Today I would like to share with you my perspective of how technology can benefit fisheries management, as well as other aspects of the fisheries.

As an engineer, I tend toward a goal-driven approach to any design and development task. If I do not understand the goal, I am certain I will not achieve it. I have adopted that approach here as well.

First, I have asked the question: "If I could build the ultimate fisheries management system, without any restrictions whatsoever, what would it do?" This will be my goal.

Second, I have asked: "Where are we now?" and I provide my, somewhat limited, perspective of certain aspects of the today’s fisheries management process.

Finally, with the first two questions providing some background, I will step through a set of technologies that I think can move us from where we are today toward the ultimate fisheries management system.

What would the ultimate fisheries management system be? I envision this system being able to do the following:

1. Each fish in the ocean would, on demand, report its species, position, sex, age, size, and health. This could be done for any area of interest.
1. During harvesting, it would be possible to selectively catch only those fish that match your species, size, and sex specifications in any given area.
1. Each fishing vessel would, on demand, report its position and a complete description of all fish on-board, including their species, age, size, and sex. Furthermore, this data would be completely secure, it would only be accessible by those that have a need to know.
1. For any given area, a complete description of the fish habitat would be available, including descriptions of the oceanographic, atmospheric, and geological parameters, as well as the current condition of micro-biological, marine mammal, bird, and plant life.
1. Finally, using all these pieces of information, a model (or models) could be immediately applied that would describe the impact of the environment on fish stocks, the inter-species relationships between fish stocks, and the dependencies between fish stocks and plant, bird, and marine mammal populations. Furthermore, these models would accurately forecast the size, abundance, health, and location of fish stocks.

In addition to these five elements, this system would provide controlled access for all members of the fishing industry. As an example, a fisherman could have immediate access to his data, but would not be able to gain access to any others. Furthermore, the habitat data would be available to all people.

And, since we are dreaming, this ultimate fisheries management system would cost nothing to build, operate, maintain, and use.

Now let’s look at where we are today.

In describing a current fisheries management system, I will focus my comments on one which I am familiar, the North Pacific waters of Alaska. Again, I will look at five aspects of management.

1. First, population data is collected through a variety of means, including annual and tri-annual trawl surveys, annual longline surveys, fish ticket data, and observer reports. This process is expensive, time consuming, and, at times, incomplete.
1. Second, as both Mr. Schmitten and Mr. MacGregor had said yesterday, the North Pacific fishery is improving its ability to fish more selectively and reduce its bycatch and discard rates, but there is still more work that can be done.
1. Third, vessel reporting is done through a variety of self-compliant methods, including weekly reports from trawlers, to landing reports by pot and longline fishermen. The accuracy of this data can be very uneven. Observer proficiency varies from vessel to vessel, and fishermen are not always able to record their catch as accurately as we would like.
1. Fourth, habitat assessment is a difficult problem, at best. The tools that are available for measuring the ocean are expensive and they are not able to provide complete coverage. This problem is exacerbated by the shear volume and complexity of the data that is available, requiring a relatively sophisticated user to analyze, synthesize, and utilize this information.
1. Finally, the population models that exist today tend to focus on only a limited set of measurable parameters. For most species, these models are adequate. But, there are still many unexplainable phenomena that occur in our oceans that we have not yet been able to model and understand.

Of course, distributing the collected data is not easy either. The primary function of fisheries management is to use the data that is collected to make good management decisions. Distributing this data is a secondary function.

And, as we all know, management costs money and the budgets seem to continue to dwindle each year.

Having given you my perspective of both the ultimate fisheries management system, as well as a current management process, I will now examine ways that we can utilize technology to deliberately move us forward from where we are, toward were I think we would like to be.

As we step through these technologies, you will notice some general themes that are motivating their use:

• First, we would like to reduce the cost of data collection, but at the same time increase coverage and improve accuracy. This includes data collected both on-board fishing vessels and data from remote sensors.
• Second, we would like to be able to make better use of the data that we currently have available to us. Tools such as spatial analysis and automatic model generation can play a role here.
• Third, we would like to have faster access to all the data that is collected, which means that we need to improve the communications infrastructure in the fisheries.

Let’s begin by looking at technologies that can improve our on-board data collection activities.

There are technologies in five areas that can help to improve on-board data collection. These technologies range from species identification to automated on-board fish logging, to accurate sediment classification.

