© 2006 International Council for the Exploration of the Sea
Comparison of airborne lidar with echosounders: a case study in the coastal Atlantic waters of southern Europe
a Museo do Mar de Galicia Avenida Atlántida 160, 36208 Vigo, Spain
b NOAA Earth System Research Laboratory 325 Broadway, Boulder, CO 80305, USA
c Fundación AZTI-AZTI Fundazioa Herrera Kaia, Portualdea, Z/G 20110 Pasaia, Spain
d IPIMAR, Avda de Brasilia 1400 Lisboa, Portugal
e IFREMER, Laboratoire de Sète 1 Rue Jean Vilar, 34203 Sète Cedex, France
*Correspondence to P. Carrera: tel: +34 986247750; fax: +34 98624748. e-mail: pablo.carrera{at}museodomar.com.
The feasibility of using airborne lidar (Light Detection and Ranging) was studied to assess the early juvenile fractions of the main pelagic fish species of the coastal Atlantic waters of southern Europe (anchovy, sardine, mackerel, and horse mackerel). Field comparisons with more established echosounder methods were undertaken in the summers of 1998 and 1999 during the recruitment period of sardine and anchovy in the selected areas, in the presence of a variety of oceanographic and environmental conditions. Backscattered energies as well as the types of target recorded by both devices were compared. The distributions of energies and the shape of the targets were generally similar for both techniques, with moderate numerical correlation between sensors, demonstrating the potential of lidar for assessment of anchovy, sardine, and juvenile mackerel. However, differences in received backscattering energy were found, especially in the presence of certain plankton assemblages (to which lidar is more sensitive) and isolated schools with large vertical dimensions (for which shadowing is more significant for light than sound). Experimental ad hoc optical reflectivity measurements of fish and plankton are proposed to discriminate these two types of targets. In addition, an improvement on lidar implementation and data processing is suggested to achieve fish abundance estimates.
Keywords: airborne lidar, anchovy, Bay of Biscay, echosounder, recruitment, sardine, West Iberian Peninsula
Received 31 March 2006; accepted 19 July 2006.
 |
Introduction |
|---|
 Top Introduction Material and methodsResultsDiscussionConclusionsReferences |
|---|
One of the most important components of fishery management is the assessment of the strength of the year classes before they enter a fishery. Good survival of larvae produces strong year classes (Smith, 1985; Houde, 1996). It seems, therefore, that the strength of the recruitment of many species is already established five or six months after spawning, once the fish are already juveniles. This has been demonstrated for many populations all over the world, including California rockfish (Sebastes spp.; Ralston and Howard, 1995), walleye pollock (Theragra chalcogramma; Bailey and Spring, 1992), and several pelagic populations such as herring (Clupea harengus; Leblanc et al., 1998) and northern anchovy (Engraulis mordax; Smith, 1985). Hence, when early juveniles can be assessed, the estimates can be used to predict the relative strength of the future recruitment to the fisheries. This strategy is of special interest to manage the fisheries for short-lived species because of the short time between spawning and the exploitation of subsequent emerging recruits, as in the South African anchovy fishery (Engraulis encrasicholus, formerly E. capensis; Hampton, 1992; Butterworth and Bergh, 1993) and Icelandic capelin (Mallotus villosus; ICES, 2000a).
Echosounder methods have traditionally been applied to obtain recruitment estimates (Dragesund and Olsen, 1965). However, the assessment of early juveniles may be difficult, because many juveniles display epipelagic phases, when they remain in the upper layers of the water column. Also, they are often found in coastal areas or, in some cases or phases, even in shallow water (Mays, 1974; Alshuth, 1988; Dias et al., 1988, 1989; Lockwood, 1988; Boyd et al., 1997; Villamor et al., 1997; Leblanc et al., 1998). Such behaviour would set juveniles outside the vessel range and/or the effective observational range of an echosounder transducer (MacLennan and Simmonds, 1992). Besides, echosounders may underestimate schools if they are actively avoiding the ship (Fréon and Misund, 1999). All these potential problems suggest a need for improved echosounder methods and/or complementary techniques for detecting juvenile fish.
In the Atlantic waters of the Iberian Peninsula and in the Bay of Biscay, sardine (Sardina pilchardus) and anchovy (Engraulis encrasicholus) support important fisheries of Portugal, Spain, and France (Uriarte et al., 1996; ICES, 2000a, b; Carrera and Porteiro, 2003). Off the Iberian Peninsula, the recruitment of sardine at age 0 is mainly in the northern and central parts of Portugal between summer and winter (ICES, 1982; Porteiro et al., 1986, 1993; Dias et al., 1988, 1989, 1993; Pestana, 1989). In the Bay of Biscay, the major nursery areas for anchovy are located in the southern part (Prouzet et al., 1994; Uriarte et al., 1996).
