Our case studies are intended to give as wide a review as possible of recent examples of the practical applications of RS and GIS to aquaculture and inland fisheries. Although we attempted to balance the studies proportionately between RS and GIS and between aquaculture and inland fisheries, it was difficult to do the latter since few RS or GIS studies have been specifically directed towards inland fisheries. In lieu we have selected some relevant applications which look at various properties of the lake environment. A few of our case studies are not directly related to either aquaculture or inland fisheries, but they are included so as to give an idea of the potential which could be applied to those fields. We have concentrated where possible on studies relating to the developing world, since we feel that this is where these examples can most usefully apply. The RS studies are placed first, followed by the GIS studies, and then the one study which covers both - within these areas of RS and GIS the aquacultural studies precede the “others” studies. Finally, we have attempted to structure the content of most case studies in a similar logical sequence.
TITLE: “The Preselection of Sites Favourable for Tropical Shrimp Farming”
AUTHOR: CNES - IFREMER
PUBLICATION and DATE: IFREMER - France Aquaculture. 1987.
The French Institute of Oceanographic and Coastal Studies (IFREMER), together with the National Centre for Space Studies, CNES), are interested in promoting the use of SPOT satellite imagery for improving aquatic resource use and for coastal management generally. They have produced a quantity of promotional materials, much of which is devoted to aquaculture. The above publication is not written in great detail but it has been included since its images are clear and very relevant to mariculture. Further details on the value of SPOT to coastal resources can be found in Loubersac (1988).
Tropical prawn culture has been the most actively developed sector of world mariculture over the past decade. During this period production has doubled every two to three years. The farming of prawns and shrimp is being developed mainly in tropical areas. About 70% of the current world production, which totals 350 000 tons, is produced in S.E. Asia. South and central America produce most of the rest and production has expanded more in Ecuador recently than in any other country.
Many tropical areas tend to have complex coast/land interfaces, typically consisting of low lying silty and sandy soils, often covered with dense mangroves and interspersed with channels, rivers, distributaries, marsh, lagoons, etc. This has meant that mapping has been difficult and/or cartographic standards have been poor. For both intensive and semi-intensive commercial shrimp and prawn production, extensive flat areas are required where large earth ponds can be excavated. These need to be connected via a series of water channels to a salt water source - frequently a sheltered estuarial or deltaic environment. Figure 7.1 shows a SPOT false colour image of a typical network of ponds in Ecuador. The dark ponds are those in production (having deeper water), the light blue are presently empty (having shallow water) and the red is permanent land.
Figure 7.1 SPOT Image Showing Shrimp Ponds Near Guayaquil, Ecuador
The criteria for site selection and assessment of potential involve complex considerations ofmeteorological, hydrological, geomorphic and socio-economic factors. However, as IFREMER state, the necessary secondary data sources are not always available, or they are out of date; the relevant coastal features are either not on the maps or they are inaccurate, and in certain tropical countries aerial flights are either forbidden or made impossible by bad weather. Hence the advantages of satellite RS imagery.
Figure 7.1 gave a very clear indication of how existing sites can be imaged by SPOT. Clearly, for the purposes of preselecting sites it would not be practicable to produce images at this scale on a world wide basis. Figure 7.2 shows an image of the S.W. coastal area of New Caledonia, having an original scale of 1:150 000. Potential sites for shrimp production have been circled. Various image enhancement and enlargement processes would then be performed on the digital data covering these areas. Figure 7.3 shows an example of this, where the Bouches du Diahot area (seen in the top right circle on the Figure 7.2 image) has been enlarged to a scale of 1:25 000, and land use colouring has been applied. Any of the areas indicated with vertical black lines would be worthy of further ground inspection and more detailed site assessments.
Figure 7.2 SPOT image showing preselection sites for aquaculture in part of New Caledonia
Figure 7.3 Enlargement of a SPOT Image Showing the Bouches du Diahot Area of New Caledonia in Detail
The availability of real time and precise geographic data for identifying site potential is offered by imagery such as SPOT. This data can lead to exact locational positioning and to precise aerial quantification - a positive aid to decision-making. The digital form of this data allows for analysis under a huge array of circumstances, at selected scales varying from 1:20 000 to 1:400 000. IFREMER claim to have a selection of images for all tropical coastal areas available and, since SPOT imaging has now been in progress for more than four years, temporal analysis of coastal developments are already possible. Where airborne sensing is not possible, or where it is not easily available, then clearly the resolution achievable by SPOT is sufficient for identifying potential locations for shrimp culture. For understandable reasons, this promotional material fails to mention the additional potential which digitized SPOT imagery would have as source data for a range of more complex GIS investigations. And, although the type of “map” which Figure 7.3 represents, is a valuable and up to date aid, far more data is needed before a realistic site selection process can be accomplished.
TITLE: “Feasibility of Using Remote Sensing to Identify the Aquaculture Potential of Coastal Waters”
AUTHORS: Cordell, E.V. and Nolte, D.A.
PUBLICATION and DATE: Published by Recon Technologies, Inc. 1988.
This study is included for several reasons. It is one of the very few examples where RS, and a limited amount of GIS, has been applied directly to aquaculture. It is also extremely detailed and there is an accompanying publication - a User's Guide to Remote Sensing which is even more detailed! Because the study is so extensive, we could not do justice to it if we were to review it in the normal way. Instead we will review Chapter 5 only (using the headings given by the authors). This chapter is what is called an “Interpretation” of the results of the study.
