Advances in At-sea Electronics and Data Collection at the Alaska Fisheries Science Ctr.

Figure 1. A trawl mensuration computer in use in the
wheelhouse of an AFSC survey vessel.

A major responsibility of the Groundfish Assessment Program of the Alaska Fisheries Science Center’s Resource Assessment and Conservation Engineering (RACE) Division is to conduct bottom trawl survey assessments of marine fish and invertebrate species of the North Pacific Ocean and Bering Sea. Survey results are utilized in a range of research activities, the most important of which is to provide estimates of fish and crab abundance which are utilized as model parameters in annual stock assessments for fisheries managed by the North Pacific and Pacific Fishery Management Councils.

Many RACE Division bottom trawl surveys are conducted from a number of chartered fishing vessels. It is important to maintain consistency in how the research trawls are fished between different vessels and from tow-to-tow so that survey results reflect accurate catch rates and not differences between vessels and fishing methods. Accurate information about the performance of the trawl (width and height of the trawl opening and contact of the footrope with the seafloor) and the distance fished (the linear distance of ocean bottom swept by the trawl) are important measurements needed to calculate biomass and to determine if the trawl is fishing properly.

Over the past 25 years, RACE scientists have worked to improve their knowledge of the performance of and area covered by the trawls used in their survey work. In the past decade, improvements in computer software and hardware, in conjunction with advances in the technology of fishing gear mensuration instrumentation, have allowed us to integrate information on trawl performance, fishing effort, catch, and biological data collections aboard ship. These advances have improved the quality of data collected, improved our ability to maintain consistency in the fishing operations between different vessels, and reduced the time needed to analyze the data and produce the survey results.

Trawl Mensuration
One of the greatest challenges in conducting RACE trawl surveys has been to properly measure the area fished by the trawl during a tow. Because the survey data are used to estimate fish abundance through application of the area-swept technique, accurate estimates of the width and length of the area towed are very important. During the 1970s, experimental work conducted by Fred Wathne of the RACE Division assessed the average spread and height of RACE standard research trawls. The average values were used to derive area-swept biomass estimates. During the 1980s, acoustic trawl mensuration gear became a standard part of survey instrumentation to monitor and record the spread of the trawl wings and height of the trawl opening; however, data were not available for all hauls, so trawl mensuration parameters for many hauls were estimated using data from similar tows for which mensuration data was available. In the 1990s, use of trawl mensuration equipment and measurements became the standard on most tows (Figure 1 above). Today, most tows have measured values for trawl height and width.

Originally, the length of the tow (both in time and distance traveled) was estimated from events during the fishing process that could be observed from the vessel’s bridge or deck. A tow began with the setting of the trawl winch brakes and ended when the winches began to retrieve the trawl from the bottom. The distance covered by the vessel during a tow was calculated as the time between these events multiplied by the average vessel speed. The area swept by the trawl was estimated by the length of the tow times the average net width for the tow.

With the application of global positioning system (GPS) technology to RACE surveys it became possible to more accurately measure the distance traveled during a tow and coordinate all the trawl mensuration data with an actual position. The RACE Division currently uses a military-grade GPS receiver. Scangraph, the in-house software program developed by the RACE Division to collect trawl mensuration data on personal computers (PC) , was modified to simultaneously collect trawl configuration measurements and GPS position data. The current version of Scangraph allows the individual who monitors trawl performance from the vessel’s bridge to observe height and width measurements, geographic position, and elapsed time in a windowed-PC environment in real time during the tow. Using the program, the user can log important events such as setting the brakes of the winches, beginning of towing configuration, haulback, and other events while the trawl is towed. A histogram of trawl height and spread allows tracking trawl configuration and performance over the length of the tow (Figure 2).

In 1993, RACE scientists began collecting depth and temperature data with an MBT (microbathy-thermograph) attached to the trawl (Figure 3).

photo as described in figure 3 caption (59765 bytes)

Figure 3. Scanmar headrope height sensor microbathy-
thermograph mounted on the headrope.

The instrument collects data during a tow and stores the data in memory. The data stored in the MBT is downloaded through a standard serial port to the trawl mensuration computer at the conclusion of the tow. The MBT and trawl mensuration data provide information on the behavior of the trawl, showing descent and ascent rates, trawl height and width, and bottom contact. However, these data do not provide information on contact of the footrope with the ocean floor, an important aspect of trawl performance that minimizes escapement of fish under the net.

In 1994 Scott McEntire of the RACE Division developed a bottom contact sensor (BCS) to attach to the footrope of the trawl to record the angle of the BCS in relation to the footrope
(Figure 4).