Direct estimation of the recruitment at age 1 for both species by echo integration is routinely done in spring at the time of spawning (Massé et al., 1992; Porteiro et al., 1993, 1996; Scalabrin and Massé, 1993; Massé, 1996). Although an echosounder survey is conducted in November off Portugal to estimate the recruitment of sardine aged 0 (Dias et al., 1996, Porteiro et al., 1996), a systematic study of juvenile age-0 anchovy and sardine off southern Europe had never been performed. Among other reasons, juvenile shoals of both fish species have an epipelagic distribution and, in the case of sardine, a shallow water one too (Cort et al., 1976; Martín, 1989; Soares, 1995). This could make it difficult to estimate fish biomass with echosounders with any accuracy.
In the 1990s, research began into the development of airborne lidar (Light Detection and Ranging) surveys for detecting fish schools (Hunter and Churnside, 1995; Gauldie et al., 1996), based on earlier feasibility studies (Fredriksson et al., 1978; Squire and Krumboltz, 1981). Over its operational range (about 30–40-m depth) lidar was tested as a technique to detect tuna (Thunnus spp.) in the Pacific (Oliver et al., 1994), and pelagic species such as sardine (Churnside et al., 1997), capelin (Brown et al., 2002), mullet (Mugilidae; Churnside et al., 2003), and even zooplankton (Churnside and Thorne, 2005). According to these experiments, lidar was a potentially useful tool to map the distribution of schools of juveniles close to the surface. As it can be operated from a small aircraft, it is an accessible tool to survey shallow coastal waters. Processing the lidar signal to obtain quantities proportional to the number of fish within the operational depth range has been demonstrated, although conversion into biomass requires species identification and experimental knowledge of fish reflectivity and size (Churnside and McGillivray, 1991; Krekova et al., 1994; Churnside and Hunter, 1996; Churnside et al., 1997).
In this context, to improve estimation of recruitment at age 0 of the pelagic resources in the Atlantic waters of the Iberian Peninsula and the Bay of Biscay, a research project named "Experimental Surveys for the Assessment of Juveniles" (JUVESU-FAIR CT97-3374) was developed. The major goal of the project was to evaluate airborne lidar systems for surveying the distribution and relative abundance of early juvenile sardine and anchovy, and comparing the results with those from the traditional shipborne echosounder systems. The experimental surveys were made in 1998 and 1999 around the western Iberian Peninsula, and in the southern part of the Bay of Biscay (Figure 1). These areas represent remarkable oceanographic differences and contain nurseries of several pelagic species of major interest for various European fisheries (mainly sardine and anchovy, and secondarily mackerel, Scomber scombrus, and horse mackerel, Trachurus trachurus).
|
This manuscript describes the results of the analysis of the relative performance of the lidar and echosounder systems for detecting juveniles of the main pelagic species arising from these experimental surveys. In addition, direct analysis between sensors was made in discrete areas with particularly homogenous distribution of juvenile school type. Finally, we discuss the potential of lidar technology for future assessment of pelagic fish juveniles in the study areas.
|
Material and methods |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionConclusionsReferences |
|---|
Two experimental surveys were performed, in August/September of 1998 and 1999, to test the performance of lidar and echosounder systems in detecting fish, primarily juveniles. These study areas, shown in Figure 1, were the southeastern Bay of Biscay (43°N–46°N, 1°W–5°W), the Galician Rías and the shelf off the northwest Iberian Peninsula (41.8°N–43°N, 8.7°W–9.4°W), and the northern central shelf off Portugal (39.5°N–41.8°N, 8.6°W–10.2°W). These are areas where juvenile anchovy, sardine, mackerel, and horse mackerel historically occur.
Fish distribution mapping and species identification were carried out by shipborne echosounder and fishing surveys, coupled with airborne lidar surveys. Echosounder surveys of each area were accomplished with various research vessels. Fishing to provide ground truth (McClatchie et al., 2000) was performed by the same vessels (pelagic and bottom trawl), and by chartered commercial vessels (purse-seine). Table 1 presents sampling details by year and region.
|
Surveys were conducted in two phases: (i) an extensive coverage of the target area; and (ii) an intensive coverage of a selected portion of the area. The ship and aircraft surveyed common transect lines at speeds between 7 and 10 knots and
140 knots, respectively. An Elementary Distance Sampling Unit (EDSU) was set at 1 nautical mile (hereafter referred to as mile). The extensive area was covered only during daylight. The intensive legs were located in those areas of high juvenile abundance, where repeated passes were made at different periods of the day. The main goal of this second phase was to characterize pelagic fish aggregations in space and time. For the analysis presented here, both the extensive data and the day and evening intensive data were used. The extensive data were used to compare backscatter intensities, and both extensive and intensive data were used to perform echo trace analyses. The lidar system is the NOAA radiometric lidar described by Churnside et al. (2001). With a mean altitude of 300 m, the lidar beam diameter was 5 m at the sea surface. Depth penetration varied by area and year, depending on water clarity, light level, and laser power, but was 25–30 m. In 1998, the lidar was installed in a twin-engine Casa aircraft operated by the Instituto Nacional de Tecnica Aeroespacial (INTA). In 1999, the Spanish Air Force supplied a similar aircraft. However, the electrical power system had to be augmented with dry batteries to accommodate the lidar laser power, so lidar performance was reduced to extend survey time (i.e. a lesser repetition rate).