Overall, the study aims to see how useful RS imagery can be to resource allocation in a remote, but valuable, location where other means of investigation would be time consuming and therefore costly. To provide a focus for the study, the specific aquaculture category of Pacific oyster rearing was chosen. The area chosen was S.E. Alaska; more specifically the Etolin island area. This area is remote from major population centers, mountainous and forested; there are numerous inlets and small islands; steep slopes cause many landslides and the high rainfall leads to a significant freshwater run-off with consequent turbidity. The natural water conditions are frequently ideal for oyster growth and a number of commercial farms operate in nearby island groups.
Information gathering from remote sensing
For marine environment applications, RS can provide information on:
Variables in remote sensing
There are several factors which can influence the quality of RS images and can affect whether or not they are even worth acquiring, e.g.
The imagery interpretation process
The authors set up an imagery interpretation process (Figure 7.4) which they activated upon receipt of each photographic image. Flawed images, or those having too much cloud cover, were rejected and alternatives sought. If images were good, and resolution was adequate, then information on turbidity was plotted onto a base map using either a zoom transfer scope or interpolation. Information on ice was compiled in the same way; other features such as land-use could be similarly plotted. If the scale was poor, then image enhancement using optical magnification was used. If the imagery features were difficult to see, then optical filters were used. These simple, low-cost techniques were able to detect much information on turbidity flows in the study area, and to a more limited degree, the presence of sea ice. Sea conditions were mapped using NASA U-2 colour infrared imagery. This data, along with land mass configurations, was utilized to predict areas with shelter from possible rough sea conditions.
Some comments on individual imagery were given:
Landsat. This was in 9 × 9 in. positive transparencies which provided information on suspended sediment loads. When the positive transparencies were processed into negatives, so that enlargements could be made, it was noted that the reversed colours of the negative displayed the sediment better than the original positives. The quality was good but the ground resolution (at 1:1 000 000) was poor.
SPOT. This imagery was received as a positive transparency at a scale of 1:250 000. The quality was good but it had limited utility. The spectral characteristics prevented it from detecting any sediments and, as the image was acquired at high tide, it proved to be of little value for determining shallow water depths. It did detect ice formations and provided data on land use activities.
Advanced Very High Resolution Radiometer (AVHRR). This imagery was in negative format and of good quality. It revealed water temperature differences in the study area. Resolution was poor with the smallest area of temperature difference being detected at 1.1 km2.
Heat Capacity Mapping Mission (HCCM). Imagery from this sensor was in negative format and of good quality. However, since it only revealed gross water body temperature differences, its utility was limited.
Coastal Zone Colour Scanner (CZCS). CZCS imagery was received in negative format. Its resolution was even worse than the HCCM and AVHRR. The images had extensive cloud cover so they proved to be of little value.
Alaska High Altitude Aerial Photography (AHAP). This colour IR imagery was acquired from a NASA U-2 aircraft and was in positive transparency format. It provided vast amounts of data on suspended sediments, wave directions, water currents, land use activities, shallow water bathymetry and it provided locations of stream outlets to the shoreline. It was the most cost-effective source of RS data.
Figure 7.4 RS Image Interpretation and Analysis Processes
Testing remote sensing imagery data validity
If the imagery interpreter is unfamiliar with an area, there are certain techniques that should be adopted in order to reduce image interpretation errors. Ideally the interpreter should research and read as much information as possible about the study area prior to actual interpretation, or ground (or water) truthing might provide a more accurate interpretation aid. The authors point out the costs involved in ground truthing and suggest that a sampling frame should be used.
The use of AutoCAD as a database
To compile the data from the various sources of information, the following problems had to be overcome:
There are several manual methods of overcoming these problems. These were briefly described, though they were rejected because of the many time-consuming difficulties which were involved and because there are now PC automated design programs which allow the necessary mapping adjustments to be made.
The system used in this study was AutoCAD, using an IBM PC/AT with 40 megabyte hard disk memory, a digitizing pad and an HP plotter. All of the data sets were standardised (registered) to a selected base map outline, a version of which is shown as Figure 7.5. The UTM projection was selected, and all measurements were metricated so that AutoCAD could function efficiently and so that the data base structure was simplified. Digitizing was then performed and data bases were created for the following “layers”:
The authors then describe some of the various manipulation functions which AutoCAD could perform, plus the assigning of labels and feature attributes. The program Surfer was used to draw a 3D view of the Stikine Strait area. This view (Figure 7.6) clearly shows the steep under-water topography and therefore the area's unsuitability for oyster development.
In order to prioritize the many factors that determine the suitability of a location for oyster culturing, a scoring technique was suggested. The factors scored in a preliminary example were:
The scoring, as applied to four sites which were selected to test the methodology, is shown in Table 7.1.
