJ •<-> CM +-> O) fO OJ CU CL OJ in ro E i- +-> CO • -a c +j C O OJ fO •>- -^ O --~. fC (O v4 Q..— Q. Q o >— to •.- O O Q. in •.- O +-> ^— r- 00 =) (U :3 > O O QJ U LO "O o o CO C (U ro 00 (O c » OJ (U 13 .:^ O) 03 c -M m E fO +-> CD 03 1 — ■o c-i e S- fa o •1 — -t-> s- I/O 4-> • n- E •^ s- «* O) r-- x: en -M <— T3 «\ C 1 — ro >- u ( — •1 — r3 +j •-D to =3 4- O O o < CO "^J- S) ^^ a>^ (UJ) Hld3a oc (0.) (4)''u adnivd3dN3i 88 to s- O) +J =3 o o > LU «/) (U o o C/) 2 \ ' •" \ 1 lU \ f" 9 \ r^ 1 «0 /\ 1 :^ oc z ^ o o 1 Jfi o ' 0. UJ 1 i E (0 o ' C9 UJ UJ O UJ (/) = l^ UJ O) ■a c to s- . O) +-> > O) CD O C fO fo a. E cu > O) fO -— s C O) 03 -C •I- +-> s- O) -i- -c S -M >^ S- -M CU E O OJ CD CO (U Z5 D1 X I- oc o z 89 s NYBERSEX ECHO-SOUNDER 160O-1629Z I JULY 874 12 3 4 5 WAVENUViBER (CYCLES/KM) 10 9 8 7 t 6 z NYBERSEX v-r:N,ecH0-5'X»oc9 CRCS5-AMP1TUK if-ECIRUM 16CC-1629Z 1 JJLT 1974 1 2 1 « S WAVENUMBER (CTCLEl/KM) 3.U 25 — 1 r- — 1 T -t 20 \ Jl NYBERSEX V-FIN I 1600-1629 15 - l\ 1 JULY 1974 ID - - 05 - ,V , V ,\a.<-^^.^ 10 20 30 40 to &0 WAVENUWGER (CYCLES/KM) 1 1 r 1 1 1 r r ■ - • A ■ / \ ■ /\ 1 \ /^ \ / vy "^ "^ v NYBERSEX V FIN. tCiO- SOUNDER CROSS CORRELATION FUNCTION leCO - 1E292 tAll.t'74 \> 1 1 . . L . J 1 1 1 '. • -04 -0 2 02 04 L*e DISTANCE (KILOMETERS) Figure 44. Triangle maneuver. Leg 2, spectral analysis of acoustic and thermistor data, as in figure 41 for Leg 1. 90 IT) "?r ro <\j ri o *-l ^ ■^p-t ^-1 ''^ ^ o o 00 lO ro ^.^ ' N- c ^-1 1 o ro -^ S- O) > 13 T-* IxJ If) 2 (0 E CJ H O) h- ▼-• CD _l < o CO s- +-> CJ or t; UJ > +-> «HIHB 2 4- in 3 o ««-• r) 1^ ^r-« C7) O _1 o o ■(-> to in ■o p i_ N o 4-> t/1 O E o O) h^ -o c in 03 in O 5S •1 — +-> 1/1 :3 o o < in CO ■»-• ID o o o rs- CD lo ai (lU) Hid3a g||^ (Oo)a)°^ 3dnivd3dyM3i 91 -a s- o O S- 4- -a > S- Ol to JD O U rt3 4- S- 13 OJ cn / 92 O o ULLHldBG O •r-l (J) in •9-t ,_^ 3 (T3 o c E o E 0> ^-^ o ^rH "•" lO ** 4-> lO UJ 2 rds dur to 45 m 3 o o o ^ S- O CX) o s- ^ mist as d lO s- ^ 00 O) +-> T3 I/) E -1- o fO E fO (J (U 22 •1- J3 •4-i -M 'f t/) a 3 CM o T3 CVJ m u 4-> O u O) QJ o -1 < ro o o CM OJ E UJ o lO > -M lO ^ 4-> ffi 3 to *^ o o lO OJ 0) ^-1 o 4-> lO tX2 ^ Q. 0) w^ o ^ 0> < 00* OJ J- J C7) 94 -I 1 1 1 1 r- NYBERSEX V-FIN, ECHO-SOUNDER CROSS-AMPLITUDE SPECTRUM 1810 - 1930 IJULY'74 3 6 9 12 15 18 21 24 29 FREQUENCY (CYCLES/HOUR) 3 6 9 12 15 18 21 24 FREQUENCY (CYCLES/HOUR) 1.0 0.9 -1 1 — I 1 1 1 1 r NYBERSEX V-FIN 1810 - 1930 1 JULY 1974 A AL^^^J^^ A lA I 6 9 12 15 18 21 24 27 30 FREQUENCY (CYCLES/HOUR) 1.0 0.8 0.6 0.4 0.2 -0.2 -0.4 -0.6 -0.8 1 1 1 I r NYBERSEX V-FIN, ECHO-SOUNDER CROSS CORRELATION FUNCTION 1810 - 1930Z 1 JULY '74 ■1.0 -Q J- -L -0.4 -0.3 -0.2 -0.1 0.1 0.2 LAG TIME (HOURS) 0.3 0.4 0.5 Figure 49. Spectral analysis of the on-station data shown in figure 47, 95 3.0 2.5 - o 2D - o , 15 -1 1.0 05 _ r- "T T — r — 1 — - 1 NYBERSEX ECHO-SOUNDER 1940- 2028 Z 1 JULY 1974 1 — 1 — 1 \yr^ Ai 1 5 10 15 20 25 30 FREQUENCY (CYCLES/HOUR) 10 — 1 1 1 1 1 9 - - 8 - - 7 - NYBERSEX V-FIN, ECHO SOUNDER 6 - CROSS-AMPLITUDE . SPECTRUM 1940- 2028 Z b - I JULY 1974 4 - - 3 ■ - 2 - 1 - _ \^- 5 10 15 20 25 30 FREQUENCY (CYCLES/HOUR) 05 04 y>, 03 02 01- /I NYBERSEX V-FIN 1940-2028 1 JULY 1974 5 10 15 20 25 30 35 40 FREQUENCY (CYCLES/HOUR) -01 01 LAG TIME (HOURS) 02 3 Figure 50. Spectral analyses of the on-station data shown in figure 48 96 2050 2055 2100 2105 2110 2115 2120 2125 UNIVERSAL TIME, t (min.) Figure 51. Acoustic and thermistor data related to shipboard siting of slicks. Beginning and end of passage of ship through a slick are marked by the light vertical lines on the acoustic record. 98 I ea {/) -a c m +-> O) -M (/) (U o s- M- (U C) fO E O £ c 0) < ^ o o s- •r- M- O CL 2_^ CVJ UJ « 4J > E E •>- o < 4- ^ O 1/1 O -r- C T3 r3 O n3 e CO O (O -r- -M -a u o c •I- 3 S- 4- O) CL 03 OJ to > m CD • c ID (U Lf) r— QJ OJ > J- o ZJ O <+- O S- ■4-> fC s- CD Q. S- E M- CU O O ^- E O -D Q. E CT5 O lO o ~ Oo) 3anindiAiv 3dniVd3diAi3i 103 w HORIZONTAL VIEW -H >^o h- SURFACE h,.''i VERTICAL VIEW ha.^t rry- Figure 57. Schematic of internal -wave packet showing characteristic lengths and features; (upper) view from surface; (lower) view in a vertical section (Apel et al . , 1976). 104 REFERENCES Apel, J. R., H. M. Byrne, J. R. Proni, and R. L. Charnell (1975a): Observations of oceanic internal and surface waves from the Earth resources technology satellite. J. Geophys. Res ., 80:865-881. Apel, J., R. H. M. Byrne, J. R. Proni, and R. L. Sellers (1976): A study of oceanic internal waves using satellite imagery and ship data. Remote Sensing of Environ . , 5:1 25- 1 35 . Apel, J. R., J. R. Proni, H. M. Byrne, and R. L. Sellers (1975b): Near simul- taneous observations of intermittent internal waves on the continental shelf from ship and spacecraft. Geophys. Res. Lett ., 2:128-131. Barnes, P. G. (1973): The generation of internal tides by flat-bump topography. Deep-Sea Res ., 20:179-205. Benjamin, T. Brooke (1967): Internal waves of permanent form in fluids of great depth. J. Fluid Mech . , 29:559-592. Bereyin, Y. A., and V. T. Karpman (1967): Nonlinear evolution of disturbance in plasmas and other dispersive media. Sov. Phys. JETP , English Trans- lation, 24:1049. Gargett, A. E., and B. A. Hughes (1972): On the interaction of surface and internal waves. J. Fluid Mech . , 52:179-191. Garrett, C, and W. Munk (1972): Space-time scales of internal waves. Geophys. Fluid Dyn ., 2:225-264. Halpern, D. (1971): Semidiurnal internal tides in Massachusetts Bay. J. Geophys. Res . , 76:6573-6584. Hunkins, K. , and M. Fliegel (1973): Internal undular surges in Seneca Lake: A natural occurrence of solitons. J. Geophys. Res . , 78:539-548. Korteweg, D. J., and G. de Vries (1895): On the change of form of a new type of long stationary waves. Philos. Mag . , 39:422. Lafond, E. C. (1962): Internal waves. In The Sea , Vol. 1, edited by M. N. Hill, Interscience, N. Y., 731-763. Lee, C, and R. C. Beardsley (1974): The generation of long nonlinear internal waves in a weakly stratified shear flow. J. Geophys. Res . , 79:453-462. Long, R. R. (1955): Some aspects of the flow of stratified fluids: No. 3, Continuous density gradients. Tell us , 7:341-357. Ostapoff, F. , J. R. Proni, and R. L. Sellers (1975): Preliminary analysis of ocean internal wave observations by acoustic soundings. GATE Rep. No. 14, WMO, Geneva, 392-397. Prinsenberg, S. J., and M. Rattray, Jr. (1975): Effects of continental slope and variable Brunt-Vaisala frequency on the coastal generation of internal tides. Deep-Sea Res . , 22:251-263. 105 Proni, J. R., and J. R. Apel (1975): On the use of high frequency acoustics for the study of internal waves and microstructure. J. Geophys. Res . , 80:1147-1151. Proni, J. R., F. Ostapoff, and R. L. Sellers: Preliminary acoustic observa- tions of high-frequency deep ocean internal wave groups during the GATE (in press). Shand, J. A. (1953): Internal waves in the Georgia Strait. EOS Trans. Am . Geophys. Union , 34:849-856. Woods, J. D. (1968): Wave-induced shear instability in the summer thermo- cline. J. Fluid Mech., 32:791-800. REMOTE SENSING OF ENVIRONMENT 5, 125-135 (1976) 125 A Study of Oceanic Internal Waves Using Satellite Imagery and Ship Data JOHN R. APEL*, H. MICHAEL BYRNE, JOHN R. PRONI, and RONALD SELLERS Ocean Remote Sensing Laboratory, Atlantic Oceanographic and Meteorological Laboratories, Environmental Research Laboratories, National Oceanic and Atmospheric Administration, Miami. Florida 33149 Surface manifestations of oceanic internal waves have been studied in Landsat-1 and -2 data since 1972. The internal waves appear as periodic, intermittent variations in the surface optical reflectivity and are visible from spacecraft, aircraft, and surface vehicles under certain circumstances. The Landsat data suggest