Data analysis
The primary lidar data-processing method, henceforth referred to as school processing, involves a threshold that removed the noise from each shot. The threshold was approximately the median backscatter intensities from all shots over 1 mile. School processing was applied to the entire extensive coverage. Because school processing rejected extensive juvenile layers, another method was developed, denoted "echo-integration" for its similarity to the echosounder processing method (see Figure 2 for an illustration of each). Similarly, echosounder data with a –60-dB threshold were integrated over 5-m depth layers from the transducer depth (3–5-m depth) to the bottom. Several school size and descriptors were extracted from the echosounder and lidar data but biomass estimations were not attempted.
|
For comparison, it was assumed that the sound echo-integration method provided an accurate fish echo trace characterization and biomass estimate. Bias attributable to fish avoidance was assumed to be negligible (Fernandes et al., 2000a, b), but fish responses were expected, specially diving reactions (Blaxter et al., 1981; Schwarz and Greer, 1984).
We first compared the extensive coverage of the three areas surveyed, assuming a stationary distribution and either single species or uniform mix of species. The mean backscattered energies over the surveyed area from the echosounders and lidar were compared with the expectation that they should be similar. Echosounder data to depths up to 30 m were used to correspond to mean lidar depth penetration. Each data set was log-transformed, then put in relative units by normalizing each observation by its maximum. For each data set, the spatial distribution was studied by means of geostatistic tools (Matheron, 1971; Petitgas, 1991, 1993). The analysis was Estimation Variance Analysis (EVA; Petitgas and Prampart, 1993) using a SURFER v7.0 (Golden Software) package. Kriged contour maps were constructed and compared using variogram models with the same map grid for both echosounder and lidar data. Wilcoxon paired-sample tests were performed on results obtained after the kriging process.
A second comparison was made over segments that satisfied the following constraint criteria: (i) similarity on echo traces recorded by both devices; and (ii) elapsed time between ship and aircraft coverage less than two days. These segments were called Homogeneous Areas for Geographic Comparison (HAGC). Figure 1 shows the location of selected HAGCs. Both lidar-processing methods were compared with the echosounder results for the HAGCs. The quantitative and ordinal (Pearson and Spearman) correlations between the average echo-integrated sound and light energies across HAGCs were calculated. Lidar and echo-integration energies were normalized by the average value for each area and year, and were log-transformed, i.e. for every HAGC i, the energy Ei was substituted by the result of the following expression:
|
|
E
area,year corresponds to the average value of the energy from a given sensor obtained for the corresponding year and area (Bay of Biscay, Galician, or Portuguese surveys). Correlations were made for these transformed energies. Also, acoustic echo traces were classified according to their shape in four categories (schools, pelagic layers, bottom layers, and scattered echo traces). |
Results |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionConclusionsReferences |
|---|
In Portugal in 1998, most fish schools were near the shore. The main species were juvenile and adult sardine, but there was no clear horizontal separation of the two life stages. A greater proportion of juveniles was found close to the shore from Aveiro to Figueira da Foz, so this area was selected for the intensive survey. Both the echosounder and lidar data showed the same general patterns in the coastal area (see sample echograms in Figure 3). Exceptions included: (i) plankton layers detected by lidar but not by echosounder using the –60-dB threshold; (ii) consistently greater light backscatter offshore; and (iii) some large schools detected by echosounders, but not by lidar.
|
In Galicia in 1998, most fish were near the coast, especially within or at the mouth of the Rías, as observed by both sensors (Figure 4). Juvenile sardine were the most abundant fish species, but in low abundance from a historical perspective. In 1999, sardine abundance was even lower and more restricted to coastal waters. Outside the bays, the main species (although scarce) was mackerel. No fish were found in surface waters near the 200-m isobath. That year, relative to the echosounder, the lidar detected more scatterers in the outer part, and few dense schools in the inner part of the surveyed area.
|
In the Bay of Biscay in 1999, almost all fish caught were juvenile anchovy, found off the continental shelf. The comparison of the maps produced by each sensor did not show clear patterns such as those observed in the western Iberian Atlantic, except for some differences near the beginning or end of the tracks (Figure 5). Lidar estimates were slightly higher in the peripheral areas, and echosounder estimates were slightly bigger in the central part of the surveyed area. In the Garonne area, for the single track analysed, juveniles in weak densities occupied the midwestern part of the shelf, whereas in the coastal zone, larger concentrations of adults of different species were found. Along the track, there were two gaps in the acoustic echosounder data that were not seen in the lidar data.
|
Extensive area comparison
Summary statistics for each data set (area, year) are shown in Table 2. Neither fish nor thick plankton layers were observed in the uppermost layers in some parts of the western areas, but no empty EDSU were seen in the Bay of Biscay. Raw data were in general skewed, with a few high values dominating both the arithmetic mean and the standard deviation. The weighted and log-transformed data were more comparable; the maxima and minima for each device were similar for the same area and year.