Figure 7.5 Base map of Etolin Island Showing Selected Oyster Sites and Turbidity Zones
Figure 7.6 Bathymetry of Stikine Strait Area Looking S.W.
|Site||Area Size||Mean Depth||Turbidity||Sea Ice||Shelter||Total Score|
|1. Area Size:||1 = < 1 hectare|
|2 = 1 to 2 hectares|
|3 = > 2 hectares|
|2. Mean Depth:||1 = < 5 meters or > 20 meters|
|2 = 20 to 15 meters|
|3 = 15 to 10 meters|
|4 = 10 to 5 meters|
|3. Turbidity:||1 = moderate turbidity (summer)|
|2 = low turbidity (summer)|
|3 = slight turbidity (summer)|
|4 = no turbidity (summer)|
|4. Sea Ice:||1 = winter sea ice|
|2 = possible sea ice|
|3 = no sea ice observed|
|5. Shelter:||1 = occasional high seas possible: two sides protected|
|2 = rare high seas: three sides protected|
|3 = protected on four sides|
Table 7.1 shows that, using the criteria listed, the areas around Blashke Island and Jadski Cove appear well suited for the development of oyster culture. The lack of suitability of the Stikine Strait area is well identified. However, other siting criteria might well want to be considered such as:
Using the existing criteria for site selection a number of additional areas were identified as being suitable for oyster culture and these were listed.
This section of the study concluded with a list showing the estimated accuracy of using the AutoCAD software, plus a comment on some of the additional statistical data which was produced.
TITLE: “Remote Sensing and Model Simulation Studies of the Norwegian Coastal Current During the Algal Bloom in May 1988”.
AUTHORS: Johannessen, J.A., Johannesses, O.M. and Haugan, P.M.
PUBLICATION and DATE: The Nansen Remote Sensing Center. Technical Report No. 16. September, 1988.
We will mainly concentrate on the RS aspects of this study, which have also been detailed in Petterson (1989).
There have been an increasing number of so-called “red tides” reported globally during the last decade. They constitute unexpected blooms of algae, some of which are toxic such as the Chrysochromulina Ipolylepis, which may occur in oceans, seas or lakes. The particular bloom reported above occurred during May - June, 1988 in the Skagerrak - Kattegat area of Scandinavia. It was thought to have been the result of the following environmental conditions:
For four weeks the bloom moved eastwards from southern Sweden, along the southern and south-western Norwegian coast. It caused major fish kills to both wild and caged fish. About 480 tons of caged fish (mostly salmon) worth over NOK 30 million were lost, and some 200 marine fish farms were evacuated during the bloom, usually to brackish, inner fjord waters. The study aims to show how the blooms can be detected, using either airborne or satellite sensors, and how a temporal study of bloom movement can enable modelling of future movements, both in an ongoing bloom situation and in any subsequent bloom occurrences.
The Norwegian State Pollution Authority carried out daily air patrols over a 10-day period, covering the coastal region between Oslo and Bergen. The airplane was equipped with a Side Looking Airborne Radar (SLAR), which had proved to be capable of detecting ocean fronts (the leading edges of bodies of warm or cold water). A visual colour monitoring of the bloom and the front was possible (from the aircraft) in calm sea conditions, which eliminated the necessity for ground truthing. Two other aircraft took part in the monitoring, though for shorter periods. They were both equipped with infrared (IR) sensors, which were capable of detecting temperature differences in the surface water, and they confirmed that the bloom was advancing westwards along with warmer water.
Advanced Very High Resolution Radiometer (AVHRR) data from NOAA 9 and 10 weather satellites were received at Tromso satellite station. These data contained information on the sea surface temperature. The data tapes were sent to the Nansen Remote Sensing Center (NRSC). and about 6 hours after the satellite pass, the corrected and processed images were able to give information on the sea surface temperature distribution and the advection and movement of the ocean fronts (Figure 7.7). The temperature and frontal information were used as inputs in the NRSC numerical model of the ocean circulation in the region.
Figure 7.7 Sequence of NOAA Infrared Satellite Images Showing Development of an Algal Bloom Over the Skagerrak in 1988
NOAA infrared satellite images resolving the surface temperature in the
Skagerrak and along the southern coast of Norway during the spring algal
bloom in 1988. Images are respectively from May 15 (a), 21 (b), 22 (c), 27 (d),
30 (e), 31 (f). Color code: Yellow: warm water ≈ 10°C, Red ≈ 8°C, Blue:
colder water ≈ 5°C, Green: land and warm water (on May 27) > 10°C, Black:
Clouds and snow in the mountains. The algal front followed mainly the
warm water temperature front around the southern coast of Norway, before
it culminated on May 29.
(Images processed at Nansen Remote Sensing Center).
During the first stages of the bloom, observations indicated a close correlation between the algal front and the satellite-derived surface temperature front. The spreading and advection of the algae was thus indirectly monitored by RS infrared data during cloud-free periods. Figure 7.7 shows a large amount of warm surface water (yellow to red = >8C) in the central Skagerrak. Whilst the warm water was confined to eastern Denmark and the eastern Skagerrak during late April and early May, it gradually covered the entire Skagerrak between 15 and 30 May. No similar evolution is observed in the central North Sea, where the mean surface temperature remains nearly constant as represented by the dark to light blue colours.
A combination of airborne, satellite RS and in-situ observations were able to closely monitor the movement of the bloom, noting its speed and any “eddies” and “meanders” in the algal front. Meanders in coastal ocean movements are caused mainly by bathymetric variations, i.e. usually topographic ridges.