Using a combination of broadband sonar and some clever signal processing, Scientific Fishery Systems has built a system that can identify a fish species from its sonar echo. Experiments have been conducted with over a dozen different fish species ranging from skate, flounder and halibut to cod, herring, and smelt. In all cases, we have been able to achieve at least 70% correct identification of species, and often much better.

Broadband sonar fish identification can help fish stock assessment in several ways:

• First, it can improve the coverage during trawl surveys by collecting data between trawl stations.
• Second, it can reduce the number of trawl surveys required.
• Third, it will be possible to survey more area per unit time.

Each of these features provides fisheries management with more data at a lower cost. Scientific Fishery Systems is currently building its second prototype. Data collection exercises in the Bering Sea, Lake Michigan, and the Prince William Sound are all scheduled for this summer and next. We expect to have the first fish identification system commercially available in late 1998 or early 1999.

Before moving on, I would be remiss if I did not thank the U.S. Dept. of Commerce, the National Science Foundation, the National Biological Service, and the Alaska Science and Technology Foundation for their generous support of this product development effort.

Another sonar application that can have a direct impact on fish stock assessment, is the use of acoustic arrays. There are two ways that this technology can be used: passive, where we simply listen to the ocean with a string of hydrophones, and active, where we send out a acoustic pulse and listen for echoes that return.

Let’s look at the active version first. Using an acoustic projector to ensonify the region around a vessel, it is possible to detect schools of fish within a 15 mile (24 km) radius, representing a total search area of 706 square miles (1,828 sq. km). Under funding from the National Marine Fisheries Service, Scientific Fishery Systems is currently designing such a system for locating tuna schools that are not associated with dolphin. Recent results from the U.S. Dept. of Defense have also indicated that schools of salmon can also be detected in the North Pacific using this approach.

Towing such an array at 4 knots, it would be possible to search 120 sq. mi. (310 sq. km) per hour. Such a capability can enhance the survey effort by locating, and possibly tracking, large schools of pollock, salmon, tuna, and other midwater schooling fish.

When an acoustic projector is not used, the array becomes a passive listening device. In some preliminary experiments at Scientific Fishery Systems, we found that certain species, such as cod and pollock, are quite noisy during certain portions of their life cycle; so noisy, in fact, that they can be heard as far as 10 miles (16 km) away. By either hanging an array from a buoy or laying it along the bottom, it would be possible to monitor these migratory fish stocks by placing arrays in strategic locations, such as along fishing boundaries.

Technology exists today that would allow fish stocks to be automatically cataloged as they are processed on-board. Using the combination of a video camera and a scale, a fish on a conveyor can be immediately classified, measured, and weighed. This information can then be placed in an on-board database that would periodically report its contents to the management agency. Such a system could significantly enhance current observer reporting in at least two ways: it would operate 24 hours a day, and it would provide consistent reporting across all vessels.

Fish are temperature sensitive. In many instances, if you know the temperature of the water, you know which species of fish would reside there. This fact is nothing new. Temperature directed fishing techniques have been employed by the Japanese for many years. What is new is how we can measure this temperature.

To date, the only means of measuring the temperature in the water column was by dropping a temperature probe from the surface to the desired depth, or by deploying an array of temperature sensors in the water column.

Scientific Fishery Systems has recently completed a project that demonstrated the ability to measure water temperature to better than 1° C using a hull-mounted sonar system. By collecting temperature profiles using sonar, it would no longer be necessary to conduct the time-consuming, and expensive, casts. Although these results are only preliminary, the potential impact for bycatch reduction is immediate. Now it would be possible to fish in those temperature zones that are species specific.

Recently there have been several sonar systems that have emerged that provide sediment identification. The full potential of these systems still needs to be explored, but it does represent another valuable piece of information about the habitat for fish stocks. In particular, the extension of these systems to include the classification of sediment contaminants presents some interesting opportunities.

Let’s now look at the data that is available from remote sensors. I will not spend a great deal of time here, as others in this session have already told you of its potential uses in the fisheries.

As this list shows, each of these data sources provides a different view of the fishing environment. Some data, such as ocean color and sea-surface current, provide a great deal of information about where fish stocks might be located. Other data sources, such as ice-edge movement, are more concerned with safety at sea.

I think there are some key issues that need to be addressed when looking at these data sources:

• First, some of this data is very expensive, especially SAR data.
• Second, in the North Pacific there is limited use of some of these data. In particular sea surface temperature and ocean color require a clear view of the sea. The extreme cloud cover experienced in the North Pacific presents some problems here. It is possible to provide interpolation for short gaps in the coverage, but it adds some expense to the data.
• Third, the accuracy of the data is not consistent. Corrections for atmospheric distortion and other physical phenomena represents a delay in the release of the data to the fisheries. The trade-offs need to be examined between the level of accuracy and the latency of the data.
• Finally, gaining access to each data source requires specific equipment and knowledge of the sensor. It is highly desirable to have a system that can bring this type of data together with other fisheries data, such as catch and bathymetry.