|
There was spatial autocorrelation for each device, area, and year, as shown in Table 3. The range of this correlation was around 7 miles for the echosounder data. In comparison, the ranges of the lidar were 3 miles in 1998 and 6.5–11 miles in 1999. Empirical variograms were fitted according to the models shown in Table 3. These models were used to construct kriging surfaces over the same grid (Figures 6 and 7).
|
|
|
Wilcoxon signed-rank tests performed on the kriged values showed significant differences. In Galician and Portuguese waters, lidar values were higher than those recorded by the echosounders (Z = 10.7, p = 0; Z = 35.6, p = 0; Z = 15.9, p = 0; for data from Galicia in 1998, Portugal in 1998, and Galicia in 1999, respectively). The situation was the same in the inner part of the Bay of Biscay (Z = 12.5, p = 0), suggesting that lidar detected more targets than echosounders. Conversely, in the Garonne area, echosounder records were bigger than those of the lidar (Z = 2.40, p = 0.0081).
Direct HAGC comparison
Data from both devices were directly compared over 22 HAGCs across years and areas (see Table 4 for the main characteristics of each HAGC). A comparison of normalized and transformed energy values produced moderate levels of positive correlation between sensors, which were significant for the case of school processing. The lidar echo-integration method improves slightly the comparison of echosounder and lidar for HAGCs of scattered fish, but generally worsens the comparison for other echo types. This is particularly noticeable for the few HAGCs with large-school detections, which can even result in negative correlations. Combining all types together except large schools gave significant (at 
= 5%) or almost significant correlations (with p < 0.1) with both lidar signal-processing methods. Relative to 1999, the 1998 HAGCs showed better agreement for the two sensors. Among geographic zones, Galician waters (and all western areas together) obtained the best fit between sensors, while for the Bay of Biscay, correlations were not significant.
|
Owing to changes in fish aggregation patterns, some HAGCs taken on different days gave lower correlation indices, as observed in HAGC 2 (surveyed by echosounder on 5 September, with a value of 3.23, and by lidar on 7 September, with a value of 0.44) and HAGC 8 (surveyed on 8 September, with a value of 0.66, and on 9 September, with a value of 2.86). On the other hand, HAGC 17 gave large differences in values for both lidar-processing methods, probably related to the quality of the lidar data.
Removing these three points from the data set improved the correlation to 0.65 (in linear scale) or to 0.55 (in log scale; Table 5, Figure 8). Removal of any other additional points barely increased the correlations. With this data set, the correlation for large-school echosounder detections (though based now on just three points) became positive in the school processing for lidar (not negative as before). In general, all the analysis of the school processing for lidar was improved: the 1998 set of HAGCs became statistically significant, and the diurnal correlation among HAGC values was also improved.
|
|
The last column in Table 4 summarizes the fraction of fish stocks in the upper 30 m of the water column for each region. In Portugal, the situation was ideal for lidar; the HAGCs corresponded to the shallow waters of the radials of the intensive campaign, and almost all schools were in the upper 30 m. Conditions in the Bay of Biscay were quite good also, with high percentages of schools inside the nominal lidar range, although the actual lidar penetration depth turned out to be <30 m in most areas. The worst cases were in Galicia, especially in 1999. Those HAGCs were mostly in inshore waters with fish very close to the bottom, which made separation of the fish and bottom signals difficult (Figure 8).
|
Discussion |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionConclusionsReferences |
|---|
The stationary comparisons based on the smoothed contour maps of backscattering transformed energies have shown that there is a general consistency between both sensors, so showing the ability of lidar to detect and map the distribution of the targeted fish species. However, lidar (a noisier sensor according to the range and sill values obtained in the variograms presented in Table 3) produced more detections than echosounders (Figures 6 and 7), and the differences were statistically significant. This may be because the relationship between target strengths (backscattered energies) among species (fish and plankton) for each sensor was different, which probably affected the threshold applied to each device. According to this, retained valid echo traces (i.e. energy patches that were above the threshold) would be different and, therefore, so would the subsequent integrated backscattered energies.
In Galician waters in 1999, the offshore detections made by lidar that were not detected by echosounders (Figure 2) may have been layers of plankton together with Polybius spp. Both depth and density of these layers varied along the transect, giving a patchy distribution which would be retained as a fish school by lidar. Other thick plankton layers, seen in some parts of the Bay of Biscay in 1999, also scatter light more effectively than sound. Similarly, fish without swimbladders do not scatter sound as effectively as fish with swimbladders, whereas differences in optical scattering between fish species do not depend on whether or not the fish has a swimbladder. This could explain the detections made in southern Galicia (off the Ría de Vigo) by lidar in 1999, where the echosounder detected very little. In that area, concentrations of juvenile mackerel, which do not have a swimbladder, were detected by both echosounder and fishing, but they produced very low echo-integration values, compared, for example, with sardine. Lidar, on the other hand, would produce similar backscattering energies for the same two fish species.