The experience from the operational forecasting during this algal bloom demonstrated the need for improved understanding of the connection between the chemical, biological and physical mechanisms steering the plankton distribution. For monitoring of the algae, the airborne observations were important on cloud covered days when the satellite sensors gave no surface information. However, the lack of airborne RS instrumentation capable of mapping the algal distribution limited the benefit of aircraft surveillance. It is likely that the Synthetic Aperture Radar (SAR) and Radar Altimeter (RA) instruments to be carried on ERS-1 will prove valuable in providing weather independent measurements of ocean circulation patterns, especially in areas such as Norway which has a high cloud frequency. The unexpected occurrence of this toxic algal bloom, accompanied by huge losses to the sea farming industry and the local ecological system, clearly demonstrate the need for a marine or coastal monitoring system based upon real-time extensive RS imagery.
TITLE: “Satellite Remote Sensing to Locate and Inventory Small Water Bodies for Fisheries Management and Aquaculture Development in Zimbabwe”
AUTHOR: Kapetsky, J.M.
PUBLICATION and DATE: CIFA Occasional Paper No.14. F.A.O. 1987.
Fisheries and aquaculture in artificial impoundments (small water bodies - SWBs) in Zimbabwe appear to have excellent development potential. It has been estimated that the total surface area of the SWB is between 21 000 and 45 000 ha (Kenmuir, 1981). At an average assumed yield of 300 kg/ha, a total of about 7 000 tons of fish could be produced in these SWBs each year, i.e. if the surface area were only 21 000 ha. This could make an important contribution to the overall supply of fish in Zimbabwe. Among the fundamental questions to be answered by this survey were:
If answers can be found to these questions it would be possible to devise a plan for fisheries and aquaculture development in SWBs.
The objective of this study was to show how satellite RS could be a rapid, cost-effective means of providing this required information on SWBs, using Zimbabwe as a test area. Specifically, the study includes the following elements:
We will not give equal consideration to these elements since we are primarily concerned with showing the use of RS.
The study was based on the visual analysis of two Landsat Thematic Mapper images of N.E. Zimbabwe (Figure 7.8), one corresponding to the end of the dry season (August, 1984), and the other to the end of the rainy season (April, 1985). Each image covers an area of about 32 400 km2. In order to enhance not only the SWBs, but also to identify them in the context of landforms and land uses in which they occur, three bands were used together. Colour composites of Bands 1 (water turbidity), 3 (terrestrial vegetation) and 4 (water) were made. In order to facilitate the location and identification of SWBs alone, separate images of each date were prepared using only Band 5. Each image was printed in the form of a false colour photograph (Bands 1,3,4) or a black and white photograph (Band 5) at a 1:250 000 scale.
Using the rainy season colour composite, water bodies were identified and circled on a mylar overlay. Criteria for identification were drainage context, shape, character of the surrounding area and colour. A magnifying glass was necessary to verify many of the small water bodies. The mylar overlay was transferred to the black and white (Band 5) image and water bodies not visible on the colour composite were located and circled.
In order to estimate surface areas and changes in surface areas, the same SWBs were located on the dry season image and the change in surface area was calculated. Surface areas were estimated using a dot grid overlay. Each dot represented an area of 4.44 ha. No attempt was made to estimate partial dots. In this same sample the hue of each water body was noted to provide an indication of turbidity, which may suggest levels of aquatic productivity.
In total; 906 water bodies were identified on the rainy season image, i.e. within the area covered by the RS image (shown in Figure 7.8). 82% of the water bodies were associated with intensive agriculture, 13% with rangelands (including some cropland) and the other 5% with rugged highlands. Their total aggregate area, as measured from the satellite imagery was about 10 000 ha. The highest density of SWBs was also associated with intensive agricultural land, i.e. approximately 9 SWBs per 100 m2. The density of SWBs in rugged highland areas was only 0.6 per 100 km2.
Figure 7.8 Area of N.E. Zimbabwe Covered by Landsat Thematic Mapper Image
To analyze changes in SWBs from the rainy to the dry season, a small area of the main map was used, i.e. immediately to the N.E. of Harare (see Figure 7.8). Here, the surface areas of 43 SWBs were estimated on both the dry season and rainy season images. Based on rainfall data for the 1983/84 to 1985/86 rainy seasons, it can be inferred that most water bodies were full or nearly so when the April 1985 imagery was acquired. In contrast, the August 1984 image coincided with the end of the previous dry season which was itself preceded by a rainy season of less than 600 mm total rainfall. Thus the August imagery can be taken to represent the minimum water surface remaining at the end of the dry season. At the end of the rainy season 42% of the water bodies are less than 4 ha, and 23% are between 4 ha and 9 ha in size. The total water surface area in this part of the image was 475 ha. Of the 43 SWBs measured in the rainy season image, only 32 could be found on the August dry season image. The total surface area of these SWBs was 244 ha, an overall decrease of 49%.
In practice, only the colour of the larger water bodies could be discerned by visual methods. Of the 14 SWBs in this category, half were black indicating a minimum of turbidity or productivity, four were dark grey suggesting a small amount of turbidity, one was grey, one was green perhaps indicating relatively high organic turbidity and the other was light blue suggesting high levels of inorganic turbidity.