At Scientific Fishery Systems, we have developed a software package, entitled Fisherman’s Associate, that is attempting to address this last issue. Currently, Fisherman’s Associate stores catch data and snags, and it tracks the vessel’s position. Fisherman’s Associate also provides detailed bathymetry and coastline to allow a fisherman to correlate catch with bottom features. The next step in the development of this Windows-based system is to incorporate data sources like those seen here. This information allows a fisherman to plan where to fish, when to fish, and he has some indication of what species he will catch when he fishes there. Another logical next step for Fisherman’s Associate would be to incorporate an electronic reporting function with the cognizant management agency. As a fisherman is bringing in his catch, he can immediately log it in Fisherman’s Associate. At the end of the trip, this information can be collected, formatted, and transmitted. This would eliminate errors, and it would provide more detailed catch information than is currently available.

I would now like to switch from the discussion of improving our data collection efforts, to methods of processing the data once it is collected.

There are several methods of information processing available to the fisheries. I would like to highlight just two that I think have something to offer: spatial analysis and automatic model generation.

Given all the data sources that I have described so far, it becomes apparent that we have a data glut. There is a massive amount of data, yet only a few decisions are required. In particular, where are the fish?

Spatial analysis provides us with a wealth of tools to solve just such problems. Tools such as this have been used extensively in oil exploration, planning military operations, and in agriculture. As shown here, some of the applications are in the areas of data integration, identification of bycatch areas, and species correlation analysis.

I know from first-hand knowledge, that this is a technology area that is being used by the fisheries of the North Pacific. The technical challenge is to make this technology affordable and accessible.

The second area in information processing that I would like to discuss is automatic model generation: that is, how do we generate accurate models from the data without necessarily having to explicitly define the dynamics of each parameter.

Three technologies that can be applied here: Database mining, neural network modeling, and nonlinear dynamical systems.

Database mining is a set of tools that seek to extract correlations and trends buried within large databases. This technology is used to find new stars, locate profitable locations for a real-estate development, and identifying good candidates for credit cards. In the fisheries, database mining can be used to explore such questions as: "Why is the Stellar Sea Lion population declining?" and "What are the primary factors affect the return of salmon?"

Neural networks are able to learn complex nonlinear mappings. They have been applied to problems ranging from identifying cancer cells to controlling robots. One area where neural networks have excelled is in modeling. In finance, neural networks are used to model stock prices, exchange rates, and purchase decisions. In the fisheries, neural networks could be used to model marine mammal populations, environmental changes, and fish migrations.

The last area, nonlinear dynamical systems, is concerned with modeling the changes in chaotic and semi-chaotic systems. To the extent that you might feel fish population dynamics is chaotic, these techniques may have some applicability.

The last area that I would like to mention is communications. The transfer of information from ship-to-shore is difficult. New technologies such as the proposed Teledesic and Iridium Low-Earth Orbiting Satellite systems will significantly change how information will be exchanged in the fisheries. With these systems in place, it will be possible to affordably and reliably transfer voice, data, and even imagery between any two points on the globe.

This, coupled with the emergence of the internet, as the global information repository of choice, will allow information to flow very easily between fishermen, fisheries markets, and fisheries management. Possibly more than any of the technologies I have described today, the ability to internet the ocean could have the most dramatic impact on the fisheries.

To summarize, I have attempted to expose you to some technologies that I feel can assist us with fisheries management. Some of these technologies, like Fisherman’s Associate and other products that have been presented in this session, are available today. Others, such as the broadband fish species identification system, will be available very soon. Still others, such as database mining and internetting the ocean, will take more time.

Will these technologies lead us toward the ultimate fisheries management system? I am not sure. But, I am sure that they can move us in that direction, and I am honored to be a part of that process.

In closing, I would like to thank the University of Alaska Anchorage’s Alaska Center for International Business, and Japan’s Ministry of Agriculture, Forestry and Fisheries for sponsoring this event. I would also like to thank our esteemed conference co-chairs, Mr. Minoru Morimoto and Mr. Steve Cowper, for inviting me to talk to you at this prestigous event. Finally, I would like to thank all of you for your time and attention. Thank you.