Lidar has a smaller target-strength difference than sound between plankton and fish, and between fish with and without a swimbladder. This leads to noisier data, which make it more difficult to set a signal-threshold level to discriminate plankton from fish, than echosounders, and further complicate attempts to obtain a conversion from energy to fish biomass. As a consequence, it is clear that complementary fishing surveys for species identification of lidar detections are necessary. In addition, it is necessary to study the light reflectivity properties of fish and plankton targets, including polarization effects and the effects of school patchiness in the global signal return if we are interested in making lidar biomass estimations. Nevertheless, it is always advisable to select survey periods in which the fish schools are not mixed with plankton layers or other non-target scatterers.
The varying performance of lidar under different conditions of fish aggregation structure, plankton-layer intensity, and water turbidity (caused by suspended matter, or zoo- or phytoplankton) have made it difficult to establish a single satisfactory processing method for the different times of day, area, and species. Two types of lidar signal processing have been assessed during the current study; both gave interesting results under some conditions, but further refinement is required. The reflection that lidar obtains from a school of fish must be processed to obtain quantities proportional to the number of fish within the depth resolution. This issue has been considered before (Krekova et al., 1994, Churnside and Hunter, 1996, Churnside et al., 1997), although conversion into biomass requires species identification and experimental knowledge of fish reflectivity and size.
Direct HAGC analyses were performed to understand better the energy and visual-shape properties of particular target types. If we check the correlations annually, the results show better relationships for the set of HAGCs analysed in 1998, likely because there were different types of aggregations and lidar performance was better in 1998. The latter was caused by the need to operate at a lesser repetition rate in 1999. By types, the analysis was limited by the number of observations, which made it more difficult to obtain significant quantitative conclusions. However, the detections made by both sensors of small schools, scattered fish, and pelagic layers were significantly correlated. It is not clear why echo-integration processing of lidar signals worked well for scattered fish, but was not as good as school processing for diffuse layers of fish and plankton. Lidar school processing seeks schooling structures and should provide good performance for schooling types, especially large ones, which are easier to detect. Contrary to this expectation, the correlation apparently failed for these large-school structures. This failure may be explained either by a greater shadowing effect on lidar backscattered energy, or by changes in the aggregation pattern or even shoal movements. Nevertheless it has been demonstrated that a consistency in school parameters exists across years and areas (Petitgas et al., 2001; Muiño et al., 2003), which permits this type of comparison. The lower correlation could be due to relative differences in sound target strength between fish and plankton, which are not the same as the differences in optical reflectivity. Hence, experimental research on the polarization-dependent reflective properties of fish and plankton targets should be performed as a basic step in implementing lidar as a routine survey system for fish resources.
Coastal and bottom topography and oceanography are very different throughout the region addressed here, which can affect lidar performance and operation. Along the Iberian Atlantic coasts, upwelling events are common in summer, and they contribute to the presence of fog, which diminishes the lidar capability. The Galician region has a narrow continental shelf, with depths <50 m close to the coast, a rough substratum, and narrow fjord-like Rías that occupy a large portion of the coast. The waters inside the Rías, such as those very close to shore in Galicia and Portuguese waters, are very productive and turbid, which will often limit lidar water penetration to the upper 15 m. Low-altitude aircraft operations within the Rías can be limited by safety considerations, especially at night. On the other hand, the Bay of Biscay has a narrow (south, Spain) or wide (north, France) continental shelf, with a stable water column. There, waters are neither as productive nor as turbid, so lidar penetration is optimal, reaching about 25 m (on average) during daylight in offshore waters inhabited by juvenile pelagic fish, and even deeper at night (about 35 m).
Applicability of lidar to survey juveniles of pelagic species in the area
Juvenile anchovy, sardine, and mackerel exhibit several common aggregation patterns and behaviour characteristics that make them potentially suitable for applying lidar technology. All are found in the upper layers of the water column, and, to a great extent, within the range of lidar. Also, they concentrate in schools during daylight and disperse at night in pelagic layers, barely distinguishable from plankton, so making daylight the appropriate diel period for surveying.
There are also some common drawbacks in directly applying lidar to juvenile sardine and mackerel that need to be taken into account. Juveniles of both species share coastal Iberian Atlantic waters, mixed in different proportions with respect to each other, as well as with respect to other species (either adults or juveniles). This requires that a lidar survey be combined with an echosounder and fishing survey to assure the identification of the different species in the area. Also, the area of this study has several environmental features (described in the previous subsection) that tend to reduce lidar capability compared with oceanic waters. Finally, lidar detected schools in the offshore part of the area, where echosounders did not. This is likely because patches or layers of plankton are retained as a valid fish echo trace.
Juvenile anchovy are widely distributed throughout the upper layers of the Bay of Biscay. For that species, the medium-sized and largest schools of anchovy were detected by lidar, but small schools were hard to discriminate from the normal empty signal return, or from plankton layers. Therefore, although lidar surveys during daylight should, in principle, be able to detect these juveniles (particularly given the coverage), the relatively poor correlation in the HAGC analysis for the Bay of Biscay indicates a need for further study to discriminate reliably between juvenile fish schools and aggregations of plankton.