The results of this study indicate that satellite RS can be valuable in locating and counting SWBs for potential fishery and aquaculture development, i.e. if the advantages of the technology are fully utilized and its limitations understood. As a means of checking the final results, Kapetsky compared his findings with Kenmuir's 1981 study, in which aerial photographs were used. The comparisons are shown in Table 7.2. There are several causes for differences between the RS and aerial photo results. Because of the small scale of the RS imagery (1:250 000), it is likely that many of the smallest water bodies (<1 ha), were not included. Also, the images might have been gained at different seasons, i.e. Kenmuir's images are not dated. Furthermore, the satellite-derived estimates are taken from a slightly different area and finally, with the satellite images there is a possibility for misidentification. As a follow-up, ground verification is required to obtain a better estimate of the reliability of the results.
|Satellite (Kapetsky)||Air Photos (Kenmuir)|
|Number of SWBs located in area||43||77|
|Total estimated area (ha)||485||673|
|Mean area of SWBs (ha)||11||8.7|
|Area range (ha)||4 to 53||0.8 to 45|
Because of the comprehensive coverage of the country and good quality of the imagery, the use of satellite RS for the detection of SWBs in Zimbabwe is technically feasible. Visual analyses would be less costly than computer processing, but at the same time would limit the scope and utility of results. A combination of Landsat MSS and TM data would satisfy technical requirements and also be cost-effective. Should they be available from other studies, aerial photographs or SPOT imagery could be used to supplement the Landsat data.
TITLE: “Use of Remote Sensing in the Study of the Changing Shoreline of Sarykamysh Lake”
AUTHOR: Nuriddinov, O.S.
PUBLICATION and DATE: Mapping Sciences and Remote Sensing; Vol.26, No.1. pp74– 77. 1989.
This is one of a number of recent papers showing the use of RS to monitor changes in lake size or shape. This particular study was selected because of its uniqueness in that a huge lake has been inadvertently created by man's interference with the hydrological cycle and, as will be shown, the lake now constitutes a viable inland fisheries resource. Studies showing lake size fluctuations over time could be vital for the successful implementation of inland fisheries development, and the main objective of this study was to test the feasibility of using RS photographic images to study the shoreline dynamics.
The use of extensive irrigation systems in parts of Soviet Central Asia are dramatically altering the surface and sub-surface water balances in the region (Sigalov, 1986). One of the largest reservoirs in the region, the Sarykamysh Lake, has gradually been forming and expanding since 1961 when wastewater from excessive irrigation began accumulating in a large natural depression near the delta of the Amu Dar'ya river which flows into the Aral Sea. It is estimated that between 1961 and 1972 some 18 km3 of water flowed into this depression to create the lake, and a further 30.8 km3 of inflow occurred between 1973 and 1980. The water balance of the lake during the 1970s was such that 53% of the incoming water evaporated, 6% was lost to infiltration and 41% went to further filling the lake.
Black-and-white space imagery was obtained for the period 1973–1985, and it was used to reveal changes in the appearance of the lake and to assess its current condition. A map of changes in the lake was compiled from the results of image analysis (Figure 7.9), which provided a graphic illustration of the expanding reservoir. The position of the lake's shoreline, as determined from a 1965 topographic survey, was used in the map construction. The volumetric characteristics of the lake were determined from the comparison of its aerial characteristics with observations on the depth of the water in the lake.
Analysis of the RS imagery provided a temporal picture of the development of the lake. By 1968 the eastern part of the present lake basin was inundated. With the continued inflow into the lake, the water then spread out over an extensive gently sloping area to the east, southeast and west. By 1975 practically all the south-eastern and half of the western part of the basin were flooded - the water area of the lake having increased more than 10-fold relative to 1965. As a result of the continued growth in the volume of incoming irrigation wastewaters, further displacement of the shoreline to the southeast, east and north has occured since 1975. The greatest changes, however, have occured in its southern and western parts because of the intensive flooding of a vast plain consisting of an area of saline soils and playa deposits. The water area of the lake during the 10 years from 1975 to 1985 increased by 80% and the water volume by 115% (Table 7.3).
Figure 7.9 Map of the Dynamics of the Water Areas of Sarykamysh Lake
|Type of Survey||Date||Water Area km2||Water Volume km3||Water Level m below m.s.l.|
The continuing inflow of mineralized irrigation waters and the continued rise in the water table have resulted in major changes in the area adjoining the lake. First, a narrow band of flat and low-lying land adjacent to the shore is damaged by crusts of salt rapidly forming on intensively waterlogged surfaces. Then water reaching the sloping shores dissolves these salts. This process continues sequentially. The salinized land surfaces adjacent to the shore can be detected on space imagery by their light and mottled image.
The rising water table causes small lakes to form in natural depressions. Further flooding results in the growth of these small lakes, which then merge with one another and, subsequently, with the main reservoir.
Space imagery has enabled the determination of the direction of the lake's growth and to forecast the position of its shoreline. Analysis of the space imagery for 1975 and 1976 clearly revealed areas that showed waterlogging - these areas were then the first to be flooded. And likewise, analysis of space imagery in 1980 and 1985 again confirmed that areas appearing to be waterlogged always appeared to be flooded on later images. Since this flooding is likely to continue, i.e. because there are no plans to halt the irrigation processes, then it is now possible to draw accurate maps which look at the future incremental growth of Sarykamysh Lake. The modelling of this process would clearly be an example of a useful study which could be greatly and efficiently expedited if performed as a GIS exercise.
Since the author claims that the lake “at present functions only as an enormous evaporation pan” and “is used only for catching fish” (p.77), then it is clear that, given the huge size of the lake, RS has been incremental in revealing the growing potential of a fisheries resource.
