SECTION 8
The following case studies are intended to expose the reader to a range of remote sensing techniques and applications. The case studies have been arranged in an order which reflects the primary objective of the experiment:
Fish Detection
Direct Technique:
- School dimension (Case Study No. 1)
Indirect Techniques:
- Fishing boat count (Case Study No. 2)
- Echo sounding (Case Study No. 3)
- Bioluminescence (Case Study No. 4)
- Chlorophyll (Case Studies No. 5 and 6)
- Water turbidity and ocean colour (Case Study No. 7)
- Sea surface temperature (Case Studies No. 8 and 9)
- Phytoplankton and sea surface temperature (Case Study No. 10)
- Ocean colour and sea surface temperature (Case Study No. 11)
- Ocean colour, chlorophyll and phytoplankton (Case Studies No. 12 and 13)
Invertebrate Detection
Indirect Techniques:
- Buoy count (Case Study No. 14)
- Bottom habitat (Case Study No. 15)
Marine Plant Detection
Direct Technique:
- Spectral signature (Case Studies No. 16 and 17)
Bottom Habitat Identification
Direct Technique:
- Spectral signature (Case Study No. 18)
Physical Oceanographic Parameter Measurement
- Bathymetry (Case Study No. 19)
- Sea surface temperature (Case Study No. 20)
- Currents (Case Study No. 21)
- Surface water velocity (Case Study No. 22)
| Reference: | Hara, I., 1985, Moving direction of Japanese sardine school on the basis of aerial surveys. Bull.Japan.Soc.Sci.Fish., 51 (12): 1939–45 |
| Remote Sensing Technique: | Airborne Remote Sensing: Platform - Cessna U-206G; Sensor - Visual Observation and Aerial Camera. |
| Objective: | To detect, identify and monitor the shape and movement of schools of sardines, Sardinops melanosticta, off the southeast coast of Hokkaido, Japan. |
| Experimental Rationale: | Aerial surveys are more appropriate for the observation of sardine schools than sonar measurements or underwater observations because they are more synoptic, cover a wider geographic area and facilitate mapping. |
| Method: | The shape, colour and movement of schools of sardines were determined by viewing a 1 km swath from a small aircraft at an altitude of 500 m (refer to Figures 8.1 and 8.2). The schools were subsequently measured from aerial photographs (refer to Figure 8.3). Movement of the fish schools was determined by reference to three yellow dye dispensers which were attached to floating devices (refer to Figure 8.4). |
| Results and Conclusions: | The dark blue colour of the front of the schools and the “watery blue” colour of the rear allow the determination of school movement with a single observation. |
| The relative dimensions of sardine schools in the surveyed area were 33 – 50:5:1 indicating the ratio of breadth:length:vertical extent, respectively. The greatest dimension, breadth, is usually the leading edge of a school. | |
| This experiment indicated that aerial survey methods are effective for the study of fish school behaviour. It would be desirable, however, to establish a coordinated and synchronized system of air and ship- borne observations. |
Figure 8.1 Flight line for visual observation on September 22, 1984. The number of sardine schools sighted is indicated.
Figure 8.2 Flight line for visual obsevation on September 23, 1984. The numbers of sardine schools sighted is indicated.
Figure 8.3 Changes in oval-shape school sketched form vertical or oblique photography. The time of observation is indicated. The three dots represent the airdropped floats and the two thin lines depict a yellow lpume from the attached seamaker.
Figure 8.4 The changes of the elongate and the intermediate-shape schools expressed in the same manner as Figure 8.3. The arrow indicates the direction of school movement.
| Reference: | Bazigos, G.P. et al., 1979, Aerial frame survey along the southwest coast of India. Rome, FAO, UNDP/FAO Pelagic fishery investigation project on the southwest coast of India. FIRM- IND/75/038:104 p. |
| Remote Sensing Technique: | Airborne Remote Sensing: Platforms - Twin Beech and E18-S-Type Aircraft; Sensor - Human Eye (Aerial Spotting). |
| Objective: | 1. To assess the area of distribution and level of localization of the inshore sector of the fishing industry; |
| 2. To provide current estimates of the number of operational fishing craft seen by type and by major fishery, i.e. non-mechanized fishery and mechanized fishery; | |
| 3. To provide information on the level of fishing activity of the operational fishing boats at the beginning of the fishing season. | |
| Experimental Rationale: | This survey was based on the ability of well trained fishery biologists to differentiate types of fishing craft and gear from an aircraft flying at a height of 500 feet and moving at a speed of 120 nautical miles per hour. |
| Method: | Two groups of experienced researchers (3 in each group) were briefed before the flight. The responsibility of group I was to obtain measurements of the fishing boats seen on the coastline/landing places. The duty of group II was to obtain measurements of the fishing boats seen on the water. Separate records were kept during the survey flight for the non-mechanized and mechanized boats. |
| The Aerial Frame Survey (AFS) included only operational fishing boats in its list. The survey area covered the entire eastern coast and southern tip of India. This long coastline was divided into time/space units in order to increase the accuracy of eye estimation. | |
| An error that occurs in a survey of this nature is the omission of the mechanized boats operating beyond the areas covered by the AFS. In order to unbias the calculated estimates of the total number of mechanized boats seen on the water, a small-scale coverage check survey (CCS) of the AFS was carried out. | |
| The coverage check survey was comprised of a sample sonar-exploratory fishing cruise conducted for the project soon after the completion of the AFS. The findings of the survey (refer to Figure 8.5) were divided into two categories: | |
| 1) mechanized boats and non-mechanized boats seen on the water operating within the area covered by the AFS; | |
| 2) mechanized boats and non-mechanized boats seen on the water operating beyond the area covered by the AFS. | |
| The analysis of data was carried out by the following methods: | |
| 1) determination of the co-efficient of fishing activity, i.e. the percentage of fishing boats in the water; | |
| 2) determination of the concentration of fishing boats, i.e. the number of standard craft per unit area or unit length. A medium size canoe was taken as the standard; | |
| 3) determination of the geographic relation between non-mechanized and mechanized boats; | |
| 4) determination of the spatial distribution pattern of all types of fishing boats. | |
| Results and Conclusion: | When compared with government statistics, the results showed a substantial discrepancy in the number of craft for certain areas. This was mainly due to government data not being up-to-date. In a situation when the existing statistics are not reliable, there is no method to test the validity of an AFS. |
| Because fisheries in the project area were more dynamic in character than anticipated, two aerial frame surveys were recommended during the fishing season. | |
| In the statistical analysis, an attempt was made to assess the operational distribution and level of localization of the surveyed fisheries. A mathematical model was developed to express the spatial distribution. |
Figure 8.5 Pictorial chart of the distribution of the marine fishing boats covered by the AFS in the project area by one degree of latitude.