that the source of the waves is semidiurnal and diurnal tidal action at the edge of the continental shelf. A study of the wave characteristics yields considerable insight into the physics of their excitation, propagation, and dissipation. Packets have been observed from the Gulf of Maine to Cape Hatteras and in images taken off the U.S. and African east and west coasts, the Gulf of Mexico and the Caribbean, The Gulf of California, the Sulu Sea, and the Baltic. The internal wave groups show an orderly variation in wavelength from front to rear of the packet, due to a combination of frequency dispersion and nonlinear amplitude effects. An oceanographic cruise was carried out in synchronism with two 18-day Landsat-1 cycles, and data were taken on temperature, density variations, acoustic echoes, and surface slicks accompanying the internal waves. The data were satisfactorily correlated with spacecraft and U-2 imagery taken simultaneously. I. Introduction Surface manifestations of oceanic in- ternal waves have been observed by many investigators, and their depen- dence upon wave-associated surface cur- rents has been recognized for at least a generation (Ewing, 1950). The surface signatures are most evident on a rela- tively calm sea and take the form of a quasi-periodic, long-length modulation of the capillary-ultragravity wave spec- trum; the scale of the modulation is usually of the order of hundreds of meters (Gargett and Hughes, 1972). Visi- ble manifestations on the ocean surface are theorized as being due to at least two mechanisms, either of which modulates the short surface wave structure rather than causing a change in optical reflec- tivity or absorptivity at depth. The first theory suggests that the high velocity of surface water arising from the large internal wave amplitude sweeps together surface oils and debris to form a slick in regions of surface water convergence, thus increasing the specular reflection and decreasing the diffuse scattering over the convergence region. The second theory predicts that the small waves are concentrated in the convergence regions due to wave-current stresses, thereby *Present address: Pacific Marine Environmental Laboratory, NOAA, Seattle. Washington 98105 ©American FIsevier Publishing Company, Inc., 1976 126 JOHN R. APEL ET AL. decreasing the specular reflection and increasing the scatter over the conver- gence region. II. Spacecraft Observations We have observed these types of peri- odic features in photographs and imag- ery taken from various spacecraft over v/idely separated regions of the world. To date, quasi-periodic features on the ocean surface have been seen on images made from Skylab, the Earth Resources Technology Satellite- 1, its successor, Landsat-2, the DMSP meteorological sat- ellite, and the Apollo-Soyuz mission, taken over both the continental shelf and deep ocean areas; they have been observed off the North American and African east and west coasts, in the Gulf of California, the Caribbean, the Baltic, and the Sulu Seas (Apel et al, 1975-a;-b). These features are almost certainly surface manifestations of inter- nal waves discussed above. A particularly dramatic example of internal waves may be seen in Fig. 1, FIG. 1. Landsat-l negative image, computer enhanced, showing internal wave packets south of long Island, New York, on 24 July 1973 (lower right-hand corner). SATELLITE IMAGERY OF OCEANIC INTERNAL WAVES 127 which is a computer enhanced, negative Landsat-1 image of the ocean south of Long Island, New York (I.D. 1366-15079). The Appendix discusses the enhancement techniques used. The size of the picture is approximately 140 by 140 km^. The waves are most appar- ent in the lower right-hand corner as two sets of curved, quasi-periodic variations in brightness having a packet-like behav- ior, a general north-south orientation, and with phase fronts convex towards the west. The wavelengths are of order 1000 m. In addition to the waves, a light-dark "marbling" pervades the image, probably brought about by light winds advecting oils on the surface and causing more or less random changes in optical reflectivity. Also visible are two acid waste dumping events in the ocean (center-left) near the inner portion of the New York Bight. (Charnell et al.. 1974). Figure 2 is an interpretive line drawing of the internal wave packets taken from Fig. 1 but with the area covered extended farther to the south in order to illustrate the wide extent of tlic internal wave signatures. The figure also shows bottom bathymetry so as to illustrate the general orientation of the lines of constant wave phase with isobaths, as well as the concentration of wave activ- ity near the Hudson Valley on the continental shelf. Repeated Landsat overpasses show these to be highly per- sistent characteristics of the waves (Saw- yer and Apel, 1976; Apel c/ a/. , 1975-a). Also shown are calculated isophase fronts (dashed lines), arrived at by as- suming that a plane internal wave is initially generated nearby and parallel to the edge of the continental shelf, which then propagates on the shelf under the intluence of varying water depth and density. The lines are spaced at arbitrary intervals. For convenience the calcula- tion assumes a water column stratified in two layers of thickness h^ and /12 and nearly equal densities p, — P2 — p. The dispersion equation, co = cj (k), relating frequency 00 and horizontal wave num- ber k, for the two layer system is obtained from Eq. (l)(Lamb, 1932): cj"* (coth kli2 coth kill + \)- co^ gk (coth khi + coth /c/zj ) + (Aplp)g^ k'^^0. (1) For a density contrast, Ap/p= 1 ^, a mixed layer depth of 50 m, and a wave- length 2Tr/k= 1000 m, the phase speed 00/k is approximately 0.25 m/s. depend- ing on the water depths shown in Fig. 2. The refraction of the calculated fronts on the shelf is due to the intluence of the deeper water over the Hudson Val- ley. The overall behavior of the model is in reasonable agreement with the space- craft observations. A study of Landsat-1 and 2 imagery has been completed for all satellite over- passes occurring during the summers of 1972 to 1974, between Cape Hatteras. N.C., and Nova Scotia, Canada; the results are being assembled in an atlas that will be published in the near future (Sawyer and Apcl. 1976). The study shows persistent internal wave activity all along the conthiental shelf and some limited patterns seaward of the slope. A more detailed analysis shows they have several characteristics which are perva- sive enough to be worthy of note: (1) The waves occur in groups or packets of width L=5 to 5 km, usually 128 JOHN R. APEL ET AL. FIG. 2. Geographically corrected line drawing of internal wave fields observed in Fig. 1, superimposed on bottom topography. The dashed lines on the graph show isophase contours as calculated from a two-layer model. landward of the continental slope and separated by distances, D, which are of the order of either 15 or 30 km; taken together with both the observed and calculated phase velocities, these facts suggest a semidiurnal or a diurnal origin ; (2) the crests are nearly always ori- ented parallel to the local bottom topography or can be loosely associa- ted with some topographic feature seaward of their observed position, or both; (3) the wavelengths fall between 200 and 10,000 m, depending on the geo- graphical area; within a given packet, SATELLITE IMAGERY OF OCEANIC INTERNAL WAVES 129 there is often a monotone decrease in wavelength, varying from X^ at the front to X < A„ at the back of the group; (4) the lengths of the crests, W, fall between a very few and perhaps 100 km, with a decrease in crest length occurring from front to back of the group; (5) the widths of the slicks, l^, are often small compared to the