TITLE: “Mapping Potential Effluent Pathways in the Long Point Region of Lake Erie from Landsat Imagery”
AUTHORS: LeDrew, E.F. and Franklin S.E.
PUBLICATION and DATE: J. of Coastal Research; Vol.3, No.2. 1987.
We have selected this study because it shows the possibility of mapping liquid effluent dispersions in a large lake using RS imagery - clearly the ability to monitor this sort of movement could be of interest to both aquaculture and inland fisheries.
Industry has been attracted to the Long Point Bay area on the north shore of Lake Erie in Canada because of greenfield site availability, inexpensive water transport and nearby markets. Figure 7.10 locates the area and gives an indication of bathemetry. As the industry and its supporting urban structure expands, it can be anticipated that effluent discharges to the lake will increase. The study seeks to find out what the ultimate long distance destinations of the effluents are. There are suggestions that the currents within the Long Point Bay area may form a closed gyre (spiral) so that accumulations of polluted sediments accrue in the eye. However, this is uncertain. What is certain is that the southern shores of the bay are very sensitive ecologically.
In this study Landsat imagery of the Long Point region is examined to identify the geometry of the surface currents for several dates, and for several different wind directions, to determine whether there is a gyre or alternative current configuration and what its spatial characteristics are.
The authors commence this section with a brief look at the problem of which spectral bands give the best indicator of sediment movements in water. There is conflicting evidence on this which results from the large number of variables concerned, e.g. size of sediment particles, amount of sediment, time of image, other pollutants present, depth of water, etc. Band 4 is certainly able to achieve greater depth penetration (up to 20 m) but this is of little value in sediment laden waters. There is a lack of historical records which relate measured sediment densities to particular Landsat images. Consequently, the authors map sediment tendrils as proxy indicators of the current directions on the presumption of mass transport down the density gradient, i.e. even though a single image is a static representation and does not reveal the dynamic nature of a surface.
Figure 7.10 Development Areas and Bathemetry Around Long Point Bay, Lake Erie
Sediment patterns for eight dates were mapped, having selected fiche records of Landsat imagery to identify images with sufficient gradation of tone within Long Point Bay to map sediment tendrils. For each date the prevailing wind pattern in the area was determined from a nearby ground station in order to establish a “regional air flow”, the prevailing directions and an average speed.
Digital image processing techniques were applied, plus some necessary image enhancement, and one image was geometrically corrected to UTM map co-ordinates. Other images were registered to this. The most useful image enhancement technique, for the purpose of delineation of the fine structure of the sediment tendrils, was a histogram stretch of Band 4. In this procedure the original limited range of reflectance values was redistributed over the entire total range by assigning equal memberships to each possible class of reflectance values. Band 7, which has negligible reflectance from water, was used to identify the shoreline. To quantitatively map the sediment contours, a density slice of Band 4 was used. This breaks the reflectance signal into discrete classes that may be colour coded in a map. The density slice was implemented by applying a parallelepiped classification to Bands 4 and 7 so that information in Band 7 could be used to mask the land surface. The sediment contours were mapped from the density sliced digital image, and the arrows depicting the authors interpretation of current direction are based upon evaluation of the contours and the histogram enhancement of Band 4. Black and white renditions of the original colour enhancements and classifications were provided for each case. Figure 7.11 gives an example of one of eight selected images studied.
The authors then describe in some detail their interpretation of four of the eight enhanced Landsat images. We need only summarize the details for Figure 7.11. On July 6, 1974 (incorrectly dated as July 7, 1979 in the journal) there were surface winds, averaging 4.8 km/hr, blowing from the S.W. On the previous day the mean wind was from the N.W., but it was from the S.W. from 1 to 4 July. East of Nanticoke, there was an east component to the littoral drift, while to the west it was westwards. Sediment concentrations were higher in the Nanticoke region than to the west and south of Long Point Bay, which is indicative of higher erosion rates in this region. To the south of Long Point, the littoral transport was strongly to the east until the coast curves north. At this point the current became detached from the shore, meandered southwards and then evolved into a pronounced anti-clockwise gyre. It appears that the current had an inherent instability that became evident in a vortex after it left the confines of the shallows.
There are several pronounced features which can be inferred from this study. The easterly drift along the southern shore of the Long Point Spit is a major feature in most weather patterns. Towards the end of the spit, where it curves northwards, the drift becomes detached before starting the large anti-clockwise gyre to the east and north of the spit. Only at times when the wind has a northerly component is this pattern broken. This easterly drift along the spit is consistent with a “coastal jet” type current. It is strong, able to carry large quantities of sediment, is probably causing erosion, and more importanlty to the aim of the study, is able to efficiently carry pollutants. Any industrial development to the west of the spit would mean that any effluents would be transported eastwards and, having passed the spit, they might curve around northwards and finally settle in the calmer waters of Long Point Bay. The general westwards movement of sediments along the north of Long Point Bay appears to be a component of the gyre within the bay.
It appears that the urban/industrial complex sited on the north shore of the bay (see Figure 7.10) is critically located as far as effluent dispersal is concerned. Had it been located a few miles further west, then most of the effluent would have been dispersed to the Long Point Bay gyre - and probably finish up polluting much of the bay as well as critically affecting the local coastal ecology. As it is, a good proportion of the effluent is likely to get caught in an easterly or off-shore water movement pattern and it is probably distributed into the eastern basin of Lake Erie.