| Reference: | Blindheim, J., G.H.P. de Bruin and G. Saetersdal, 1979, A survey of the coastal fish resources of Sri Lanka. Report No. 2, April - June 1979. Reports on surveys with R/V DR. FRIDTJOF NANSEN. Bergen, Institute of Marine Research, 63 p. |
| Remote Sensing Technique: | Echo-Sounding: Platform - Research Vessel (DR. FRIDTJOF NANSEN); Sensors - Echo-Sounder, Sonar. |
| Objective: | To describe and assess the demersal, semi-demersal and pelagic resources available in the coastal region of Sri Lanka. |
| Experimental Rationale: | Echo-sounding can be used as a sampling technique for fish populations. Assuming that the distribution of fish in a particular area is uniform and that the speed of the ship is constant, the area of a fish trace made on the echo-sounder recording can be considered to be proportional to the size of the fish school. The conversion factor for this proportionality was deduced by test experiments. The character of the sea bottom can also be determined by an echo-sounder recording. |
| Method: | Fishing and echo-sounding surveys of the waters around Sri Lanka were carried out with the fishery research vessel DR. FRIDTJOF NANSEN for three seasons, August- September 1978, April-June 1979 and January-December 1980. Fishing experiments were made on a total of 133 fishing stations with different types of fishing gear including bottom trawl, pelagic trawl, bottom long line and oyster dredge (refer to Figure 8.6). |
| The remote sensing equipment used in this study was two echo-sounders (38 KHz and 120 KHz) and a searchlight sonar (18 KHz). | |
| The echo-integrator attached to this acoustic equipment recorded the analog data at a rate of 1 mm per nautical mile. | |
| The bottom condition of the shelf floor was classified into 4 groups: “even flat”, “uneven”, “very rough” and “steep slope”. By analyzing the echo recording of the bottom, the general characteristics of the continental shelf of Sri Lanka were found to be “even flat” with an “uneven” and “very rough” outer edge. Mapping of the sea bottom with echo-sounders, coupled with experimental fishing, identified the favourable fishing grounds. | |
| For the sake of navigation and safety, charting of the sea was confined to regions deeper than 10 metres. Species could be differentiated by the pattern of the recordings. Small sized pelagic fish, e.g., clupeoids and scads, were found in well defined schools. Semidemersal fish, e.g., snappers, breams, jack mackerel, etc., were found in looser aggregations. Refer to Figures 8.7, 8.8 and 8.9 for examples. The discrimination of the fish into different species with echo-recording was confirmed by the fish catch. | |
| Although no detailed quantitative analysis of the echo- recording was made, it was possible to differentiate echo-traces to three arbitrary types regarding fish biomass: very scattered (1–10 mm per nautical mile), scattered (11–20 mm per nautical mile) and dense (greater than 20 mm per nautical mile). The echo- integrator readings in mm per nautical mile are relative measures proportional to fish density, i.e. one unit of 1 mm per nautical mile represents a certain square nautical mile. A conversion factor is needed to determine absolute fish biomass from the relative echo- integrator values. | |
| This survey also included the determination, by conventional methods, of temperature, salinity and dissolved oxygen in nine sections across the shelf. | |
| Results and Conclusions: | Observations on the bottom condition revealed that good trawling grounds were limited to the shallow inshore parts of the shelf and to the shallow northern area. |
| Hydrographic observation permitted a description of the water masses on the continental shelf. The dissolved oxygen content in waters of the Northwest, Northeast and East coasts was found to be insufficient to support commercial fish species. The survey described the distribution of fish of commercial importance along the coast from Negombo to Pedro Bank; the total standing crop was assessed at approximately 500,000 tonnes. On the basis of some simple assumptions concerning mean density of biomass in the northern shallow waters not covered by this survey, the total biomass of Sri Lanka's coastal shelf and immediately adjacent waters was estimated to be 750,000 tonnes. The sustained annual potential yield from these resources was calculated to be about 250,000 tonnes, of which 80,000 tonnes represent large demersal or semi-demersal species. This estimate may well prove to be conservative, but it does indicate that the present level of catch of 100,000 tonnes at least can be doubled. |
Figure 8.6 Dimensions of the sound beam form the echo sounder at 20 m echo depth in relation to the distance between trawl doors and wind ends of the trawl net.
Figure 8.7 Example of “Type A” echo recordings of demersal ans semi- demersal fish.
Figure 8.8 Example of “Type B” echo recordings of dispersed pelagic fish.
Figure 8.9 Example of “Type C” echo recordings of schooling small pelagic fish among recordings of dispersed larger pelagic fish (Type B).
| Reference: | Roithmayr, C.M., 1970, Airborne low-light sensor detects luminescing fish schools at night. Commer. Fish. Rev., 32(12):42–51 |
| Remote Sensing Technique: | Airborne Remote Sensing: Platform - Aircraft; Sensors - Aerial Camera, Spectrometers. |
| Objective: | To assess fish stocks at night using low-light sensors. |
| Experimental Rationale: | Bioluminescence is light produced by living animals and plants including plankton. The responsible light- emitting substance, luciferin, when disturbed absorbs energy and emits light. When fish schools disturb the water, many of these organisms emit light by which the presence of fish can be detected. |
| Method: | During 1968, tests were conducted in waters off Florida aboard a commercial seiner. A starlight scope was used to detect bioluminescence associated with Spanish mackerel schools. With the use of a scope coupled to a closed circuit television camera, the image of luminescing schools was recorded on video tape. |
| During dark nights luminescing fish schools were recorded on video tape with a SANOS (Stabilization Airborne Night Observation System) scope and closed- circuit television. Luminescing schools at the surface could be detected by a low-light detector at an altitude of 1,600 m (5,000 feet). | |
| Results and Conclusion: | Low-light detectors proved to be efficient in detecting fish schools on moonless nights (refer to Figure 8.10). This can be of great assistance to fishermen in reducing their search time. It also will give scientists an opportunity to study the behaviour of fish schools at night. |
| The airborne sensor could greatly assist in resource assessment by providing “real time” observations on the number and size of fish schools. |
Figure 8.10 A large luminescing school of thread herring, 160m (500 ft.) in diameter, amplified by airborne low-light sensor
| Reference: | Cram, D.L., 1979, A role for the NIMBUS-9 coastal zone colour scanner in the management of a pelagic fishery. Fish.Bull. /Visserij-Bull.,Cape Town, (11):1–9 |
| Remote Sensing Technique: | Satellite Remote Sensing: Platforms - NIMBUS-G, LANDSAT-1; Sensors - CZCS, MSS. |
| Objective: | To detect and measure chlorophyll concentrations with the aid of CZCS imagery to determine pilchard distribution off the coast of South West Africa. |
| Experimental Rationale: | Pilchard schools tend to avoid areas of dense phytoplankton and zooplankton abundance. Because of the difference in spectral signatures between brown- green phytoplankton (Fragilaria karstenii), which was the most dominant species in South West African waters, and sea water, the former could be discriminated in CZCS imagery. This provided an important method of delineating pilchard distribution. |
| Method: | CZCS imagery was analysed in order to delineate areas of phytoplankton occurrences. Variations in ocean turbidity, deduced from LANDSAT-1 data, were compared with pre-existing fisheries, and oceanographic and remotely sensed data in an attempt to identify biological features off the south-west coast of Africa. |
| An attempt was also made to discriminate areas of zooplankton abundance, mainly euphausiids, from LANDSAT images. | |
| Results and Conclusions: | The CZCS of NIMBUS-G imagery was found to be efficient in detecting and measuring the chlorophyll-a content of water. LANDSAT MSS imagery was used to delineate areas of zooplankton (refer to Figure 8.11). With the knowledge of the areas of occurrences of phytoplankton and zooplankton, it was possible to infer the occurrence of pilchard, mainly because pilchard tend to occupy zones where the phytoplankton and zooplankton are present in intermediate densities. |
| This information could be used to minimize the search component of fishing effort and to refine assessments of pilchard stock by means of catch-per-unit-effort calculations. |
Figure 8.11 Plankton distribution and observed positions of plichard shoal (school) groups-cumulative over 10 days
| Reference: | Caraux, D. and R.W. Austin, 1983, Delineation of seasonal changes of chlorophyll frontal boundaries in Mediterranean coastal waters with NIMBUS- 7 coastal zone colour scanner data. Remote Sensing Environ., 13(3):239–49 |
| Remote Sensing Technique: | Satellite Remote Sensing; Platform - NIMBUS-7; Sensor - CZCS. |
| Objective: | To determine the seasonal changes of major chlorophyll boundaries in the northwestern Mediterranean (Golfe du Lion) using CZCS imagery. |
| Experimental Rationale: | The Coastal Zone Colour Scanner (CZCS) of NIMBUS-7 was primarily designed to detect chlorophyll content of the coastal waters. |
| Delineation of chlorophyll frontal boundaries using CZCS imagery has been carried out for several years. | |
| The concentration of chlorophyll was considered to be a function of the difference of reflectance values in two channels. Varied combinations of bands and values for constants in the mathematical model were used for different types of water, i.e. reflectance values changed according to the concentration and composition of suspended particles, etc. | |
| Method: | Cloud free images of the Golfe du Lion were obtained for the year 1979. The images were corrected for atmospheric attenuation. Using specific algorithms, maps were produced depicting the distribution of chlorophyll for the area under investigation (refer to Figure 8.12). |
| Conclusion: | The chlorophyll-like pigment boundaries determined in this study showed a resemblance to those previously identified by other workers using thermal infrared radiometry. |
| The results demonstrate that phytoplankton distribution is a good indicator of seasonal variations of oceanic fronts, with which many fish species are associated. Features such as coastal upwellings, cyclonic eddies or plumes could be monitored. |
Figure 8.12 Monitoring of chlorophyll frontal boundaries in the Golfe de Lion throughout 1979. Boundaries were delineated after attributing different colours to chlorophyll concentration ranges obtained on CZCS image. The different tilt and scan angles of the CZCS radiometer account for variations in the area covered.