The BCS remains vertically in the water column until the footrope contacts the bottom, whereupon the BCS is forced to move into a more horizontal position until the footrope leaves the bottom. The data from the BCS is downloaded at the conclusion of each tow via an optical data shuttle and then downloaded to the trawl mensuration computer. The combination of data received from the net mensuration system, the GPS receiver, the MBT, and the bottom contact sensor give an accurate measure of the configuration and performance of the trawl during towing and provide the information necessary to determine if the tow is valid or if it needs to be repeated.

At the conclusion of the tow, the trawl mensuration and GPS data are integrated with the data downloaded from the MBT and BCS. The Scanplot software integrates these data. To standardize the time of each event measured and recorded during the tow, the clocks of all instruments are set to the trawl mensuration computer clock which is updated each haul directly from the GPS satellites. Scanplot provides the user plots of the trawl height and spread, MBT data, and BCS data in four separate stacked panels with a moveable cursor bar (Figure 5). The user moves the cursor bar to line up the events from all devices and determines the official beginning and end of the tow based on all available data, with priority given to the results from the BCS showing bottom contact of the footrope. A tow is defined as beginning when the footrope has contacted the seafloor and ending when contact ceases after haulback. A plot from a standard tow shows the vertical opening of the trawl decrease and the width of the trawl increase as bottom contact is initiated. A subsequent leveling off of the depth shown from the MBT depth data also occurs. Since the MBT and BCS data are not available until after the tow is completed, the trawl height and spread data displayed in real time are used to estimate the “on bottom” event and intitiate timing the duration of the tow.

Deviations from the standard trawl configuration can signal potential trawl or seafloor substrate problems (trawl caught on bottom, for example) which can be examined after the tow is completed and when the MBT and BCS data also are available. Observations from sequential tows allow close estimation of tow length from the combination of available data if one source of data is absent due to instrument failure. After determination of the start and end times of the tow, the Scanplot program calculates the time and distance fished (linear length) of the tow, average height and width of the trawl, average bottom depth and temperature, and other summary tow data, such as number of signals received from the trawl mensuration transducers.

These advances have substantially improved our knowledge of trawl performance during the tow and improved our estimates of the area “swept” by the trawl and, consequently, our resulting estimates of fish and crab abundance.

At-Sea Data Entry
Advances in electronic technology have allowed us also to improve the methodology for sorting, weighing, and enumerating the catch, and the methods used to collect biological data (length, age structures, food habits, etc.) Traditionally, biological data on RACE bottom trawl surveys were recorded on waterproof paper forms, and the data were later entered into a computer. With the subsequent availability of small portable computers we began entering data at sea. During early surveys, the data entry was accomplished at the AFSC facilities in Seattle after a survey was completed. As computers have become more powerful, we have introduced more features into the data entry software and improved our data editing at sea. Because timely reporting of survey results is critical to resource management, we have emphasized improvements that help make survey results available more quickly.

During the 1990s, the RACE Division instituted a computer program to reduce errors during data entry at sea and to verify and correct data during the time of data collection rather than months later in the office. Although data have been logged at sea for more than 10 years, a vast majority of data editing has been done after the data were transferred to a mainframe computer at the AFSC Sand Point facility in Seattle. Data errors were difficult to evaluate and correct months after the data were collected. As a result, RACE computer specialists worked to develop a data entry and editing program that would bring data entry transcription errors to the user’s attention as soon as possible after data collection.

The first step was to eliminate errors made in the entry of fish length data by replacing the length board’s traditional pencil-marked plastic strips, which required manual length measurements which were then transcribed to paper forms and then keyed into the computer. The strips next were washed to reuse for the following length sample, so if errors were made, the original data were lost. Our new Polycorder system measures length electronically with bar-coded length strips which are read by a portable, waterproof, hand-held computer equipped with a light pen. The fish is placed on the bar-coded length strip and the pen is used to read the bar code associated with each fish length (Figure 6).

Figure 6. Polycorder fish length board in use during a RACE
bottom trawl survey.

Length strips contain other command bar codes which are used to change the species or sex codes in the data logger as fishes in the length sample are processed. Because of the large number of fish species that are sometimes measured for length (more than one hundred species in some surveys), the bar-coded length intervals on the length strips also double as species codes. The length of our standard board is 120 cm, which serves both purposes well. After all lengths have been collected, the data are downloaded to a PC for entry into a database. The weight and number of each species occurring in the catch are still collected on waterproof sheets , as are specimen data, and entered into the computer after the catch is fully processed later in the day.