Analysis of local plume dispersion by RS imagery can only form part of a study of the impact of potential pollution. The cumulative effect of several developments and the associated urban infrastructure in the region must also be considered. Additionally, it would be advisable to monitor sediment dispersal patterns over a much wider region. Long-term RS imaging should prove the ideal means of doing this and of eventually showing the final destination of effluent concentrations.
Figure 7.11 Density Sliced Image, Black and White Image and Interpretation of Sediment Plumes in Long Point Bay for July 6, 1974
TITLE: “A Geographical Information System for Catfish Farming Development”
AUTHORS: Kapetsky, J.M., Hill, J.M. and Worthy, L.Dorsey
PUBLICATION and DATE: Aquaculture. Vol.68. 1988. pp311–320.
The authors recognize that there is a need to increase protein supplies world-wide using various “aquafarming” techniques but, until recently, very little attention had been given to evaluating the water, land and human resources necessary, or to look at various environmental and economic constraints to further development. Planning for suitable locations for increasing inland fish production is mostly carried out at a “middle” governmental level, e.g. by districts, county or provinces, though it might be local fisheries officers who would advise on specific sites for production, and who would need information to help with this.
This study is confined to Franklin parish (county) in Louisiana, U.S.A., and the species is Channel Catfish. There were more than 1 000 ha of catfish ponds here in 1986 producing nearly 1 000 tons of intensively farmed fish from 40 different farms. Franklin is the leading parish in Louisiana for catfish production. Catfish farming in the U.S.A. is highly competitive and careful attention to siting can result in significant economic advantages. The primary objective of the study was to show how a GIS could be used to locate and inventory areas suitable for aquaculture at the county level, and to test the results against the actual locations of catfish farms.
The GIS assessment was linked to the Parish, rather than to a geomorphological unit, because most data is reported by administrative unit and because development decisions are likewise implemented.
Since almost all of Franklin parish is flat (elevational range = 11 to 42 meters) then potentially catfish ponds could be developed almost anywhere. The main criteria for development used by the authors was on the physical characteristics of soils, i.e. in terms of their suitability for:
Detailed information was available for all soils in Franklin parish. This categorized them into 7 general, and 28 more detailed, soil map units (Figure 7.12 - 1). It was possible to “rate” all the soil categories, for each of the five land uses given above, so as to arrive at a “pond suitability score” for the whole parish (Figure 7.12 - 2).
A second location criterion used was suceptibility to flooding, and use was made of a map showing all areas with more than a 1% probability of flooding each year (100-year floodplain map) (Figure 7.12 3). A map showing the location of the existing catfish farms was also drawn up.
Figure 7.12 Maps of Franklin Parish, Louisiana, Showing Suitability for Catfish Farms
Fig. 1. The general soil map units of Franklin
Parish: (1) Sharkey-Tensas, (2) Dunde, (3)
Calhoun-Calloway-Loring, (4) Gilbert-Gigger-Egypt,
(5) Necessity-Foley-Deerford, (6)
Forestdale-Sharkey,and (7) Dexter-Liddleville-Necessity.
Turkey Creek Lake is 8.
Fig. 2. Suitabilities of soils for catfish farming development.
Fig. 3. The area of Franklin Parish within the 100-year floodplain.
Fig. 4. Suitabilities of soils for catfish farming which are outside of the 100-year foodplain.
Fig. 5. Suitabilities of soils for catfish farming compared with the locations of existing catfish farms. Proximity to a processing plant is also shown.
|General map unit||Detailed map units (% of area)||Area of parish (%)||Numerical ratings||Final relative rating||Observations|
|Ponds||Levees||Roads||Equipment||Buildings||Overall numerical rating|
|3||Calhoun-Calloway- Loring||33||2.4||1.3||1||1.9||1.3||7.8||Fair||Level to gently undulating|
|4||Gilbert-Gigger- Egypt||29||3||1||1.3||1.8||1.3||8.4||Good||Level to gently undulating|
|5||Necessity-Foley- Deerford||8||3||1.4||1||1.7||1||8.1||Fair||Level to gently undulating|
|6||Forestdale-Sharkey||6||3||1||1||1||1||7||Fair||Fo.higher, less wet|
|7||Dexter-Liddieville- Necessity||3||1.4||1.1||1.2||2.9||2.8||9.4||Poor||Permeable; high relief|
The soil map and the 100-year floodplain map were digitized. This information, along with the necessary attributes and the fish farm location map, were input to a computer at the Louisiana State University Remote Sensing and Image Processing Laboratory - an ELAS GIS software package was used for data processing.
Table 7.4 gives the numerical ratings and final relative ratings for the 7 general soil units, and on Figure 7.12 - (4) the areas are ranked in order of suitability. It can be seen that the final relative ratings do not necessarily concur with the total soil suitability ratings because an assessment of other criteria has been introduced. For example, “general map unit” 7 (Dexter-Liddleville-Necessity) was rated first for soils, but its overall relief was disadvantageous to pond construction resulting in a final “poor” rating. Map unit 4 was considered to be the overall best area for pond construction on account of its suitable soils, its relatively level land, very little of which was flood prone. It is interesting to note from Figure 7.12 - (5) that about half of the existing catfish farms are located in areas indicated by the GIS as being most suitable. Only two of the 40 farms are located in the least suitable areas.