| Reference: | Kemmerer, A.J., 1980, Environmental preferences and behaviour patterns of Gulf menhaden (Brevoortia patronus) inferred from fishing and remotely sensed data. ICLARM Conf. Proc., (5):345–70 |
| Remote Sensing Technique: | Satellite and Airborne Remote Sensing; Platforms - LANDSAT-1 and 2, Aircraft; Sensors - MSS, Aerial Camera, T.V. |
| Objective: | To evaluate and subsequently to demonstrate the feasibility of using remotely acquired data to enhance the management and utilization of coastal pelagic fishery resources, especially menhaden. |
| Experimental Rationale: | The distribution of menhaden was found to be correlated with water turbidity and colour. The distributional pattern of these fish can therefore be predicted in relation to these parameters. |
| Method: | Most of the data considered in this paper were collected from three areas in the Northern Gulf of Mexico in 1972, 1975 and 1976. These areas were selected because of the LANDSAT coverage, logistics and location of the fishing fleet. In 1972 aerial photography and a low-light level imaging television system were used extensively to collect data on menhaden distribution and abundance. |
| In addition to the commercial fish catches, oceanographic data were also collected, including: sea surface temperature, salinity, light penetration, colour, surface chlorophyll-a and water depth. | |
| Results: | The only oceanic parameters that had any correlation with the distribution of menhaden were light penetration and ocean colour. Although the distribution of this fish could not be inferred from LANDSAT images, the radiance values of each spectral band were found to be significantly correlated. On the basis of turbidity and colour the LANDSAT images were classified to give the highly probable areas of menhaden occurrences (refer to Figure 8.13). In 1976 a satellite-aided fishery harvest and assessment system was demonstrated and was subsequently established. It is now possible to produce in near real - time probability charts of menhaden occurrences using LANDSAT MSS imagery. |
| Conclusion: | The ability to infer fish distribution patterns synoptically from environmental data collected by spacecraft will have profound management implications. It will enable resource investigators to improve sampling design and subsequent analyses for more efficient and accurate stock assessments. Long term monitoring of these patterns will enable resource managers to detect and subsequently to predict natural and man-induced resource perturbations. |
Figure 8.13 Classification of LANDSAT MSS data from May 20, 1975, into high and low probability menhaden fishing areas for the eastern half of the Mississippi Sound.
| Reference: | Lasker, R. et al., 1981, The use of satellite infrared imagery for describing ocean processes in relation to spawning of the northern anchovy (Engraulis mordax). Remote Sensing Environ., 11:439–53 |
| Remote Sensing Technique: | Airborne Remote Sensing: Platform - NOAA-6; Sensor - AVHRR. |
| Objective: | To relate variations in mesoscale (approximately 200 km) sea surface temperature distributions with anchovy spawning and to identify and delineate important ocean processes that might influence the survival of fish eggs and larvae. |
| Experimental Rationale: | This project attempted to evaluate the influence that surface water temperature has on the movement and reproduction of the Northern Anchovy, by means of satellite imaging of sea surface temperature patterns. This relationship could lead to predictive methods for the identification of optimal fishing sites. |
| Methods: | The AVHRR data were converted to sea surface temperatures. Temperature calibration was done by relating pixel grey scale values to same-day sea surface temperature observations from ships at several locations. |
| Oceanographic and biological parameters were acquired by the NOAA research vessel David Starr Jordan using a thermosalinograph, plankton net and midwater trawl. | |
| Results and Conclusions: | The distribution of first-day anchovy eggs clearly showed that nearly all spawning was confined to the Southern California Bight. The seaward extent of spawning was apparently confined by the southward intrusion of recently upwelled water indicated by the 14°C isotherm (refer to Figure 8.14). |
| In March-April 1980 anchovy did not enter water colder than 12.5°C nor warmer than 17°C. In earlier years appreciable spawning took place at temperatures above 16.5°C and in water colder than 14°C. The modal temperature for anchovy spawning was 14°C – 15°C from 1969 to 1979 but was 15° to 17°C in 1980. For this reason the authors were led to the conclusion that temperature was not the sole environmental variable responsible for anchovy distribution in 1980. Temperature, however, can be used as an indicator of water mass distribution. | |
| Wind, chlorophyll and sea state sensors used in conjunction with the AVHRR should provide more reliable information on fish egg and larval distribution. |
Figure 8.14 Distribution of anchovy aggs superimposed in the thermal image of the Southern California Bight
| Reference: | Cornillon, P. et al., 1986, Sea surface temperature charts for the southern New England fishing community. Marine Technology Society Journal, 20(2):57–65 |
| Remote Sensing | |
| Technique: | Satellite Remote Sensing: Platform - NOAA-7; Sensor - AVHRR. |
| Objective: | To evaluate the potential assistance of satellite- derived SST charts to fishing activities on the continental shelf. |
| Experimental Rationale: | A number of fish species have a preference for seawater which has a specific and limited temperature range. Fishermen, therefore, can take advantage of sea surface temperature (SST) charts to assist in the locating of optimal fishing areas. |
| Method: | During 1983 and 1984, the Sea Grant Marine Advisory Service and the Graduate School of Oceanography at the University of Rhode Island (URI) prepared 37 SST maps from AVHRR data and mailed them to Southern New England fishermen. Throughout the program, user response was sought and, in order to meet user requirements, three different products were delivered: a temperature map covering the full study region (refer to Figure 8.15); a temperature map featuring an enlargement of a subregion of particular interest (refer to Figure 8.16); and a temperature gradient map showing areas of high horizontal SST gradients (refer to Figure 8.17). |
| The AVHRR data were received at the University of Rhode Island 24 to 36 hours after their acquisition. They were then processed and the results were immediately mailed to fishermen who received them the next day. | |
| Results and Conclusions: | The interaction between URI and the users during the experiment allowed the development of products that fishermen recognized as being useful. Many fishermen, however, wanted greater spatial resolution and several considered the charts to be outdated by the time they were received. Although fishermen considered that they saved money due to decreased search time, few were willing to spend more than $50 per year to receive the charts. Two problems were identified: infrared sensors such as the AVHRR do not see through clouds, resulting in periods where no SST data can be acquired. Microwave sensors, however, can produce SST data through clouds but these sensors are not currently available on civilian satellites. Secondly, the delay between the satellite pass and the delivery of the SST chart to the fishermen was excessive. This delay could be greatly reduced by the use of electronic data transfer methods. |
Figure 8.15 NOAA/NESDIS (National Environmental Sarillite Data and Information Service) oceanographic analysis chart for June 18, 1984
Figure 8.16 Subsection of the NOAA/NESDID chart shwon in Figure 8.15, midified by NMFS (National Marine Fisheries Service)
Figure 8.17 Enlargement of the region east of Long Island, mailed to fishermen
| Reference: | Laurs, R.M. et al., 1984, Albacore tuna catch distributions relative to environmental features observed from satellites. Deep- Sea Res., 31(9):1085–99 |
| Remote Sensing Technique: | Satellite Remote Sensing: Platforms - NOAA-7, NIMBUS-7; Sensors - AVHRR, CZCS. |
| Objective: | To determine the relationship between hydrographic patterns, as imaged by satellite sensors, and albacore distribution. |
| Experimental Rationale: | It is well known that the occurrence of albacore is closely related to oceanographic conditions in the North Pacific. The AVHRR allows the recognition of thermal fronts while the CZCS provides an accurate measurement of chlorophyll content, hence, phytoplankton distribution. Used together, they should provide valuable information on albacore distribution and should guide fishermen to optimal fishing regions. |
| Method: | The satellite data collection and processing network utilized for this experiment is indicated in Figure 8.18. The CZCS radiances were converted to phytoplankton pigment concentration using a pigment algorithm (band ratio), while the AVHRR radiances were converted to temperature and corrected for atmospheric effects. |
| Finally, the sea surface temperature images were registered to control points taken from the CZCS image so that the two images could be co-registered (superimposed). | |
| Albacore catch data were obtained from daily logs submitted by fishermen. The catch data for periods from two days before to two days after each pair of satellite passes were plotted on the temperature and phytoplankton pigment images. | |
| Results and Conclusion: | The results clearly showed the existence of a relationship between oceanic fronts and albacore distribution. Shoreward intrusions of oceanic water were particularly favourable sites for albacore aggregation. The results also showed that, in offshore water, commercial concentrations of albacore were associated with oceanic boundaries marked by colour fronts resulting from phytoplankton distribution and concentration, but without sea surface temperature gradients. Thus phytoplankton concentration seemed to be at least as important as temperature in explaining and predicting the occurrence of albacore (refer to Figures 8.19 and 8.20). |
Figure 8.18 Satellite data collection and processing network utilized by US National Marine Fisheries Service (Southwest Fisheries Centre).