lead wavelength X^ : (6) the crests are curved in a horizon- tal plane with their convex sides pointed in the direction of propaga- tion; the radii of curvature. Re, range from essentially infinity to a few kilometers; (7) as the packet progresses up on the shelf, there is some evidence of a continued increase in the wavelengths throughout; an accounting for this may be had by a combination of linear dispersion and nonlinear effects akin to solitary wave behavior. Other data off southwest Africa (Apel et a!., 1975-a) show six distinct packets apparently radiating from a small source, perhaps a submarine canyon, approxi- mately 150 km offshore, with spacings indicative of once-a-day excitation. If such is the case, these data provide evidence that internal waves may have lifetimes of several days, even on the continental shelf. ni. Shipboard Observations A cruise with the acronym NYBER- SEX: New York-to-Bermuda Remote Sensing Experiment— was undertaken during June and July of 1974 to verify the interpretations offered above; it was scheduled so as to coincide with three consecutive Landsat-1 overpasses of the New York Bight, plus a NASA U-2 flight at the conclusion of the cruise. The ship was instrumented with a salinity-temper- ature-depth sensor, expendable bathy- thermographs, thermistors towed in a vee-fin, and a 20 kHz, downward-looking acoustic echo-sounder to delineate inter- nal wave motion at depth (Apel et ai, 1975-b; Proni and Apel, 1975). Because of weather, it was not possible to obtain one set of simultaneous observations involving all of the available instrumenta- tion. Nevertheless, an ample quantity of cross-correlated data was gathered to enable us to assert unequivocally that the spacecraft imagery does indeed con- tain surface manifestations of internal waves. The U-2 also obtained photogra- phy and multispectral scanner imagery of internal wave packets in and near the Hudson Canyon; the latter sensor is the prototype for the Nimbus-G Coastal Zone Color Scanner (Hovis. 1975). Figure 3 is a yellow-filtered, color infrared photograph made with the Vin- ton 70 mm camera on the U-2 at 20.000 m altitude, showing surface and internal waves and oil slicks in the sun glint patterns at the head of the Canyon on the shelf edge. For this packet, X„ s 500 m. Figure 4 illustrates shallow water in- ternal wave data from the echo-sounder (upper) and the towed themiistor (low- er), illustrating a packet approximately 150 km due east of Sandy Hook, New Jersey in some 80 m of water. The water was sharply stratified into two layers 130 JOHN R. APELET AL. SATELLITE IMAGERY OF OCEANIC INTERNAL WAVES 131 having upper and lower temperatures of about 17 ° C, and 8 ° C, separated by a 10 m thick gradient region centered at approximately 30 m depth. Clearly de- fined surface slicks accompanied the underlying internal wave field. The ship was proceeding at about 2.5 m/s from the rear to the front of the wave group while the data of Fig. 4 were taken. Two layers of acoustic-reflecting mate- rial were concentrated in regions where the macroscopic curvature of the vertical temperature profile, 9^ r/8z^ , was the largest; these layers are clearly seen in Fig. 4. This is in accord with models of acoustic scattering from fluctuations in the acoustic index of refraction (Proni and Apel, 1975). Thus the measurement yields some evidence that, in addition to scattering from biological material-the mechanism that is usually invoked to explain such reflections—, these signals may be partially returned from micro- structure and turbulence in the water column. Both the acoustic measurement and the coincident towed thermistor record yield a peak-to-trough amplitude, 277, of about 15 m; the latter record was used to evaluate 2r}—{Tu-Ti) {bT/dzT^ , with Tu and Ti being the maximum and minimum temperatures occurring during an oscillation. At 1625 UT, a 135° turn was exe- cuted because the forward edge of the packet, as defined by its slick pattern, had been passed. The wave coherence on the new course is somewhat lower, as may be seen on Fig. 4. A power spec- trum for the portion of the echo-sounder record to the left of 1625 UT is shown 50 15' 14* 13» 12* 11 101 9 8 7* 6«« 1600 i BOTTOM "■ S;ECTld)N FCiUR JULY 1 , 1974 ^ ULvyV-- 1605 1610 1615 1620 1625 1630 1635 1640 1645 UNIVERSAL TIME , t (min ) FIG. 4. (Upper) internal wave packet in 80 m water off New Jersey as observed by 20 kHz acoustic echo-sounder; (lower) temperature record from a towed thermistor at 30 m depth. A L^5 ° turn was executed shortly after 1625 UT. 132 C/) o < a. 24 - 1 1 1 1 1 18 - NYBERSEX 16 - ECHO-SOUNDER " 14 - 1600- 1629 Z 1 JULY 1974 12 - - 10- - 8 ■ K 6 1 \ 4- \ 2 n _ 11/ K_. . JOHN R. APEL ET AL. r 12 3 4 5 WAVENUMBER (CYCLES/KM) (a) ^ — 1 1 1 1 1 r— — 1 r- - ^^ A - ^"" / \ - /\ 1 \ /\ f "V. / ^ ^ \l NYBERSEX - V-FIN. ECHO-SOUNDER CROSS CORRELATION FUNCTION • 1600 - 1629 2 1 JULY '74 - 1 1 1 1.1 ._! 1 1 1.0 2.0 3.0 4.0 50 6.0 WAVENUMBER (CYCLES/KM) (b) 10 - -1 1 1 1— 1 1- 9 - / 8 - - 7 - NYBERSEX tn V-FIN, ECHO-SOUNDER t 6 - z 1 CROSS-AMPITUDE A SPECTRUM ^ 5- ll 1600- 1629 Z ? 1 1 JULY 1974 CD a - < ■ 3 - ^ 2 - h M. 171 1 - \\ \\\ . ^ V V. .^.. 12 3 4 5 6 -I \J *U.O -u.o -'_'.•» - U. «. V W.C. \J.f V,U V.V k.V LAG DISTANCE (KILOMETERSI WAVENUMBER (CYCLES/KM) (C) (d) FIG. 5. (a) Power spectrum from the portion of the acoustic echo-sounder record to the left of 1625 UT on Fig. 4; (b) spectrum for the thermistor temperature record from the same interval; (c) acoustic temperature cross-correlation function; (d) acoustic-temperature cross-amplitude spectrum. The dominant spectral component at 2 cycles/km corresponds to the dominant wavelength observed in Fig. 4. SATELLITE IMAGERY OF OCEANIC INTERNAL WAVES 133 in Fig. 5; the acoustic time series has been demeaned, detrended, and cosine- tapered in order to obtain the spectrum shown. Similarly, vee-fin temperature spectra, and acoustic-temperature cross- correlation functions and cross-spectra have also been evaluated and are repro- duced in Fig. 5. These show that the oscillations observed using the acoustic remote sensor are highly correlated with the temperature observations, a more conventional method of internal wave measurements. There are several significant features to be noted in Fig. 4. First, the oscilla- tions are indeed packet-like, having a sharply defined leading edge, as was observed from the Landsat and U-2 images. Second, the leading oscillation, occurring at about 1618 UT, is downgo- iiig; this is in accord with the solitary wave interpretation mentioned above (Lee and Beardsley, 1974). Third, both the amplitudes and the wavelengths de- crease toward the rear of the packet (i.e., earlier times); the decreasing amplitude is consistent with the shorter crest lengths observed in the spacecraft im- ages, since a small amplitude is accompa- nied by reduced currents, which in turn implies a reduced surface signature. Fourth, the mean depth of the upper mixed layer is lowered during the pas- sage of the wave packet. Fifth, the packets observed from the ship had spacings consistent with a 12 hr genera- tion interval. Because of clouds or bad weather, no one packet has been simultaneously identified from ship and spacecraft; how- ever, all of the data, including observa- tions of the disappearence of surface effects during winds above 6-8 m/s, are consistent with the picture offered above. IV. Summary of Packet Characteristics A schematic diagram that summarizes the salient features of the packets is shown on Fig. 6. This figure defines several quantities used in an empirical mathematical model for the amplitude of the packet, Eq. (2); some of the quantities have been defined above in the list of seven wave characteristics deduced from the satellite irhagery. In terms of a moving coordinate system with its origin at the front of the packet and its ;c-axis increasing toward the rear, an expression for the vertical displace- ment, 77, is given by r?