Though the results show that a GIS can be used to locate, map and inventory areas which are suitable for catfish pond development, there are many ways in which it could be improved. Other location criteria would certainly be worthy of addition. These include:
The study shows that an effective GIS for aquaculture planning need not be complex or expensive. Once the data bases have been captured, updating is easy and the data can be manipulated to achieve various simulations, e.g. what effect would raising levee bank heights have on diminution of flood prone areas?
TITLE: “Where Should Trout Farms be in Britain?”
AUTHOR: Meaden, G.J.
PUBLICATION and DATE: Fish Farmer; Vol.10,No.2. 1987.
Trout farming in Britain has undergone considerable expansion over the past decade. This has been in response to rising demands, to the greatly increased cost of wild-catch fish, to a diffusion of knowledge about trout farming potential and to the needs of farmers to diversify in diffusion of knowledge about trout farming potential and to the needs of farmers to diversify in times of food stockpiles and insecure farm incomes. Where should prospective trout farmers be looking to find commercially viable situations for farming in England and Wales? The basic objective of this study was to outline a methodology which could be used to answer this question. It was an exercise in applied spatial analysis, i.e. the author was concerned that so many business enterprises had been unsuccessful simply because they were poorly located, and the importance of the location decision to business success has been grossly underestimated.
To determine location suitability it was first necessary to establish those factors of production (production functions) which controlled trout farming, i.e. mostly physical and economic factors. From a large number of factors identified, those which showed spatial variability were then isolated (see Table 7.5).
|iv)||Availability of underground water.|
|vii)||Nearness to the source of the waterway.|
|x)||Access to road transport.|
|xi)||Distance from catering markets.|
|xii)||Distance from wholesale markets.|
|xiii)||Distance from restocking markets.|
|xiv)||Distance from competing trout farmers.|
The methods used to map and compare the suitability of different areas for trout farming involved eight stages:
For mapping purposes the area of England and Wales was divided into a grid of 10 km2 cells, i.e. a total of 1 706 cells.
A map was produced for each of the production functions listed in Table 7.5, showing its distribution throughout England and Wales.
A “score”, ranging from 0 to 10, was allocated to each cell according to its ability to provide for each production function. Those which had no monetary value (the physical functions) could then be “costed” on a similar scale to economic inputs.
All trout farmers in England and Wales were sent a questionnaire which asked them to “weight”, on a scale of 0 to 6, the relative importance of each function in helping them to achieve commercial viability.
A “mean weighting” could then be established for each production function, showing its relative importance to trout farming.
The “mean weighting” given to a function was multiplied by the “score” allocated to that function (in iii above) for each cell to achieve a “weighted production function score”.
The 14 “weighted production function scores” were aggregated to produce an “aggregated weighted score” for each cell. These scores could be ranked to show the order of suitability of cells for trout farming potential.
The “aggregated weighted scores” formed the basis of a final map (Figure 7.13) showing areas of relative suitability for trout farms by 10 km2 cells.
Figure 7.13 Areas of Suitability for Trout Farms, by 10 km2 cells
The simple mathematical functions (multiplication and aggregating) were performed on a VAX minicomputer and mapping was done using a GIMMS cartographic software package, though this manipulation and plotting could now be performed using less sophisticated devices. The author then lists a number of simplifying assumptions which are inevitable in any process modelling the real world.
Figure 7.13 clearly highlights the suitability of central southern England for trout farming, with its favoured areas of chalk or limestone hills providing high quality water, of a uniform temperature and having a consistent flow rate, i.e. little variability. These areas also have good access to road transport, high agglomeration potential, reasonable land costs and they are not too far from the various market outlets. Other major areas showing potential are the Yorkshire, Norfolk and Lincolnshire Wolds and an area in the West Midlands.
It is of interest to note that some of the areas which had traditionally been seen as being good trout farming areas, in fact scored very poorly. These included upland areas such as much of Wales, the Lake District and the Pennines. In these areas water temperatures and flow rate variability are poor, climatic conditions and relief are adverse, though the major problem is the distance from most markets. The author also explains why other areas are unsuitable.
It is worth testing the methodology by seeing if existing trout farms are located in areas which were shown to be suitable. An analysis revealed that 32% of all farms were located in cells which were ranked as being in the top 10% in their suitability for trout farming. From the data gathered, it was also possible to work out those parts of England and Wales which were either over- or under-represented as far as the number of trout farms was concerned - information which could be very valuable to prospective fish farmers.
The analysis was very detailed and data capture was a lengthy exercise which relied on the availability of a large amount of secondary data. Once captured, however, the necessary updating could be simply carried out and the data bases could be modified as either more refined data were found or if the exercise was to be carried out at a different scale. The study also showed that different results could be produced if, at the “weighting” stage (described in (v) above), only questionnaire responses from selected groups of trout farmers were selected, e.g. farmers who were producing for the restocking market or farmers who were utilizing a particular type of production system (perhaps using polythene tanks). This would produce different “mean weightings” as different groups of farmers might see different production functions as being important to their type, or sector, of trout farming. Clearly different final maps would then result. Indeed, it would not be necessary to use groups of fish farmers at all, i.e. any potential or actual trout farmer could use weightings according to his perceptions of their relative importance.