Modeling statistical performance of an airborne lidar survey system foranchovy

摘要

The objective of our study was to model the performance of an airborne lidar survey system for northern anchovy in terms of survey accuracy and precision. Our analyses indicated that swath width would have little or no effect on the probability that at least one fish school would be encountered. In typical coastal waters off California (attenuation coefficient=0.1/m), about half of the schools were detected by the lidar during the day and about 64% during the night. A greater proportion of schools were detected during the night because anchovy have a shallow vertical distribution, whereas in the day, schools may extend down to 155 m; schools below about 40 m depth were not detectable to the laser. Although schools tended to be more diffuse during the night than during the day, even the very diffuse schools of anchovy (0.5 fish/m(3)) were detectable at night throughout the upper 20 m of the water column with a lidar. With a substantial increase in instrument and survey costs, it would be possible to increase the equivalent laser-pulsed power by a factor of 10 over that of the "off-the-shelf system," as used in our model. Such a change would increase the maximum detection depth of the lidar system by about 10 m but would have a negligible effect on the probability of detecting schools during the day owing to the skewed vertical distribution of anchovy schools. More effective approaches for improving the accuracy and precision of potential lidar surveys for fisheries would be to improve school detection algorithms and to develop a lidar survey model based on line transect theory to obtain an unbiased estimate of abundance. To produce an accurate reconstruction of the average vertical distribution of schools for a particular season and region, a synthesis of acoustic and lidar surveys of school distribution is required.