(r) = VoRcix.y)W(z){x/i,)e-''''^^ (2) Here i?o is a normalizing amplitude; R^. ix,y) describes the curvature, R^-, of the crests in the horizontal plane; fV(z) is the solution to the linear, vertical-velocity eigenvalue problem for the density pro- file existing; the function x exp(-.v) together with the constant z, and the phase, -IT, give a downgoing initial pulse whose mean depth and envelope decay toward the tail of the packet: and the linearly increasing wave number, kg (\+x/li), accounts for the decreasing wavelength. The /'s give scales for the various features described. This expression, while only a model of the wave behavior, nevertheless contains the important features observed to date. Work is in progress to derive a theory 134 JOHN R. APEL ET AL. W ♦ Rr HXoK SURFACE HORIZONTAL VIEW h,,/', ■- VERTICAL VIEW BOTTOM FIG. 6. Schematic of internal wave packet showing characteristic lengths and features; (upper) view from surface; (lower) view in a vertical section. from a basic viewpoint that contains a more rigorous rendition than this admit- tedly quahtative behavior. V. Conclusions structure or biological material im- bedded in the internal wave motion is sufficiently intense. These two remote sensing techniques together constitute a set of new tools for investigations of high-frequency internal waves. High-resolution satellite imagery has been shown to be useful in studying internal waves on the continental shelf under conditions of clear skies and light winds. Amplitude information can be obtained from acoustic echo-sounding without the necessity for temperature sensors whenever scattering from micro- VI. Appendix Figure 1 is a contrast-enhanced nega- tive image of Landsat-1 image I.D. 1366-15079, Channel MSS-6. This wavelength interval, 700 to 800 nm, shows near-surface effects most readily SATELLITE IMAGERY OF OCEANIC INTERNAL WAVES 135 of the four MSS Channels. The contrast enhancement was done digitally by ap- plying a linear stretch to the original digital numbers 9-13 (on a scale of 127) so that they occupy the full dynamic range of the film. The optimum stretch appears to be one for which the intensity histogram for the oceanic part of the scene (which is nearly gaussian) has its stretched mean at approximately 64 and has about 3% of the energy falling beyond the limits and 127. Figure 1 is geographically distorted in the east-west direction by a factor of 3600/2400 because of the size of Landsat pixels; nor is it corrected for earth rotation. The interpretive line drawing, Fig. 2, is geo- graphically correct, however. Maul (1974) has discussed more fully some enhancement techniques for Land- sat oceanic images of this type. A portion of this work was supported by the Defense Advanced Research Pro- jects Agency, which assumes no responsi- bility for the accuracy of this report. The authors thank the National Aero- nautics and Space Administration for special data taking from both Landsat- 1 and the NASA U-2. References internal and surface waves from Earth Re- sources Technology • Satellite, / Geophys. Res. 80, 1147. Apel, J. R., J. R. Proni, H. M. Byrne, and R. L. Sellers (b), (1975); Near simultaneous obser- vations of intermittent internal waves on the continental shelf from ship and spacecraft, Geophys. Res. Lett. 2, 128. Charnell, R. L., J. R. Apel, W. Manning, III, and R. H. Qualset (1974), Utility of ERTS-1 for coastal ocean observation: The New York Bight Example, Marine Technol. Soc. J. 8 p. 42. Ewing, G., Slicks ( 1950), Surface films, internal waves,/. Mar. Res. 9, 161. Gargett, A. E., and B. A. Hughes (1972), On the interaction of surface and internal waves, /. Fluid Mech. 52, 179-191. Hovis, W. (1975), Ocean color imagery-coastal zone color scanner, Proc. NASA Earth Re- sources Survey Symp. VI-c Houston, Texas, p. 1989. Lamb, H. (1932), Hydrodynamics, Cambridge Univ. Press, London, p. 370. Lee, C, and R. C. Beardsley (1974), The generation of long nonlinear internal waves in a weakly stratified shear flow, J. Geophys. Res. 79,453. Maul, G. A., R. L. Charnell, and R. H. Qualset (1974), Computer enhancement of ERTS-1 images for ocean radiances, Remote Sensing of Environ. 3, (3), 237-25 2. Proni, J. R., and J. R. Apel (1975), On the use of high-frequency acoustics for the study of internal waves and microstructure, J. Geo- phys. Res. 80, 1 147. Sawyer, C, and J. R. Apel (to be published). An atlas of internal waves from Landsat, NOAA-ERL Technical Memo. Apel, J. R., H. M. Byrne, J. R. Proni, and R. L. Received 22 December 1975; revised 11 March Charnell (a), (1975), Observations of oceanic 7 976 Vol. 2, No. 4 CEOPHY.SICAL RLSLARCll LHTTLiRS April 19' NEAR-SIMULTANEOUS OBSERVATIONS OF INTERMITTENT INTERNAL WAVES ON THE CONTINENTAL SHELF FROM SHIP AND SPACECRAFT John R. Apel, John R. Proni, H. Michael Byrne, and Ronald L. Sellers Ocean Remote Sensing Laboratory Atlantic Oceanographic and Meteorological Laboratories Environmental Research Laboratories National Oceanic and Atmospheric Administration Miami, Florida 33149 Abstract . Internal waves on the continental shelf off New York have been observed from ship and the ERTS-1 spacecraft, and positive correla- tions made between surface and subsurface measure- ments of temperature, acoustic volume reflectivi- ty, and surface slicks. The spacecraft imagery senses the quasi-periodic variations in surface optical reflectivity induced by the internal waves. The waves appear to be tidally generated at the shelf edge and occur intermittently in packets, which propagate shoreward and disappear in water near 50-m depth. Introduction Surface manifestations of internal waves have frequently been identified during observations from ships, aircraft, and offshore towers, with "surface slicks" or "regions of enhanced capil- lary waves" being terms used to describe the state of the sea surface overlying the internal wave field. In a recent publication, we report on what appear to be surface signs of internal waves detected in visible and near- infrared imagery obtained from the multi-spectral scanner on the NASA Earth Resources Technology Satellite, ERTS-1 ( Apel et al. , 1975). These signs usually take the form of periodic varia- tions in optical reflectivity of the ocean sur- face; they have wavelengths of order 500 to 5000 m and appear repetitively in several groups, or packets, which are separated by intervals ranging from about 15 to 40 km. The surface manifesta- tions usually occur during conditions of light winds and relatively clear skies. While the absolute identification of the features in the spacecraft imagery as being due to internal waves had not been attempted until now, it has nevertheless been possible to synthe- size a simple, consistent internal wave model that accounts for most of their major character- istics. They appear intermittently on the conti- nental shelf at intervals which suggest their generation by semidiurnal and diurnal tides at the edge of the shelf ( Halpern , 1971) ; they seem to be refractlvely controlled in phase speed and propagation direction by the mixed layer and water depths and the Brunt-Vaisala profile; and they largely disappear where the mixed layer comes to occupy a substantial fraction of the water column. Nonlinear, dispersive behavior is Indicated by certain of their characteristics. Copyright 197S by the American Geophysical Union. including a sharp onset and the frequent appear- ance of the longest wavelengths at the front of the packet ( Lee and Beardsley , 1974) . New York-to-Bermuda Remote Sensing Experiment (NYBERSEX) A