Figure 8.19 Central California daily albacore catches, 27 September to 2 October, 1981, superimposed on the NOAA-7 AVHRR sea surface temperature, 30 September, 1981,14:02 PST.
Figure 8.20 Central California daily albacore catche, 27 September to 2 October, 1981,Superimposed on the NIMBUS-7 CZCS blue-green colour ratio and phytoplankton pigment concentration, 29 September, 1981, 11:30 PST.
| Reference: | Montgomery, D.R. et al., 1986, The applications of satellite-derived ocean color products to commercial fishing operations. Marine Technology Society Journal, 20(2):72–86 |
| Remote Sensing Technique: | Satellite Remote Sensing: Platform - NIMBUS-7; Sensor - CZCS. |
| Objective: | To assess the utility and benefit of specially prepared environmental charts designed for commercial fishing operations. |
| Experimental Rationale: | It appears that ocean colour and sea surface temperature data are complementary in defining the environmental limits of the distribution of albacore tuna, Thunnus alalunga. Satellite-derived data can provide a wide geographic and synoptic coverage in comparison to that obtained by ship or aircraft. |
| Method: | The CZCS data from the NIMBUS-7 satellite were received and processed by the Scripps Satellite Facility in California. The image pre-processing steps included atmospheric correction, blue/green ratio image creation and resampling to Mercator projection. Data analysis involved the identification of water masses, the creation of a false-colour composite and, finally, the production of boundary charts from the interpreted image. |
| Since the transmission of colour facsimile is a complex and expensive process, the data were converted to a black and white chart, and annotated with water type codes (refer to Figure 8.21). Water colour information, so obtained, was integrated with other fisheries-aid charts which were prepared by meteorologists/oceanographers working from a variety of sources (refer to Figure 8.22). Fishermen were able to obtain these data by mail or by radio transmission (facsimile and voice). | |
| Results and Conclusion: | The program demonstrated that conventional and satellite-derived data on the marine environment, when properly combined and correlated, can offer to the commercial fisherman tactical tools which can result in the selection of fishing strategies for more efficient and economical operations. This program found limitations in operation due to inconsistent CZCS coverage caused by satellite programming and cloud cover. Other problems were related to the high cost of preparation of the ocean colour charts (approximately $ 1000 each) and to the difficulty of transmitting the charts by high frequency radio to the near shore regions. |
Figure 8.21 Chart forwarded by telecopier to radoi facsimile broadcast stations for subsequent transmission to participating fishing vessels.
Figure 8.22 Representative chart generated as part of fisheries demonstration effort.
| Reference: | Feldman, G.C., 1986, Variability of the productive habitat in the Eastern Equatorial Pacific. EOS Transactions, American Geophysical Union, 67(9):106–8 |
| Remote Sensing Technique: | Satellite Remote Sensing: Platform - NIMBUS-7; Sensor - CZCS. |
| Objective: | To show that satellite ocean colour data can be used to define the spatial extent of the region of enhanced biological production (productive habitat) in the eastern equatorial Pacific (refer to Figure 8.23). To determine the degree of interannual variability in the areal extent of the productive habitat and in the estimated primary production of the region. |
| Experimental Rationale: | The changes in ocean colour detected by the CZCS provide a quantitative measure of phytoplankton pigment concentrations in the surface layer of the ocean. These concentrations are an index of phytoplankton biomass and may be empirically related to primary production. Examination of a series of large-scale images, covering the entire eastern equatorial Pacific, allows the determination of the temporal and spatial scales of oceanic processes and of the resulting variability in the distribution and abundance of phytoplankton. Phytoplankton represent the first link in the food chain and their patterns of distribution in time and space may indicate how oceanographic processes regulate primary production. |
| Method: | A sequence of CZCS scenes was processed to derive images of chlorophyll-like pigment concentration coregistered to a uniform spatial grid covering the eastern equatorial Pacific region. Subsequently, these images were composited to produce seasonal mean pigment maps for the 1978–79, 1979–80 and 1982–83 winter periods. Finally a comparison of the results obtained for each of these three periods was made with reference to previous descriptive and modelling studies of the eastern equatorial Pacific environment. |
| Results: | Significant coherence in the distribution and abundance of phytoplankton was found, in both time and space, within each of the three periods considered. The primary production estimates from the CZCS data show a close agreement with those from ship sampling obtained in the same periods. The time/space composite images retain the major features observed during each period and appear to be the best means for quantifying the high degree of interannual variability evident from the imagery. This interannual signal was found to be greater than that observed over the shorter time scales involved in constructing the seasonal composites. Surprisingly, the largest variability occurred between the 79–80 period and the other two periods, i.e. 78–79 and 82–83, which actually have similar characteristics in spite of the El Nino event of the 82–83 period (refer to Figure 8.24). In the 79–80 winter, the area classified as productive habitat (pigment concentrations greater than 1 mg/cu.m) was about one order of magnitude larger than in the other two winters, reaching almost 30% of the study area versus 3–10% in 78–79 and 82–83 respectively. Therefore, the major question raised by this study does not revolve around El Nino, but rather in trying to understand the reasons for the variability between 78–79 and 79–80, since in these periods the conditions throughout the region have been characterized as being close to normal (refer to Figure 8.25). |
| Conclusion: | This work demonstrates the potential of remotely sensed pigment measurements for the assessment of primary production and productive habitat extent on a regional or even global scale. For the eastern equatorial Pacific, there is evidence that there may be significant large-scale oceanic and atmospheric differences even when El Nino type phenomena are not active. Variations in the strength, location and timing of intensified undercurrent flows (e.g, the Equatorial and Peru Undercurrents) could alter the large-scale patterns of vertical mixing and nutrient input, i.e. upwelling along the Peru coast, thereby influencing phytoplankton production and the fish population sustained by this production. Perhaps the system is periodically purged. Low primary production during certain periods of time results in a significant reduction in the abundance of herbivores such as copepods and anchovies. The associated reduction in grazing pressure would then allow a large increase in planktonic abundance if accompanied by sufficient nutrient levels. A “boom and bust” type of cycle could then be established in the ecosystem of the region. The quantitative information derived from the satellite images allows the primary production to be estimated for the entire study area, as well as the production arising from specific regions. The repetitive character of the information makes it possible to follow the evolution of this production. Marine resource exploitation strategies may be identified byøm such analyses.ates the potential of remotely sensed pigment measurements for the assessment of primary production and productive habitat extent on a regional or even global scale. For the eastern equatorial Pacific, there is evidence that there may be significant large-scale oceanic and atmospheric differences even when El Nino type phenomena are not active. Variations in the strength, location and timing of intensified undercurrent flows (e.g, the Equatorial and Peru Undercurrents) could alter the |
Figure 8.23 Map of the eastern equatorial Pacific Ocean, showing the major features of the submarine bathymetry.