Therefore, for these potentially suitable species, lidar surveys could supplement a minimal echosounder cruise that would provide identification and biological sampling and estimation of the biomass of schools. From this perspective, it would seem that lidar surveys could improve the accuracy of an echosounder survey by adapting the number and the length of surveyed tracks to the total distribution of these juveniles shown by the lidar survey. Such a survey design would speed up the survey and reduce the total cost.
|
Conclusions |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionConclusionsReferences |
|---|
The positive correlation between lidar and echosounder backscatter, along with the visual correspondence observed in the kriged maps, suggest that lidar is able to detect and map aggregations of juvenile anchovy, sardine, and mackerel. However, fish and plankton are difficult to discriminate in lidar data, and there are virtually no data available on optical target strengths for the different fish and plankton species present in the surveyed areas. Additionally, both the school and echo-integration processing algorithms failed to detect all possible aggregations of juvenile fish. Therefore, for airborne lidar to provide quantitative measures of fish distribution and abundance, significant progress must be made in the following areas: (i) modelling and measuring optical target strengths by size for the dominant fish and plankton species; (ii) improving species discrimination algorithms in lidar signal processing; and (iii) combining airborne lidar, shipborne echosounder, and direct sampling into a survey design to exploit the advantages of each platform and technique.
|
Acknowledgements |
|---|
We thank our colleagues and the crew members who constituted the teams contributing to this multidisciplinary fieldwork. We are also indebted to the anonymous referees who greatly contributed to the improvement of the draft manuscript. The research was supported by the project JUVESU (FAIR CT97-3374) of the EU.
|
References |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionConclusionsReferences |
|---|
-
Alshuth S. (1988) Seasonal variations in length-frequency and birthdate distribution of juvenile sprat (Sprattus sprattus). ICES Document CM 1988/H: 44.
Bailey K.M. and Spring S.M. (1992) Comparison of larval, age-0 juvenile and age-2 recruit abundance indices of walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska. ICES Journal of Marine Science 49:297–304.
Blaxter J.H.S., Gray J.A.B., Denton E.J. (1981) Sound and startle responses in herring shoals. Journal of the Marine Biological Association of the UK 61:851–869.
Boyd A.J., Shannon L.V., Schülein F.H., Taunton-Clark J. (1997) Food, transport and anchovy recruitment in the southern Benguela upwelling system off South Africa. In Durand M.H., Cury P., Mendelssohn R., Roy C., Bakun A., Pauly D. (Eds.). Global vs. Local Changes in Upwelling SystemsParis edition de l'Orstom pp. 195–210.
Brown E.D., Churnside J.H., Collins R.L., Veenstra T., Wilson J.J., Abnett K. (2002) Remote sensing of capelin and other biological features in the north Pacific using lidar and video technology. ICES Journal of Marine Science 59:1120–1130.
Butterworth D.S. and Bergh M.O. (1993) The development of a management procedure for the South African anchovy resource. In Smith S.J., Hunt J.J., Rivard D. (Eds.). Risk Evaluation and Biological Reference Points for Fisheries Management. Canadian Special Publication in Fisheries and Aquatic Sciences 120: pp. 83–99.
Carrera P. and Porteiro C. (2003) Stock dynamics of the Iberian sardine (Sardina pilchardus, W.) and its implication on the fishery off Galicia. Scientia Marina 67:Suppl. 1, 245–258.[Web of Science]
Churnside J.H., Demer D.A., Mahmoudi B. (2003) A comparison of lidar and echo sounder measurements of fish schools in the Gulf of Mexico. ICES Journal of Marine Science 60:147–154.
Churnside J.H. and Hunter J.R. (1996) Laser Remote Sensing of epipelagic fishes. In Feigels V.I. and Kpilevich Y.I. (Eds.). Laser Remote Sensing of Natural Waters: from Theory to Practice. Proceedings SPIE 2964: pp. 3–53.
Churnside J.H. and McGillivray P.A. (1991) Optical properties of several Pacific fishes. Applied Optics 30:2925–2927.[Web of Science]
Churnside J.H., Tatarskii V.V., Wilson J.J. (1997) LIDAR profiles of fish schools. Applied Optics 36:6011–6020.[Web of Science]
Churnside J.H. and Thorne R.E. (2005) Comparison of airborne lidar measurements with 420 kHz echo-sounder measurements of zooplankton. Applied Optics 44:5504–5511.[CrossRef][Web of Science][Medline]
Churnside J.H., Wilson J.J., Tatarskii V.V. (2001) Airborne lidar for fisheries applications. Optical Engineering 40:406–414.[CrossRef][Web of Science]
Cort J.L., Cendrero O., Iribar X. (1976) La anchoa, Engraulis encrasicholus (L.), del Cantábrico. Boletín del Instituto Español de Oceanografía 220:3–34.
Dias C. A., Marques V., Soares E. (1993) Spanish–Portuguese sardine acoustic survey off Iberian Atlantic coast (ICES Sub-areas VIII and IX) – "Ibersar–92", November 1992. Working Document of the ICES Working Group on the Assessment of Mackerel, Horse Mackerel, Sardine and Anchovy.
Dias C. A., Pestana G., Soares E., Marques V. (1996) Present state of sardine stock in ICES Divisions VIIc and IXa. Working Document of the ICES Working Group on the Assessment of Mackerel, Horse Mackerel, Sardine and Anchovy.