cruise was scheduled during June and July 1974 aboard the R/V Westward , a 30-m auxiliary staysail schooner selected in part for her quiet acoustic characteristics. The experiment was primarily Intended to observe internal waves in the New York Bight coincident with three conse- cutive ERTS-1 overpasses of that region (12 June, 30 June, 18 July), during the early summer, a season when previous ERTS observations of the periodic surface features had been most frequent ( Apel et al ., 1975). The ship was instrumented with a salinity- temperature-depth device (STD) , expendable bathy- thermographs (XBT) , a horizontally towed under- water hull containing temperature and depth sensors, an echo sounder (20-kHz, 2-ms pulse, 12° X 15° beam) that viewed vertically downward to delineate internal wave motion at depth ( Proni and Apel , 1975) , and color and infrared films for photographing surface features. Preliminary Results of the Experiment Measurements were made during approximately 20 days of ship operation and 16 separate space- craft overpasses. Because of weather, it was not possible to obtain one set of simultaneous observations involving all of the available instrumentation. Nevertheless, an ample quantity of cross-correlated data was gathered to enable us to assert unequivocally that the spacecraft imagery does indeed contain surface manifesta- tions of internal waves. Figs. 1(a) & (b) illustrate the geographical area under investigation on 17 July 1974; a portion of the photographically enhanced ERTS-1 negative image (numbered 1724-14475-6) taken in the near-infrared between 700 and 800 nm; and a line drawing interpretation of the periodic surface features. The ship's track is located in the lower left-hand corner of Fig. 1(a), with the circle denoting its position at the time of the satellite overpass — 1447.5 GMT — and the triangle the location of a well-defined internal wave field encountered later on, at approximately 2115 GMT. Four packets are visible in the image. 128 Apel et al . : Internal Waves 129 1 - 1 il 1 ^' 1 ' Figure 1(a) (Left) . Line drawing showing internal wave packets south of Long Island (solid lines) overlain by isobaths in meters (dotted lines) . Ship track at bottom shows position at time of ERTS-1 overpass (dot) and at encounter with internal wave packet (triangle) . Vertical dashed line is western edge of space- craft image. Figure 1(b) (Right) . Section of negative image from ERTS showing - internal wave field from which Figure 1(a) was derived. Dimensions 154 x 108 km. Black areas are clouds. generally oriented along isobaths and separated by 24 to 33 km. They have wavelengths at the front of the packets ranging from approximately 400 to 1000 m and appear to be propagating shore- ward, a view that is substantiated in part by the absence of packets seaward of the shelf break and In part by the observed doppler shifts discussed below. No surface slicks were visible from the ship at the time of the overpass, nor are there any apparent at that point in the image. Nor would any waves be expected there, since the ship was beyond the edge of the shelf and hence outside of the boundary of the hypothesized generation region. However, at the position where the internal wave field was encountered (at the triangle), slicks were visible from the ship. Color photographs taken from the ship show that the slicks have the color of reflected skylight, in the low winds of 3 to 4 m/s; see Fig. 2. Figure 3 illustrates temperature and acoustic data gathered during a maneuver involving course reversals that was executed through the region indicated by the triangle on Fig. 1. The upper trace shows the temperature at a depth of 29 m Figure 2 . Photo of internal wave slicks, taken from deck height (photo courtesy of R. L. Hughes) 130 Apel et al . : Internal Waves O S^ 20 X 15 a: UJ 10 20 wmm mmmmmmmmmmmmmmmmii^mmm E ■ — 40 N X 60 h- Q. UJ O 80 v» X X X X X WNW ESE DISTANCE ALONG TRACK X (km) Figure 3 (upper) . Temperature trace at 29 m depth taken near triangle on Figure 1; (lower) . acoustic reflections from two scattering layers at 20 and 33 m depth, and from bottom at 80 m. Ship course was reversed midway through. as obtained from the towed thermistor. The left half of the record was obtained during a 2 . 5-km segment of a west-northwest course, during which the internal wave packet of about 2 km total length was encountered; the right half shows the trace obtained during the reverse course. The left half shows down-shifted and the right half up-shifted oscillations. The maximum tempera- ture excursions were of order 6 C. The lower record was obtained from the 20-kHz echo sounder and clearly shows two acoustic reflecting layers oscillating vertically with peak-to-trough ampli- tudes of approximately 15 m. The temperature and acoustic records are in very good correlation, with a 180 phase shift and with the troughs of the wave corresponding to downward-moving warm water from the mixed layer and the crests to upward-flowing cold water. An STD cast made approximately one-half hour before the onset of this wave packet showed a temperature profile which, if oscillated vertically by 15 m, would yield approximately the temperature excursions shown in the upper portion of Fig. 3. This same graph can be used to derive the dominant wavelength. A, period, T, and the pro- jection of the internal wave phase velocity, ?', along the ship velocity, u, by assuming negative and positive doppler shifts due to the ship motion on the reciprocal courses. From the down- and up-shifted wavelengths on that figure, one obtains approximately A ^ 500 T iii 20 min c ^ 0.35 m/s. where a ship speed u = 1.5 m/s has been used. The phase and group speeds for a 500-m wave- length are about 0.5 m/s for A = 500 m, as cal- culated from the observed density profile, for which 2-n/N - 140 s. . ™3X Returning to Fig. 1, the wave packet that is closest to the one shown on Fig. 3 has a wave- length of approximately 600 m and a packet length of about 3 km, values quite near to those ob- served from the ship. In addition, the isophase lines of the packet appear to be nearly parallel to the isobaths, and it is quite reasonable to assume that the isophase fronts of the wave group marked by "0.6 km" can be extended toward the southwest parallel to the bottom topography, out of the ERTS picture and to the region of ship observation. During the seven-hour time delay between satellite overpass and shipboard measurements, the waves would have propagated only about ten km toward the northwest and could thus be expected at about the location shown. Assuming constant phase and group speeds of 0.35 m/s, the wave packet would progress about 33 km shoreward in 25 hr. This is close to the spatial intervals observed on Fig. 1, which suggests once-daily generation. Additional spacecraft and shipboard data show weakly developed internal wave striations on 12 June 1974, a calm, moderately clear day. However, on 13 June 1974 no evidence of slicks was visible either on the ERTS image or from the ship; this was a day of fresh, 7-8 m/s northeasterly winds, and slicks apparently do not form under such conditions. Another experiment (a) established that the echo-sounder was indeed quantitatively observing internal waves; and (b) showed the location of the reflecting layers relative to small-scale temperature gradients. Figure 4(a) (left) shows two XBT traces obtained on 1 July 1974; one was taken at the crest and the other at the trough of an internal wave packet that was propagating past the ship while it was adrift . The packet was accompanied by a well-defined series of slicks. The temperature traces show the thermo- cline oscillating up and down with approximately a 20-min period by an amount