Figure 8.24 Cumulative frequency distributions of satellite derived phytolankton pigment concertration (in milligrams per cubic metre) versus the percentage of total cloud-free suface area covered by each concentration range for the eatern equatorial Pacific, as observed by the CZCS.
Figure 8.25 Frequency distributions of satellite-derived phytoplankton pigment concentrations (in miligrams per cubic metre ) versus the percentage of total cloud-free surface area covered by each concentration range for the region 0o-100S, 870-78oW, as observed by the CZCS.
| Reference: | Barale, V. et al., 1986, Space and time variability of the surface color field in the Northern Adriatic Sea. J.Geophys.Res., 91(C11):12957–74 |
| Remote Sensing Technique: | Satellite Remote Sensing: Platform - NIMBUS-7; Sensor - CZCS. |
| Objective: | To investigate the variability of the surface bio- optical field in terms of phytoplankton pigment concentration derived from CZCS data and its correlation with possible steering factors, over the entire northern Adriatic basin at interannual, seasonal and monthly scales. |
| Experimental Rationale: | A time series of corrected, coregistered CZCS images can be used to provide some simple statistics of the surface colour field, describing the heterogeneity of phytoplankton distribution. The patterns of primary productivity have spatial and temporal characteristics associated with the main oceanographic properties of a basin such as the Northern Adriatic Sea. By comparing these patterns with a suite of environmental factors, it is possible to investigate their relationship and the conditions and processes which regulate food chain dynamics. |
| Method: | The available CZCS data for 1979 and 1980 were used to construct a two year time series of Northern Adriatic Sea visible imagery. The original images were corrected for atmospheric contamination, processed into maps of chlorophyll-like pigment concentration and remapped to a standard geographical grid with a pixel resolution of about 1 km × 1 km. This allowed the derivation of yearly and monthly mean and standard deviation images and of two image sequences suitable for empirical orthogonal function analysis. These statistics were compared with analogous statistics of air temperature, wind velocity and coastal runoff. In particular, quantitative estimates of the spatial scale of high pigment concentration patterns, derived from the monthly averaged images, were correlated with the corresponding averages of outflow from the major river (the Po) entering the Northern Adriatic Sea. The results were interpreted in terms of the established oceanographic knowledge of the basin and of numerical modelling and laboratory experiments. |
| Results: | The phytoplankton pigment concentrations derived from CZCS data exhibited a high correlation with sea truth measurements performed during seven surveys in the summer of both years. A comparison of the mean pigment fields from 1979 to 1980 indicates an increase in the concentration values and the spatial extent of coastal features (refer to Figure 8.26). This variability may be linked to the different patterns of nutrient influx due to coastal runoff. The distribution of surface features is consistent with a general cyclonic circulation pattern, even though surface colour can not be considered a passive tracer of the flow in every instance. The pigment heterogeneity appears to be governed by fluctuations of freshwater discharge. The dominant wind fields do not appear to have important direct effects. The Po River discharge results in a plume which spreads over most of the northern basin with scales positively correlated with its outflow. The spatial scales of the western coastal layer in which the southward component of the cyclonic circulation is confined are instead negatively correlated with the Po River outflow and the plume scales (refer to Figure 8.27). Both results are consistent with theoretical and experimental results which indicate a dynamic balance between nonlinear advection and bottom friction, with the predominance of one of the two effects on an alternating basis. |
| Conclusion: | The analysis of time series of visible imagery can be a powerful tool for the assessment of ecological conditions and relationships in the marine environment. For the test site and for the two years considered in this study, the freshwater input from coastal runoff had a profound influence on the pigment field in addition to the thermohaline condition of the basin. The planktonic environment is significantly affected by both long and short term variations in runoff, possibly as a result of the associated variations of nutrient influx. |
Figure 8.26 Average conditions of the surface colour field in the Northern Adriatic Sea: 1979 yearly (a) mean and (b) standard deviation of phytoplankton pigment concentration; 1980 yearly (c) mean and (d) standard deviation of phytoplankton pigment concen tration. In all images, obtained from 35 individual CZCS scenes per year, high pigment concentrations (above 2 mg/cu.m) are shown in red; intermediate concentrations (between 1 and 2 mg/cu.m) in yellow and green; lower concentration (below 1 mg/cu.m ) in shades of blue.
Figure 8.27 Compariosn of monthly averaged Po river outflow (in cubic metres per second) with Po river plume scale and western coastal layer scale (in kilometres) for the period form August 1978 to December 1980. The scales of surface colour features are indicative of regions with very high phytoplankton pigment concentrations (above 1 mg/cu.m).
| Reference: | Pringle, J.D. and R.E. Duggan, 1983, A remote sensing technique for quantifying lobster fishing effort. Can.Tech.Rep.Fish.Aquat.Sci., 1217:16 p. |
| Remote Sensing Technique: | Airborne Remote Sensing: Platform - Aircraft (Aztec Pa-23); Sensor - Aerial Camera. |
| Objective: | To determine the lobster fishing effort in the Scotia- Fundy region of Atlantic Canada with the aid of colour aerial photography. |
| Experimental Rationale: | The lobster fishing effort was considered to be proportional to the number of traps in use for a given time period and area. The number of traps had a direct relationship to the number of buoys to which they were attached (1:1 in the present study). The brightly coloured buoys could be identified in colour aerial photographs. |
| Method: | Based on the results of a series of trials, it was decided to utilize colour positive film at an altitude of 915 metres (3000 ft), giving the photographs a scale of 1:6,000 and a coverage of 1,840.2 square metres. A 2 km stretch of coastline along St. Margaret's Bay, Nova Scotia, was photographed using a Fairchild K-17 aerial camera with a 15.24 cm lens and a Kodak HF-3 filter. A surface survey of buoys was also conducted for ground verification. Using a back-lit digitizing table, buoys identified on the photographs were digitized and mapped with a Geographic Information System (refer to Figure 8.28). |
| Results: | There was a difference of 28.5% between ground truth and aerial photograph counts; this was attributed to poor weather conditions. In areas where weather was favourable the error was reduced to 11.3%. |
| Conclusion: | Aerial photography using a Fairchild K-17 or Wild RC-8 camera with a 15.24 cm lens, Kodak no. 2448 colour positive film and a Kodak HF-3 filter gave sufficient resolution to detect buoys when the maximum altitude flown was 915 metres and the sun angle was less than 25°. |
| Recommendations: | 1. Flights should be made during calm weather conditions to minimize error. |
| 2. An evaluation of the test area to identify other fisheries using the same type of buoys should be made. | |
| 3. This technique is appropriate for spot-checking critical geographic areas because of its accuracy, rapidity and aerial coverage. The cost, however, is relatively high. |
Figure 8.28 Map of buoy locations - actual fishery.