Dias C. A., Soares E., Marques V. (1988) Distribuição e estimação acústica da abundância de sardinha na costa Portuguesa em Março de 1988. Resultados do cruzeiro "SAR88MAR" do projecto "SARAS". Anexo II do relatório de Progresso do Projecto JNICT 87344/Mar, Novembro 1988. INIP. 40 pp.
Dias C. A., Soares E., Marques V. (1989) Acoustic abundance estimation of sardine (Sardina pilchardus, Walb.) of the Portuguese coast, July–August 1988. ICES Document CM 1989/H: 52. 24 pp.
Dragesund O. and Olsen S. (1965) On the possibility of estimating year-class strength by measuring echo-abundance of 0-group fish. Fiskeridirektoratet Skrifter Serie Havundersoekelser 13:47–75.
Fernandes P.G., Brierley A., Simmonds E.J., Millard N.W., McPhail S.D., Armstrong F., Stevenson P., Squires M. (2000) Fish do not avoid survey vessels. Nature 404:35–36.[CrossRef][Medline]
Fernandes P.G., Brierley A., Simmonds E.J., Millard N.W., McPhail S.D., Armstrong F., Stevenson P., Squires M. (2000) Addendum to fish do not avoid survey vessels. Nature 407:152.[CrossRef]
Fredriksson K., Galle B., Nyström K., Svanberg S., Öström B. (1978) Underwater laser-radar experiments for bathymetry and fish-school detection. GIPR-162 Göteborg Institute of Physics, Göteborg, Sweden.
Fréon P. and Misund O.A. (1999) Schooling behaviour. In Fréon P. and Misund O.A. (Eds.). Dynamics of Pelagic Fish Distribution and Behaviour: Effects on Fisheries and Stock Assessment(Fishing News Books, Oxford) pp. 56–100.
Gauldie R.W., Sharma S.K., Helsley C.E. (1996) LIDAR applications to fisheries monitoring problems. Canadian Journal of Fisheries and Aquatic Sciences 53:1459–1468.
Hampton I. (1992) The role of acoustic surveys in the assessment of pelagic fish resources on the South African continental shelf. South African Journal of Marine Science 12:1031–1050.
Houde E.D. (1996) Evaluating stage-specific survival during the early life of fish. In Watanabe Y., Yamashita Y., Oozeki Y. (Eds.). Survival Strategies in Early Life Stages of Marine Resources(Balkema, Rotterdam) pp. 51–66.
In Hunter J. R. and Churnside J. H. (Eds.). Airborne Fishery Assessment Technology – a NOAA workshop report. (1995) South West Fisheries Sceince Center Administrative Report, La Jolla, LJ-95-02. 71 pp.
ICES. (1982) Report of the Working Group for the Appraisal of Sardine Stocks in Divisions VIIIc and IXa. ICES Document CM 1982/Assess: 10. 41 pp.
ICES. (2000a) Report of the Northern Pelagic and Blue Whiting Fisheries Working Group. April/May 2000. ICES Document CM 2000/ACFM: 16.
ICES. (2000b) Report of the Working Group on the Assessment of Mackerel, Horse Mackerel, Sardine and Anchovy14–23 September 1999Copenhagen ICES Document CM 2000/ACFM: 5.
Krekova M.K., Krekov G.M., Samokhvalov I.V., Shamanaev V.S. (1994) Numerical evaluation of the possibilities of remote sensing of fish schools. Applied Optics 33:5715–5720.[Web of Science]
Leblanc C.H., Mello L., Dupuis H.M.C. (1998) Analysis of juvenile Atlantic herring catches in the southern Gulf of St Lawrence September groundfish surveys. Canadian Technical Reports of Fisheries and Aquatic Sciences 1997: 35 pp.
Lockwood S.J. (1988) The Mackerel. Its Biology, Assessment and the Management of a Fishery(Fishing News Books, Oxford) 181 pp.
MacLennan D.N. and Simmonds E.J. (1992) Fisheries Acoustics(Chapman and Hall, London) 325 pp.
Martín I. (1989) Catch of small pelagic species by the live bait boats of the Basque Country in 1987 and 1988. ICES Document CM 1989/H: 34.
Massé J. (1996) Acoustic observations in the Bay of Biscay: schooling, vertical distribution, species assemblages and behaviour. Scientia Marina 60:Suppl. 2, 227–234.
Massé J., Scalabrin C., Liorzou B., Weill A. (1992) Distribution and spatio-temporal description of fish school's acoustic detections observed in the Bay of Biscay. Workshop Echospace18–20 May 1992Montpellier.
Matheron G. (1971) The Theory of Regionalised Variables and its Applications. Les Cahiers du CMM, 5, Ecole Nationale Superior de Mines Paris, Fontainebleau. 211 pp.
Mays K.F. (1974) Pelagic fish surveys in the California Current. California Fish and Game, Fish Bulletin no. 162: 79 pp.