ri(z) that depends on depth z; by differencing the two traces, one may obtain the displacement, which is shown on the right. The lowest order mode is clearly Apel et al . : Internal Waves 131 i 20h 40 60 T — I — r 1 1 1 1 r T T F/^/^1SV\l REFLECTING LAYERS I I ' \. I l__ 10 15 TEMPERATURE T|z), CO 20 2 4 6 8 10 DISPLACEMENT -rym.im) Figure 4 (Left) . Temperature profile at internal wave trough (solid line) and crest (dotted line). Cross-hatching shows regions of intense acoustic reflec- tions, as in Figure 3. (Right) . Displacement of isotherms shows dominant first- order mode. the dominant one, but higher order modes are also present. Coincident STD data demonstrate that the water column is statically stable, with the temperature inversion shown being offset by Increased salinity in that region. The coincident towed thermistor and echo- sounder records also yielded data on this inter- nal wave field. These traces (not reproduced here) appear quite similar to those on Fig. 3. The acoustic record shows two reflecting laminae that undergo vertical excursions whose ampli- tudes and phases are in excellent agreement with those shown on the towed temperature and XBT records. The locations of these reflecting layers are indicated by the hatched regions of Fig. 4(a) and are clearly associated with depths where the macroscopic curvature of the temperature record, 3 T/8z , is the largest. This is in accord with models of acoustic scat- tering from fluctuations in the index of re- fraction. Thus the experiment yields some evidence that, in addition to scattering from biological material — the mechanism that is usually invoked to explain such reflections — these signals may be partially returned from microstructure and turbulence in the water column. Conclusions The combined measurements of horizontal and vertical temperature variations, density struc- ture, acoustic reflections, and spacecraft imagery have shown that the periodic features seen in ERTS Images taken over the New York Bight are Indeed surface slicks associated with oceanic Internal waves; that such waves appear to be generated at the edge of this part of the continental shelf approximately once a day; that they propagate as intermittent high-frequency packets whose wavelengths, phase speeds, and periods are consistent with the vertical density profiles and depths for the area; and that they are detectable by high-frequency acoustic echo sounders. Thus the combination of spacecraft imagery, which yields synoptic scale data, and echo sounding, which provides amplitude infor- mation on a more restricted scale, appears to offer useful tools for investigations of higher- frequency internal waves on the continental shelves. Acknowledgements . The authors are grateful to NASA for gathering open-ocean ERTS-1 data and to the officers, crew, and apprentices of the Westward for their help. This research was partially supported by funding from the Advanced Research Projects Agency, which assumes no responsibility for the content of this report. References Apel, J. R., H. M. Byrne, J. R. Proni. and R. L. Charnell, Observations of oceanic internal and surface waves from the Earth Resources Technology Satellite, J. Geophys. Res. , 80 , 865, 1975. Halpern, D. , Semidiurnal Internal tides in Massa- chusetts Bay, J. Geophys. Res. , 76 , 6573. 1971. Lee, C, and R. C. Beardsley. The generation of long nonlinear internal waves in a weakly stratified shear flow, J. Geophys. Res. , 79 . 453, 1974. Proni, J. R. , and J. R. Apel. On the use of high- frequency acoustics for the study of internal waves and microstructure, J. Geophys. Res. (to be published), 1975. (Received November 11, 1974; accepted March 14, 1975.) VOL. 80, NO. 6 JOURNAL OF GEOPHYSICAL RtStARCH FEBRUARY 20, 1975 Observations of Oceanic Internal and Surface Waves From the Earth Resources Technology Satellite John R. Apel, H. Michael Byrne, John R. Proni, and Robert L. Charnell Atlantic Oceanographic and Meteorological Laboratories, Environmental Research Laboratories, NOAA Miami, Florida 33149 Periodic features observed on the ocean surface from the Earth Resources Technology Satellite I have been interpreted as surface slicks due to internal wave packets. They appear to be generated at the edge of the continental shelf by semidiurnal and diurnal tidal actions and propagate shoreward. Nonlinear effects apparently distort the wave packets as they progress across the shelf. This observational technique con- stitutes a new too! for delineating two dimensions of the internal wave field under certain limited con- ditions. Surface Observations Visible manifestations of internal waves on the ocean sur- face may be due to at least two mechanisms, either of which modulates the short surface wave structure rather than causes any change in optical reflectivity or light absorption at depth. The first mechanism suggests that the high velocity of surface water arising from the large internal wave amplitude sweeps together surface oils and materials to form a slick in regions of surface water convergence, and thus the surface reflectivity is increased [Ewing, 1950; Shand, 1953]. The second mechanism predicts that capillary wave energy is focused in the con- vergence zones due to surface stress, and thus the small-scale roughness is enhanced and the surface reflectivity is decreased [Gargeli and Hughes, 1972; Thompson and West, 1972]. In either case, quasi-periodic surface features will be produced under light wind conditions, and such eff"ects are very often seen at sea. Figure la is a photo taken from an aircraft in the Caribbean during Bomex in 1969 that shows a series of what almost cer- tainly must be internal waves propagating from the left overlaid at nearly right angles by much shorter wind waves (R. B. Grossman, private communication, 1974). A ship wake crosses from left to right. The length of the internal waves is to a very rough approximation near 200 m. Figure \b illustrates internal waves being generated by a tide rip and shows at least three distinct wave trains propagating at difl'erent angles [La- fond. 1962]. Periodic surface slicks may generally be seen from the air under conditions favorable for viewing oil on the sea, i.e., at low sun angles or in the sun glitter, when light or moderate winds prevail. Generation Mechanisms Internal waves can be generated by a wide variety of mechanisms. Wind stress that excites a spectrum of surface waves is one source [Phillips, 1966]. In this case, wave-wave scattering between two surface oscillations of wave vectors and frequencies x,, «2 and o),, 0)2, respectively, generates an internal wave whose parameters are set by the conservation rules K3 = K, - Kj and W3 = CO, - 0)2. The amplitude of the internal wave is partially determined by the surface wave amplitudes; the in- terference pattern decays exponentially with depth with a scale given by I/|k3| and thus makes itself felt at levels where inter- nal waves may propagate. Copyright © 1975 by the American Geophysical Union. A second mechanism for generating internal waves is the scattering of barotropic tidal energy into internal baroclinic tides by bottom roughness or by the discontinuities presented by the edges of continental shelves or island arcs [Rattray, 1960; Halpern, 1971]. The relative importance of different topographies in transforming tidal motions into internal wave motions is unknown. This phenomenon is difficult to observe in nature; extant evidence suggests that up to Vi of the 3 x 10" W energy dissipation required to account for the lengthening of the day may be ascribed to such scattering [Hendershott, 1973]. Thus island arcs, continental shelf edges, and shallow submarine sills become important regions in which to make observations of