| Reference: | Bour, W., L. Loubersac and P. Rual, 1986, Thematic mapping of reefs by processing of simulated SPOT satellite data: application to the Trochus niloticus biotope on Tetembia Reef (New Caledonia). Mar.Ecol.Prog.Ser., 34:243–9 |
| Remote Sensing Technique: | Airborne Remote Sensing: Platform - Aircraft (Simulated SPOT); Sensor - Deadalus radiometer (Simulated HRV). |
| Objective: | To determine the potential stock size of Trochus (Trochus niloticus) by means of an assessment of the areal extent of its coral habitat. |
| Experimental Rationale: | Exploitation of Trochus (Trochus niloticus) is important to Pacific Islanders. The shell yields valuable mother-of-pearl of which 2000 tons were exported from New Caledonia in 1978. Knowledge of the exploitable stock requires a good estimate of the surface area occupied by this gastropod. Trochus live in shallow water on reef flats consisting of dead coral slabs with many crevices and scattered coral rubble. This bottom type can be found on most of the reef formations in the lagoon (fringing reef, inner lagoon reef and barrier reef). The environment inhabitated by trochus stocks represents a large portion of the 20,000 km² of the New Caledonia lagoon. It is, however, difficult to access and synoptic methods of estimation are required such as SPOT. |
| Method: | In December 1983 the Groupement pour le Développement de la Télédetection Aérospatiale (GDTA) coordinated a SPOT Simulation Program focussed on New Caledonia. In addition to observation of reef environments, the program included bathymetric studies, identification of aquaculture sites and geological studies. |
| SPOT, launched into sun-synchronous orbit in February 1986, passes vertically over New Caledonia every 26 days at 10:15 local solar time. Because of off-nadir viewing the same area (60 × 60 km²) of the equator can be targeted 7 times during the 26 day cycle. Simulated SPOT imagery, recorded before the launching of the satellite, was achieved by radiometric methods using an airborne Deadalus radiometer. These data were restructured to give a radiometry equivalent to that of the HRV channels. | |
| The different spectral responses of the four channels are used to discriminate the different reef types: | |
| Channel 1 0.50 – 0.59 green 20 m resolution | |
| Channel 2 0.61 – 0.69 red 20 m resolution | |
| Channel 3 0.79 – 0.89 (near IR) 20 m resolution | |
| Channel 4 0.51 – 0.73 panchromatic 10 m resolution | |
| The green and red channels penetrate water to various depths; they are used together to differentiate underwater features at depths of 0 to 5 metres. At greater depths only Channel 1 (green) allows bottom types to be discriminated. The high resolution panchromatic channel provides morphological details. | |
| A first classification defined 5 classes of pixels. Ground verification and reference to black and white aerial photographs supported these 5 general themes and areas (refer to Figure 8.29). | |
| The pixels of the class identified as reef flat or hard bottom covers were then isolated, analyzed by principal components and classified. Five bionomic themes were identified and corroborated by ground verification (refer to Figure 8.30). | |
| Conclusion: | The processing of simulated SPOT image data successfully located the trochus biotope and permitted the calculation of the surface areas with reasonable precision. The repetitiveness of current SPOT coverage will permit the monitoring of habitat changes with time. |
Figure 8.29 Tetembia reef: general themes.
Figure 8.30 Tetembia reef: hard bottom themes.
| Reference: | Jensen, J.R. et al., 1980, Remote sensing techniques for kelp surveys. Photogramm.Eng.Remote Sensing, 46(6):743–55. |
| Remote Sensing Technique: | Satellite and Airborne Remote Sensing: Platforms - Aircraft, LANDSAT, SEASAT; Sensors - Aerial Camera, X-Band Radar, LANDSAT MSS, L- Band SEASAT Radar. |
| Objective: | To survey and monitor giant kelp resources off the California coast, with images obtained from LANDSAT, MSS, CIR photography, X-band airborne radar and L-band SEASAT radar. |
| Experimental Rationale: | The spectral signature of healthy Macrocystis pyrifera (kelp) was found to be similar to that of terrestrial orange brown vegetation. In the visible region of the electromagnetic spectrum, chlorophyll absorption in the green band was apparent with slightly greater reflectance in the blue and red. Kelp reflects 60–70% of the incident radiant flux in the region between 0.7- 11.0 micrometres. In contrast, water absorbs the majority of infrared radiation flux in this region, thus providing a good contrast between kelp and water. |
| The sensors which are sensitive to the spectral difference between kelp and water include normal colour and colour infrared (CIR) photography and multispectral scanning systems. Radar can also be used to discriminate kelp beds from the ocean, provided there is a difference in either surface roughness or dielectric constant between the two classes. Kelp normally exhibits a vertical relief of approximately 1 to 2 cm above the water surface while calm glassy ocean attains a surface relief of about 0.1 to 0.3 cm. It is hypothesized that this difference in surface roughness is sufficient to discriminate between kelp and water at certain radar wavelengths. | |
| Method: | During the period from 1975 to 1977 a base line inventory of kelp beds was obtained on a quarterly basis with the aid of 70 mm large-scale (1:24,000) colour photography and CIR vertical aerial photography. In this study high altitude CIR photographs (1:125,000) were obtained also from a NASA U-2 aircraft. The dates of the high altitude photography were chosen to coincide as nearly as possible with the dates of the large-scale inventories. |
| LANDSAT imagery was obtained on a repetitive basis for 18 days and classification procedures were conducted to discriminate kelp areas from water. | |
| An important advantage of microwave remote sensors over other types is their ability to image through fog. An extensive radar survey of the marine target was conducted with X-band horizontally polarized radar at a height of 6,500 feet. A vertically polarized X-band synthetic aperture radar system, equipped with 16 ft and 8 ft antenna, was also flown simultaneously at 5,500 feet. | |
| SEASAT-A images also were analysed for the detection of kelp beds. | |
| Results: | The kelp bed areas, calculated by means of low altitude photography, showed a very high agreement with those calculated by the high altitude photography. These results suggest that the analysis of high altitude CIR aerial photography can provide accurate measurement of kelp areal extent. The estimate derived from LANDSAT imagery was highly correlated with that of the aerial photography inventory. |
| The spectral nature of kelp and ocean was remarkably consistent in its clustering for three of the four LANDSAT images. The separation of kelp from ocean was due solely to the water penetration capability of band 4. | |
| Kelp acreage statistics obtained from manual interpretation of X-band radar imagery provided relative accuracy. The synthetic aperture system, however, consistently yielded overestimates while the real aperture system yielded underestimates. | |
| Conventionally derived kelp acreage statistics were not available for comparison with those obtained from SEASAT-A data. It was shown, however, that SEASAT radar signatures could be used for accurate kelp surveys. | |
| The most striking difference between X and L band imagery was the signature reversal between kelp and ocean. In the X-band imagery kelp was bright and the water was dark whereas in the L-band imagery, kelp was dark and water was bright. | |
| Conclusion: | This study showed that high altitude CIR photography (refer to Figure 8.31) and X-band radar imagery can provide areal extent data on kelp at approximately the same level of accuracy as conventional large-scale inventories. |
| LANDSAT data (refer to Figure 8.32) also provides accurate statistics if the consistent underestimation is corrected by using a simple linear equation. Given these results multispectral sensors in the 1980's offer potential for the operational monitoring of renewable kelp resources. |
Figure 8.31 Example of high altitude CIR photograhy (original scale 1:125,000) and manually interpreted kelp acreage surveys on four dates.