McClatchie S., Thorne R.E., Grimes P., Hanchet S. (2000) Ground truth and target identification for fisheries acoustics. Fisheries Research 47:173–191.[CrossRef][Web of Science]
Muiño R., Carrera P., Petitgas P., Beare D.J., Georgakarakos S., Haralambous J., Iglesias M., Liorzou B., Massé J., Reid D.G. (2003) Consistency in the correlation of school parameters across years and stocks. ICES Journal of Marine Science 60:164–175.
Oliver C. W., Armstrong W. A., Young J. A. (1994) Development of an airborne lidar system to detect tunas in the eastern tropical Pacific purse-seine fishery. NOAA Technical Memorandum NOAA-TM-NMFS-SWFSC-204, La Jolla. 67 pp.
Pestana G. (1989) Manancial Ibero Atlântico de Sardinha (Sardina pilchardus, Walb.) sua Avaliaçao e medidas de gestao. Thesis, Instituto Nacional de Investigaçao das Pescas. 192 pp. (in Portuguese).
Petitgas P. (1991) Contribution géostatistiques à la biologie des pêches maritimes. Thèse doctorat, Centre Géostatisteque, Fontainebleau. 211 pp.
Petitgas P. (1993) Geostatistics for fish stock assessments: a review and an acoustic application. ICES Journal of Marine Science 50:285–298.
Petitgas P. and Prampart A. (1993) EVA (Estimation VAriance). A geostatistical software on IBM-PC for structure characterization and variance computation. ICES Document CM 1993/D: 65.
Petitgas P., Reid D., Carrera P., Iglesias M., Georgakarakos S., Liorzou B., Massé J. (2001) On the relation between schools, clusters of schools and abundance in pelagic fish stocks. ICES Journal of Marine Science 58:1150–1160.
Porteiro C., Alvarez F., Pereiro J. A. (1986) Sardine (Sardina pilchardus, Walb.) stock differential distribution by age class in ICES Divisions VIIIc and IXa. ICES Document CM 1986/H: 28. 19 pp.
Porteiro C., Carrera P., Miquel J. (1996) Analysis of Spanish acoustic surveys for sardine, 1991–1993: abundance estimates and inter-annual variability. ICES Journal of Marine Science 53:429–433.[CrossRef][Web of Science]
Porteiro C., Miquel J., Carrera P., Meixede M., Cabanas J. M. (1993) Abundance estimation and distribution of sardine in northern Spain (North of IX and VIIIc Divisions). ICES Document CM 1993/H: 28. 19 pp.
Prouzet P., Metuzals K., Caboche C. (1994) L'anchois du golfe de Gascogne: Caracteristiques biologiques et campagne de peche francaise en 1992. Rapport CNPM-IMA-IFREMER. 29 pp.
Ralston S. and Howard D.F. (1995) On the development of year-class strength and cohort variability in two northern California rockfishes. Fishery Bulletin US 93:710–720.
Scalabrin C. and Massé J. (1993) Acoustic detection of the spatial and temporal distribution of fish shoals in the Bay of Biscay. Aquatic Living Resources 6:269–283.[CrossRef]
Schwarz A.L. and Greer G.L. (1984) Responses of Pacific herring, Clupea harengus pallasi, to some underwater sounds. Canadian Journal of Fisheries and Aquatic Sciences 41:1183–1192.
Smith P.E. (1985) Year-class strength and survival of 0-group clupeoids. Canadian Journal of Fisheries and Aquatic Sciences 42:Suppl. 1, 69–82.
Soares E. (1995) Contribuçao para o Estudo da Distribuçao e do Comportamento da Sadinha na Costa Continental Portuguesa. Trabalho de síntese apresentado para provas de acesso à categoria de Assistente de Investigaçao. Area Cientifica de Recursos Haliêuticos, Instituto Nacional de Investigaçao das Pescas. 100 pp.
Squire J.L. and Krumboltz H. (1981) Profiling pelagic fish schools using airborne optical lasers and other remote sensing techniques. Marine Technological Society Journal 15:27–31.
Uriarte A., Prouzet P., Villamor B. (1996) Bay of Biscay and Ibero Atlantic anchovy populations and their fisheries. Scientia Marina 60:Suppl. 2, 237–255.[Web of Science]
Villamor B., Abaunza P., Lucio P., Porteiro C. (1997) Distribution and age structure of mackerel (Scomber scombrus, L.) and horse mackerel (Trachurus trachurus, L.) in northern coast of Spain, 1989–1994. Scientia Marina 61:345–366.[Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. H. Churnside, E. Tenningen, and J. J. Wilson Comparison of data-processing algorithms for the lidar detection of mackerel in the Norwegian Sea ICES J. Mar. Sci., July 1, 2009; 66(6): 1023 - 1028. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Scalabrin, C. Marfia, and J. Boucher How much fish is hidden in the surface and bottom acoustic blind zones? ICES J. Mar. Sci., July 1, 2009; 66(6): 1355 - 1363. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Churnside and P. L. Donaghay Thin scattering layers observed by airborne lidar ICES J. Mar. Sci., May 1, 2009; 66(4): 778 - 789. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||