internal waves. A third mechanism that can cause internal waves is shear flow instability [Rattray, 1960; Phillips, 1966]. In regions of the ocean having large baroclinic currents accompanied by horizontal stratification, small fluctuations in depth of a given stratum can grow to large amplitudes in an oscillatory way by mechanisms generally akin to the Kelvin-Helmholtz instabili- ty. An indication of the onset of this growth can be obtained by evaluating the gradient Richardson number Ri, where Ri = N'^/{dUo/dzf. Here A' is the Brunt-Vaisala frequency, and dujdz is the vertical shear in the horizontal baroclinic current speed Uo(2), which may be due to a mean motion or a tidal current. The onset of the growth generally occurs when Ri ap- proaches '/4 (or less). The oscillation amplitude is usually a maximum at some intermediate depth, i.e., is an unstable in- ternal wave. This mechanism may be responsible for the gen- eration of short-period internal waves from baroclinic internal tides of semidiurnal periods or in regions of western boundary currents. Erts 1 Observations of Internal Waves Periodic or quasi-periodic features in or on the ocean have been detected on many Erts 1, Skylab, and U.S. Air Force Dapp meteorological satellite images taken over both con- tinental shelf and deep ocean areas; to date we have observed these features off" the North American east and west coasts, the Gulf of California, the African east and west coasts, northern South America, the central Pacific, and the Celebes Sea. These features are almost certainly surface manifestations of internal waves that are made visible to the spacecraft by one of the mechanisms described above that serve to 'tag' the waves dur- ing winds which are well below whitecapping speeds. The observations will first be presented, and then arguments will be made to support the internal wave interpretation. 865 866 Apel et al.: Satellite Observations Fig. la. Aircraft observations of internal waves during Bomex (1969) (latitude 9.5°N, longitude 59.5°W). Also shown are surface waves, a ship wake, and wind rows. . .n\" •:..•;-<• i^f Fig. \b. Additional aircraft photo of internal waves [Shand, 1953). Surface effects on the continental shelves. Figure 2 is a stan- dard Erls image taken with the Erts I mullispeclral scanner southeast of New York; it is printed as a positive image and shows Long Island at the top and New Jersey at the left. The scene dimensions are 184 X 184 km'; north lies appro.ximately 102° counterclockwise from the horizontal edges. The scene was taken in orange-red light between 6(X)- and 7(X)-nm wavelength, al approximately 1007 LT on August 16. 1972. The large white objects in the center of the picture are high-altitude clouds; southeast of the mouth o'i the Hudson River two U-shaped acid-dumping events are visible, little else of oceanographic interest is apparent. Apel et al.: Satellite Observations 867 ^.•ir.^rv ' ' ■'r- 1 Fig. 2. Standard NASA Erts image. Area shown is the New York bight taken on August 16, 1972. Figure 3 is a negative print of the same image, computer- enhanced through a process known as contrast stretching, with the land portions sacrificed in favor of ocean information. There is a weahh of data observable, including the Hudson River plume, acid and sewage sludge dumps, color discon- tinuities, and in the southeast corner, periodic surface slicks. (This figure is also geographically stretched by a factor of 1.4: I in the east-west direction due to computer/photo recorder peculiarities.) Figures 4, 5, and 6 are similar computer-enhanced images of the New York bight area taken on May 31, July 6, and July 24, 1973; each shows periodic striations in the southeast corner of somewhat similar characteristics. No such features were observed on scenes taken over the same area between August 1972 and May 1973. Figure 7(a, b, c, and ophase conii>urs as calculated Ironi a simple internal wave model Apel et al.: Satellite Observations 873 Fig. 8. Eastern Long Island area, July 23, 1973 (photographically enhanced), 24 hours prior to and 110 km to the east of the area shown in Figure 6. Note packets separated by about 30 km in lower left center. phase speeds near 0.25 m/s, with a longer distance between crests and a greater extent of the packet occurring on the in- shore side and shorter wavelengths and smaller packet extent on the offshore side; the reasons for some of these characteristics are not yet clear. 6. There is some evidence of a continued lengthening of the distance between crests, especially in the front of the packets, as the group progresses up on the shelf; an accounting for this may be had by a combination of finite amplitude and dispersive effects. Surface effects observed in deep water. The data shown thus far have been taken on or near continental shelves, in water depths less than about 500 m. Very limited numbers of Erts 1 images have been taken over deep water in the open ocean, of which two scenes are shown in Figures 14 and 15. Figure 14 illustrates a photographically enhanced image of a portion of the deep Indian Ocean taken on December 4, 1972; equatorial east Africa is visible on the left. The continental shelf in this part of Africa extends only a few tens of kilometers off the coast. Quasi-periodic structures are apparent in the cloud-free center portion of the scene having a scale of 2-4 km and relatively little coherence. In the lower right there is a definite change in the orientation of the structure. These features lie approximately in the southward-flowing Somali current; the image was taken at a time of year when the current in the region usually undergoes sharp spatial and temporal 874 Apel et al.: Satellite Observations 41'>N 73°w 72°W im ZO* 2 4 JULY 7 3 2 3 JULY 7 3 39°N Fig. 9. Geographically corrected line drawing showing wave packets from Figures 6 and 8. variations due to the shifting monsoon. It is thought that the structure in question may be a surface signature of deep water internal waves; however, no sea truth is available for this area, and so such an interpretation must remain speculative. Another ocean scene is shown on Figures 15 and 16. These Erts ! data were taken in the North Pacific at about 40°N, I60°W in September 1973. Figure 15 has dimensions of 92 X 184 km^ and is a computer-enhanced negative print showing cumulus clouds in the south and a peculiar surface mottling in the north having a scale of a few kilometers, in this latter region, a microscopic examination shows an expanse of sur- face waves approximately 200 m in length propagating toward the southeast away from a low pressure cell several hundred kilometers away, having wind speeds of 10 to 15 m/s. A more detailed look at a 1 3 X 1 8 km sector of this region is shown on Figure 16a, which clearly illustrates two trains of surface gravity waves of nearly equal wavelength intersecting at approximately 15°, as well as the mottling mentioned above; the latter is seen to be very roughly periodic with an average 'wavelength' of about 2.5 km and approximately a north-south orientation. Figure \6b is the logarithm of the two-dimensional Fourier transform of Figure 16a and shows the spatial frequency components of the mottling near the origin and those of the surface waves as two concentrations of energy separated by about 15° at nearly equal wave numbers. Under the suspicion that the origin of the mottling might be found either in ( 1 ) surface wave interference effects or (2) in the kind of internal wave generation process arising from wave- wave scattering that was discussed earlier, we have applied the wave vector conservation rule k, — K2 = "a to the scene. The magnitude of K3 agrees with that of Kmott, the wave vector of the mottling, to within 10%; its angle, which is nearly at 90° to the mean value of *ti and Kj. agrees to within 20°. Such agree- ment does not prove that either of the above processes suggested are at work, of course, but such interpretations are at least consistent with the data. Obviously, further controlled experiments are required. The surface stress hypothesis might explain the visibility of internal waves on the surface of a moderately rough ocean having a wind blowing over it [Gargett and Hughes. 