Figure 8.32 Kelp acreage surveys derived form four dates of LANDSAT image processing.
| Reference: | Belsher, T. and M. Viollier, 1984, Thematic study of the 1982 SPOT simulation of Roscoff and the west coast of the Contentin peninsula (France). In Proceedings of the Eighteenth International Symposium on remote sensing of environment, Paris, France. Ann Arbor, Environment Research Institute, pp. 1161–6 |
| Remote Sensing Technique: | Airborne Remote Sensing: Platform - Aircraft (Simulated SPOT); Sensor - HRV. |
| Objective: | To give a quantitative evaluation of the seaweed cover and to determine the reliability of the technique for species differentiation. |
| Experimental Rationale: | The SPOT HRV (Haute Résolution Visible) sensor is primarily designed for land observation. Seaweed species in the intertidal zone, like land vegetation, may be classified according to their particular spectral reflectances. The spatial resolution of the HRV (20m) permits discrimination of relatively small seaweed patches. |
| Method: | The interval zone was initially isolated for analysis by the removal of water and land data from the imagery. Water has a very low reflectance in the near-infrared and the corresponding pixels are recognized by simple thresholding on channel 3 (790–890 nm). For land, interactive operations were required to superimpose the shoreline from accurate charts. |
| A derived image was generated by computing a vegetation index Iv = | |
 | |
| where CH2 and CH3 referred to the raw data of the corresponding channels. Thereafter, thresholding of this image lead to the identification of the algae production zone since non-vegetation pixels were characterized by very low relative values of this index. | |
| A principal component transform was performed on raw data of the seaweed production zone on channel 3 to identify clusters of pixels of similar spectral signature. Then these spectral signatures were linked to individual species or dominant vegetation types by correlation with ground observed data. | |
| Conclusion: | The above vegetation index could provide an estimate of seaweed cover for a SPOT pixel area. Principal component analysis permits identification of five to six species. Refer to Figure 8.33 which shows the study areas after digital processing. The colours may be interpreted as follows: |
| green = Fucus (vesiculosus and serratus) | |
| blue = Ascophyllum nodosum | |
| yellow = Sargassum muticium | |
| orange = Ulva and Enteromorpha | |
| brown = Pelvetia canaliculata |
Figure 8.33 Digital processing of SPOT image Each colur represents a particular type of seaweed.
| Reference: | Armstrong, R.A., 1983, Marine environments of Puerto Rico and the Virgin Islands: automated mapping and inventory using LANDSAT data. Caribbean Fishery Management Council, 37p. |
| Remote Sensing Technique: | Satellite Remote Sensing: Platform - LANDSAT; Sensor - MSS. |
| Objective: | To explore the possibility of discriminating marine communities of Puerto Rico and the Virgin Islands with LANDSAT imagery. |
| Experimental Rationale: | The different spectral signatures associated with submerged plant life and reefs in coastal seas can be utilized to differentiate marine communities. |
| Method: | LANDSAT imagery with less than 30% cloud cover was analyzed to discriminate coral reefs and other large marine communities for Puerto Rico and the Virgin Islands. |
| It was shown that a change in sensor mode from low gain to high gain caused a 25% increase in water penetration power from 6 m to 8 m. | |
| The digital imagery used in marine community classification was subjected to preprocessing procedures to eliminate the distortions caused by atmospheric attenuation (radiometric distortion) and curvature of the earth (geometric distortions). | |
| Supervised classification with user-defined training areas was carried out for images of St. John where up- to-date bathymetric and topographic maps were available for selection of training areas. The sea around St. John was classified into 6 regions using a digital image analysis system. Utilizing the spectral signatures obtained from the St. John image the sea around Culebra and Vieques was then classified into 5 regions, St. Croix into 7 regions and Puerto Rico east and north into 6 regions each. | |
| Results: | The sea around St. John was classified as follows (refer to Figure 8.34): |
| Water Class 1- deep water 18–31 m; | |
| Water Class 2- 18 m contour; | |
| Water Class 3- 18 m contour with submerged vegetation; | |
| Water Class 4- shallow sand bottom; | |
| Water Class 5- fringing coral reef; | |
| Water Class 6- coral reef at a depth of 4–6 m. | |
| A certain degree of classification error was found in water class 4; although shallow sand bottom was correctly identified in the St. John area, coral reef crests were identified in this water class around St. Thomas, St. Croix, and Culebra. | |
| The land classification was generally not successful although large vegetation areas of mangrove were accurately classified on Puerto Rico using hybrid supervised/unsupervised classification procedures. | |
| Conclusion: | The resolution of LANDSAT MSS imagery was found to be adequate for the discrimination of large marine communities such as coral reefs. With low gain imagery, submerged features to a depth of 6 m could be classified accurately. |
| Recommendations: | 1. Only high gain imagery with high quality band ratings for MSS-4 and MSS-5 should be used for studies of coastal environments and general marine use. |
| 2. The initial classification run should be done in an area that is familiar to the user or with adequate ground truth information. | |
| 3. LANDSAT MSS data are useful for the study of large marine communities in the depth range of 0–5 m. Submerged features which are relatively small and submerged features which have deeper bathymetric ranges, however, will remain unclassified. In such cases a high-resolution airborne MSS with several narrow bands in the 0.4–0.7 micrometre range, or the Thematic Mapper which has three bands in the 0.4–0.7 micrometre range (versus only two in LANDSAT MSS) and a higher (30 m) resolution, should be used. |
Figure 8.34 Classified image of St. John, US Virgin Islands.
| Reference: | Middleton, E.M. and J.L. Barker, 1976, Hydrographic charting from LANDSAT satellite: a comparison with aircraft imagery. In Oceans '76. Second combined conference, Marine Technology Society/ Institute of Electrical and Electronics Engineers. New York, IEEE Inc. and Washington, D.C., MTS, (CH 1118–90 EC):6 p. |
| Remote Sensing Technique: | Satellite and Airborne Remote Sensing: Platforms - LANDSAT-2, U-2 Aircraft; Sensors - MSS, OCS (Ocean Colour Scanner). |
| Objective: | To identify, quantify and isolate depth information from other factors to permit study of superimposed transient phenomena such as red tide. In pursuit of this objective, an additional application of considerable merit, namely hydrographic mapping of coastal areas, was recognized. |
| Experimental Rationale: | Analysis of imagery produced by the MSS of LANDSAT-2 and OCS on a U-2 aircraft revealed that depth-related information represents the major contribution to the total recorded radiance in those satellite/aircraft sensor channels that are expected to be of most value for phytoplankton and sediment detection. |
| Method: | LANDSAT-2 imagery of the Gulf of Mexico and Tampa Bay were analyzed for hydrographic charting. All the images used (except one) were recorded on the high gain mode. |
| Simultaneous coverage was provided by the Ocean Colour Scanner flown at an altitude of 20 km on a U-2 aircraft. A sub-area of both OCS and MSS imagery along the west coast of Florida was chosen for intensive study. It should be noted that OCS was subsequently renamed Coastal Zone Colour Scanner (CZCS). | |
| Results: | The depth was effectively discriminated with the LANDSAT MSS-4 (channel 4 – 0.5 to 0.6 micrometres) imagery taken in the high gain mode. Contours to a depth of at least 8 metres were recognized with 10% error over a radiance intensity of 42 grey levels. A 26 grey level range of MSS-5 (0.6 to 0.7 micrometres) gave depth information to at least 5 metres with 10% error |
| The simultaneous coverage of OCS and MSS on September 19, 1975 provided a good opportunity to compare the performances of the two sensors. The channels 4 and 5 of OCS were found to be the most sensitive for the discrimination of depth or depth related factors. The greatest sensitivity as indicated by a radiance intensity range of 56 grey levels was found in OCS-4 which measured depth to at least 12 meters with 5% error (refer to Figure 8.35). | |
| Conclusions and Recommendations: | The available channels of two remote sensors were analyzed spatially and temporally for their capacity to discern bathymetric information. The OCS-4 provided the greatest depth penetration and discrimination with the smallest error estimate; therefore, it is recommended for small-scale hydrographic mapping of sandy bottom coastal areas. |
| LANDSAT's MSS-4 in the high gain mode is recommended when mapping at a scale of 1:80,000 or smaller is adequate. Bottom depths and the log of radiance intensities were found to be related by a linear relationship for depths down to 12 metres. This shows the possibility of depth estimation in this range given several radiance values. |
Figure 8.35 The charted depth contour (5 m) in the study site in comparison with the binary print of the OCS-4 radiance-value distribution pattern for a single grey level in this depth range.