Most specimen data (otoliths, DNA, and maturity samples, for example) are accompanied by individual length and weight data for the specimen. Until the early 1990s, specimens were weighed with a mechanical triple beam balance. The mechanical balance worked well under ideal conditions, but was difficult to use under our frequently rough wind and sea conditions. The balance also required constant vigilance to keep track of the scale’s hanging weights to prevent misreading of the scale weight and the consequential perplexing editing problems in the data. During the past decade, motion-compensated electronic scales have vastly improved our collection of individual weight data on survey vessels. With the new electronic scales, weight data collection is no longer dependent upon good weather, and speed and accuracy of processing have improved substantially.

All biological data are recorded at sea with a catch data entry program based on Microsoft Access, which allows the data to be quickly tested and integrated into a database. Information can be integrated and tested in a database more readily than from separate flat file computer systems. The user enters the computer program from a master form that contains all the main functions of the program. The length data from the polycorder is downloaded into the database first and is used in conjunction with weight-length regression parameters determined from earlier surveys to provide the user with an estimated weight for the measured subsample of fish. Because length subsamples are all weighed and recorded separately from the nonlength measured part of the catch, this serves as a reasonable cross check of both length and weight subsamples. If there is a substantial discrepancy between the estimated weight from the length sample and the weight recorded on the deck catch sheet, it can be checked immediately while the work is still fresh in everyone’s mind. After the length data has been vetted, the polycorders are cleared and prepared for the next tow. Ideally, all data should be entered after each tow, but only the polycorder data must be downloaded and checked immediately. If tows are spaced closely together, catch and specimen data are entered later as time allows. Catch data are entered into the Access database using an interactive form with drop-down menus (Figure 7). A list of all species encountered on the survey in previous years is presented in a drop-down selection box. Species names must be selected from this list or a new name and species code can be documented and added. Selecting species from a list enforces the integrity of the species codes in the list and reduces the chance of transcription errors in entering the species name. As data are entered, the weights are again checked against estimated weights from the length data and the user is warned if the difference is significant.

Although it is no longer necessary, length data can be entered manually as well. A separate form allows entry of Pacific halibut data for International Pacific Halibut Commission scientists who often participate in our surveys. An editing utility allows editing of any data from previous hauls.

Our standard methodology to determine the species composition of each catch is to sort and weigh the catch of each species from every tow, but occasionally we get a haul that is too large to entirely sort and weigh even with our 20,000 kg load cell. On these occasions, we estimate weight using volumetric methods. In these cases, we determine either the volume of catch in the trawl or in bins on deck and then convert the volume to an estimate of weight by applying an estimated catch-specific gravity to the volume. The data entry program has a feature to facilitate this process (Figure 8). Users can choose a series of geometric shapes to estimate volume and are prompted for appropriate dimensions. The program then provides a uniform printout of estimated density and total weight.

The catch entry program has user-initiated backup utilities built in. Normally the data are backed up daily to hard and floppy disks. All files produced by the data entry program are standardized to the comma delimited with field names in the first row format used in all our programs.

One advantage of building the data entry program within a database environment is that the entire database can be returned from the field to the laboratory for final editing. Several editing functions are built into the database, including functions to check the data structure and hierarchical relationships between tables, testing numbers in the catch data versus numbers in the length data, testing for unusual average weights, and testing for weight or length outliers in specimen data. When accompanied by final line-by-line checking of the data at the lab, this helps ensure a high level of data integrity.

Future plans
The next step in the application of modern weighing, measuring, and computer technology to RACE survey field work will be to integrate specimen data collection into an on-deck, field computer environment. Currently, length, weight, specimen number, and sometimes other data such as maturity stage are measured and recorded on a waterproof form and later transcribed to the computer through the data entry program. Because our electronic scales can communicate with external devices during the weighing of the specimens, we plan to integrate data collection through a waterproof field computer that will collect lengths with a bar code wand and weights directly from the electronic scale, as well as keeping track of specimen numbers.

Catch composition data collection can be improved with direct readout from an electronic scale to a handheld computer. Because the catch data collection more frequently involves taking notes on deck, it is likely this will have to be done with a screen-oriented notepad computer.

The final step will be an integration of the computers on the boat. Currently, one computer is dedicated to haul data, trawl performance, and fishing effort data (usually on the bridge) and another is dedicated to data entry, because these events may occur simultaneously. With the advent of simple intercomputer communications built into operating systems, a local area network of computers on a survey vessel at sea is feasible. With all computers communicating on the boat, the field computers could act as expert systems to provide feedback on sampling decisions, such as suggesting specimen data collection strategies by monitoring the status of collections within strata.

All of our past advances in measurement of trawl performance, fishing effort and the electronic recording and editing of data at sea, as well as those planned for the future, improve the quantity and quality of the data collected during our resource surveys which are used in the conservation and management of our living marine resources.

Advances in At-Sea Electronics and Data Collection

Advances in At-Sea Electronics
and Data Collection