1972]. The horizontal surface currents associated with internal waves are large, being of the same order as the phase velocity of the waves themselves; they are oriented approximately either parallel or antiparallel to the direction of propagation and have opposite directions in parts of the wave separated in phase by (2n - l)7r. A wind field superimposed on the currents will excite a capillary/ultragravity wave spectrum that is steeper where the wind opposes the internal wave current than where it is parallel to it. This periodic variation in the slope spectrum would lead to periodic variations in the optical reflectivity that might serve to delineate the internal wave in the deep sea to high-altitude sensors, much as surface slicks serve to do in continental shelf waters. The ditTerence in the two cases is that the slick cannot maintain its integrity at wind 900 Variation in Wavelength: Eastern Long Island g 600 - J. 400 - o c • 5 200- o ——^ South Sector T -• North Sector 07/23/73 1 D B 80 m h, a 50-60m • ■ — t i o- L f - 1 2 3 Position In Wave Pocket , km — Fig. 10. Variiition in internal wavelength as a function of distance from the front ol the upper and lower packets seen on Figure 8. Apel et al.: Satellite Observations 875 Fig. 11. Enhanced Erts photo made on November 28, 1972, oil' soulhwesl Al'rica jusl north of Cape Town, showing six packets of surface slicks apparently radiating from a small source at the center left. This photo suggests internal waves may have lifetimes on the shell of several days. 876 Apel et al.. Satellite Observations 18°00' E 17°00' E 30° 00' S 31*00' S 32*00' S 30°00' S 31*00' S 32*00' S 17*00' E 16*00' E Fig. 12. Line drawing of packets of Figure I! together with bottom topographs. Wav« Packet Characteristics ^ Southwest Africa 11/26/72 Fig I.V Leading wavelength and interpaeket separation as a func- tion of distance, from figures 11 and 12. velocities above whitecapping speeds, whereas the above mechanism may. Surface effects in the Florida current. Figure 1 7 is presented as a final example of possible internal waves observed from Erls 1. This is a contrast-stretched negative image of the Florida current made on August 18. 1972. \Mth Miami located near the upper portion of the photograph. At the bottom center near the mean axis of the current is a group of periodic striations oriented roughly northwest-southeast and extending approximately 30 km along the direction of mean tlow. The distance betvseen stripes is about HOG m. It is thought that these are surface signs of internal waves excited b> and propagating along the Florida current. Periodic surface slicks in this general region are coinmon at sea. and acoustic echo-sounding from ships has indicated that internal vsaves often underlie Apel et al.: Satellite Observations 877 such slicks [Proni and Apel, 1975]. The mechanism for their generation is quite possibly a shear flow instability. On the westward side of the current in the region of cyclonic horizon- tal velocity shear the vertical velocity and density gradients are sometimes such that the Richardson number for the flow ap- proaches '/4. Duing's [1973] data give values for Ri that indicate incipient instabilities may occur. Once again, the theoretical conditions that indicate the existence of a mechanism for generating internal waves have to be verified by simultaneous field and satellite measurements. Summary and Conclusions Convincing proof of the assertions that (1) the periodicities observed on Erts 1 images are surface slicks and (2) the slicks tag internal waves under conditions of light winds has been ob- tained by the authors in a recent cruise between New York and Bermuda. The data are only partially analyzed and will be published in the near future [Apel et al.. 1974]. The limitations of the satellite technique appear to be as follows: light wind and clear sky conditions must prevail, high- frequency internal waves on the shelf are most visible, and coverage is limited by satellite dynamics and sensor characteristics. No amplitude information is currently de- rivable from the spacecraft images. However, as the ac- companying paper shows [Proni and Apel, 1975], this missing dimension to the internal wave field may be partially ob- tained via acoustic echo sounding, so that a more nearly complete picture of the three-dimensional field may be obtained. ■ rr,. •• • v>'"'' J'sk^i, Fig. 14. Pholographically enhanced Erts image of a portion of the deep Indian Ocean otf equatorial east Africa taken on December 4, 1972. 878 Apel et al.: Satellite Observations Fig. 15. Enhanced Erts image (negative) taken over Ihe North Pacific al 40°N. 160''W in September 197.V Image dimensions are 92 X 184 km'. Note the peculiar surface mottling and surface wave lield. K ■w)! Z00/9P^5 I- I I I I I I I I I i I t I I o i JO 6 V3 — — i« o = S ^ u o ^ u t— >^ fo ^^ ^ '-> . '— ^ o ° '■'= 5 o ill " 5 c jf 2 ?i -: § u u 'J 1 = i > '•J- x. ^ u C l> o > — a _ ? -o ^ Kl o E c ao -^ o X 5b X M u c c E cd r 00 ^ o . c ^ X •o E £ Q c o 1^ 3 ^ ? F w u ^ C > 00 c §^ cib n e1 E o o- 880 Apel et al.; Satellite Observations Kig. 17. Contrast stretched negative Erts image of the Horida current made on August 18, 1972. Note the periodic striations present at the bottom center ol the image. Apel et al.: Satellite Observations Acknowledgment. This project was partially supported by funds from the Advanced Research Projects Agency, which assumes no responsibility for the correctness of the results. References Apel, J. R., J. R. Proni, H. M. Byme, and R. L. Sellers, Near- simultaneous observations of intermittent internal waves from ship and spacecraft, submitted to Geophys. Res. Lett., 1974. Duing, W., Observations and first results from ProjcQt Synop 71, Sci. Rep. UM-RSMAS 73010. Univ. of Miami, Miami, "Fla., 1973. Ewing, G., Slicks, surface films and internal waves, J. Mar. Res., 9, 161, 1950. Gargett, A. E., and B. A. Hughes, On the interaction of surface and in- ternal waves, J. Fluid Mech., 52. 179-191, 1972. Halpern, D., Semidiurnal internal tides in Massachusetts Bay, J. Geophys. Res., 76. 6573-6584, 1971. Hendershott, M., Ocean tides, Eos Trans. AGU, 54, Ib-ib, 1973. Lafond, E. C, Internal waves, in The Sea, vol. 1, edited by M. N. Hill, pp. 731-763, Interscience, New York, 1962. Lamb, H., Hydrodynamics, p. 370, Cambridge University Press, Lon- don, 1932. Lee, C, and R. C. Beardsley, The generation of long nonlinear inter- nal waves in a weakly stratified shear flow, J. Geophys. Res., 79(3), 453-462, 1974. Phillips, O. M., The Dynamics of the Upper Ocean, pp. 161-197, Cam- bridge University Press, London, 1966. Proni, J. R., and J. R. Apel, On the use of high-frequency acoustics for the study of internal waves and microstructure, J. Geophys. Res. , 80, in press, 1975. Rattray, M., On the coastal generation of internal tides, Tellus, 12, 54-62, 1960. Shand, J. A., Internal waves in the Georgia Strait, Eos. Trans. AGU, 34, 849-856, 1953. Thomson, A. J., and B. J. West, Interaction of non-saturated surface gravity waves with internal waves. Tech. Rep. RADC-TR-72-280, Phys. Dyn. Inc., Berkeley, Calif., (ARPA order 1649), October 1972. (Received July 24, 1974; accepted November 6, 1974. <>GPO: 1978-796-417/116-10 I PENN STATE UNIVERSITY LIBRARIES ADDDD7mMSM DT 1