| Reference: | Roy, S.E., 1978, Sea surface temperature and related measurements of the South Caribbean Sea, utilizing GOES, NOAA and GOSSTCOMP data for locating structures. In Proceedings of the Seventh annual remote sensing of earth resources conference. Tullahoma, Tennessee, University of Tennessee, Space Institute, pp.261–87. |
| Remote Sensing Technique: | Satellite Remote Sensing: Platforms - GOES, NOAA; Sensors - VISSR, VTPR, VHRR (Very High Resolution Recorder), SR (Scanning Radiometer). |
| Objective: | To determine the sea surface temperature (SST) and related measurements associated with ocean currents and structures such as upwelling and eddies by the analysis of images from GOES and NOAA satellites. |
| Experimental Rationale: | The temperature of an object is related to the IR radiation coming from it. Since the launching of the first NOAA satellite, the National Environmental Satellite Service (NESS) has developed a process to convert IR data into sea surface temperature information. Subsequently, interactive processing of NOAA satellite data using atmospheric attenuation data obtained from a Vertical Temperature Profile Radiometer (VTPR) has led to an improved quality of SST measurements. A further improvement of analytical procedures led to the development of an operational and highly flexible model called Global Operational Sea Surface Temperature Computation (GOSSTCOMP) which permits rapid retrieval and updating. |
| Method: | The data for this study was obtained from a geo- stationary satellite of the SMS/GOES series and distributed by NESS. The visible and IR Spin Scan Radiometer (VISSR) data recorded on tape and relayed by telephone line for facsimile reception was the chief source of data for the Caribbean Sea study area. |
| The surface temperature measurements were obtained by a statistical histogram analysis. The corrections for atmospheric attenuation were calculated from the VTPR data. | |
| The quantitative measurements of sea surface and land surface temperatures, however, are still affected by noise, the extent of cloud and haze in the retrieval area and atmospheric moisture content. In addition, sea state condition, surface reflectance, etc. influence the accuracy of sea surface temperature measurements. For accurate temporal monitoring of ocean phenomena, the images had to be geometrically corrected; land markers were used for accurate registration purposes. | |
| Results: | The results show that the (GOSSTCOMP) charts obtained weekly were too broad to discern any local structures. Digitally enhanced images, however, clearly showed SST and defined most structures. The digital enhancement program with 12 step increments of 0.5°C over a 6°C range was found to be the most convenient for general use of GOES and NOAA data in the subtropical region. |
| Conclusion: | Thermal IR imagery from SMS/GOES can be used to derive SST on an operational basis over the Caribbean Sea region. NOAA series data, although of higher resolution, could not be acquired and analysed cost effectively in near real time. |
| However, because of excessive cloudiness, atmospheric moisture content and a high general albedo over the sub-tropics, VTPR data from the NOAA series is essential for the correction of atmospheric attenuation on SMS/GOES imagery. |
| Reference: | Mattie, M.G. and D.E. Lichy, 1980, SEASAT detection of waves, currents and inlet discharge. Int.J.Remote Sensing, 1(4):377–98. |
| Remote Sensing Technique: | Satellite Remote Sensing: Platform - SEASAT; Sensor - SAR. |
| Objective: | To demonstrate the capabilities of SEASAT Synthetic Aperture Radar (SAR) to image certain coastal and nearshore phenomena during specific orbits. |
| Experimental Rationale: | Because radar wavelengths do not penetrate the sea surface, the backscatter pattern indicates its condition. Surface winds are clearly distinguishable in addition to surface interactions which have little relationship to the wind, such as currents, eddies and ship wakes. |
| Method: | Duck X was a 2 month experiment conducted during August to October 1978 off the coast of the U.S.A. to test the SEASAT SAR. For the verification of SEASAT SAR images various types of sensors were used including airborne photographic and radar imagery, meteorological satellite imagery, land-based radars and conventional wave gauges. |
| Results and Conclusion: | The data set obtained by SEASAT SAR (orbit no. 974, 1339), although relatively sparse, contained a great deal of information on a variety of phenomena including ocean currents, ocean surface waves and coastal inlet discharge. Ocean waves with heights as low as 0.1 m and their shoaling were imaged. Hurricane Ella passed 300 km east of the SAR swath at Cape Hatteras and its effects were observed. |
| The SEASAT SAR images enabled the Gulf Stream boundary and eddy features to be identified. Coastal inlet hydraulics and marshland water levels were also seen in SAR images. The other phenomena identified in SEASAT SAR images included: wave diffraction, bottom sand waves, hydrographic features and internal waves. SEASAT SAR also supplemented information obtained by other instruments which are weather limited, such as IR photography. | |
| Figure 8.36 shows a digitally processed image indicating some of the multitude of phenomena that have been noted in the SAR imagery. The dark regions generally correspond to low surface winds. A low energy swell system with 200 m wavelength is travelling through the area from the south-east. |
Figure 8.36 SEASAT SAR image. (resolution of 25 metres, horizontal dimensions 100 km).
| Reference: | Tanaka, S. et al., 1983, Accuracy of direct measurement of mean surface water velocity of the Kuroshio using multi-temporal NOAA-6 imageries. In Proceedings of the Seventeenth International Symposium on remote sensing of environment, Ann Arbor, 1983. Ann Arbor, Michigan, Environment Research Institute, pp. 933–44. |
| Remote Sensing Technique: | Satellite Remote Sensing: Platform - NOAA-6; Sensor - AVHRR. |
| Objective: | To determine the surface water velocity of the Kuroshio current using the “Sea Mark” method in NOAA-6 images. “Sea Mark” is a point pattern on the sea surface, which can be identified in at least 2 consecutive images. |
| Experimental Rationale: | The movements of “Sea Marks” show the direction of the currents. These “Sea Marks” tend to keep their configuration intact since water drifts as a mass. Tracking the prominent points of the fronts or the centres of eddies gives the necessary current vectors. |
| Method: | Two NOAA-6/AVHRR images recorded on the same day were used in this study (refer to Figure 8.37). The time difference between these two scenes was 11 hours and 17 minutes. Both scenes were mapped in relation to the earth coordinate system to measure accurately the velocity vectors. Ground control points were used in the mapping procedures. |
| The maps produced by this method contained a maximum error of 0.2 mm regardless of mapping scale, i.e. an error of only 600 metres in horizontal distance on the 1:3,000,000 map. | |
| Results: | Because of the temperature differences between the Kuroshio and surrounding water, clear fronts could be identified by the imagery. In addition oceanic eddies probably generated on the shallow sea bottom south of Tanegashima Island were recognized and the movement of their prominent points was traced in the consecutive images. The distances these points moved within the 11 hours and 17 minutes time interval were used to calculate the average velocity of the eddy. |
| Sea current phenomena were found to vary in 12 hour intervals. Measurement accuracy of mean surface water velocity was within 0.1 knots based on an assumed “Sea Mark” orientation accuracy of “2.0 pixels”. | |
| Conclusions: | The accuracy of measuring the mean surface velocity of an ocean current such as the Kuroshio depends on the following: |
| a) the ability to recognize special floating points on both NOAA-6/AVHRR images recorded on the same day; | |
| b) the necessity to integrate (or map) the NOAA- 6/AVHRR imagery into the earth coordinate system. The ground control points necessary for this transformation should be taken from uniformly scattered points within the scene. |
Figure 8.37 “Sea Mark” chase method for current vector measurement.
