E/9
AUTOMATIC PROCESSING OF WESTERN AUSTRALIAN PRAWN FISHERMEN'S LOG-BOOKS

by

R. J. SLACK-SMITH
Department of Fisheries and Fauna
Perth, Western Australia

and

A. E. STARK
CSIRO Marine Laboratory
Cronulla, New South Wales, Australia

Abstract

A system of intensive log-book data collection from Western Australian prawn fishermen is described. To expedite computation and tabulation of this volume containing a considerable amount of data, two computer processing systems have been developed.

The first, designed for the University of Western Australia's PDP-6 computer, summarizes raw data into vessel by area by month groups, giving an immediate overall picture of the industry. Monthly summaries are forwarded to fishermen for their use in estimating the relative efficiencies of their vessels.

The second system, using the Commonwealth Scientific and Industrial Research Organization's CDC 3600 computer, is a system of storage, analysis, handling and retrieval of detailed data. Raw log-book data, landings and programs are stored on magnetic tape and all subsequent processing is carried out in stages, within the computer's high-speed store, magnetic tapes and magnetic drum storage. The results of each processing stage are stored on tape and are readily available for checking. Daily summaries of efforts, adjusted catch of each species, catch per unit of effort (C/UE) and log C/UE are tabulated for each vessel in each area fished. Further programs for this system are being developed.

The system is flexible and could be readily adapted to other research programs where large quantities of data are stored and manipulated.

TRAITEMENT AUTOMATIQUE DES LIVRES DE BORD DES PECHEURS DE CREVETTES DE L'AUSTRALIE-OCCIDENTALE

Résumé

La communication décrit un système de rassemblement des données des livres de bord des pêcheurs de crevettes de l'Australie Occidentale. Pour accélérer le calcul et la mise en tableaux de la masse considérable de renseignements, on a mis au point deux systèmes de traitement par ordinateur.

Le premier, conçu pour l'ordinateur PDP-6 de l'Université d'Australie-Occidentale, récapitule les données brutes par bateau par zone et par mois. ce qui permet d'avoir immédiatement une idée d'ensemble de l'industrie. Des récapitulations mensuelles sont adressées aux pêcheurs afin qu'ils les utilisent pour évaluer le rendement relatif de leurs bateaux.

Le second est un système de mémorisation, d'analyse, de manipulation et de récupération des données détaillées faisant appel à l'ordinateur CDC 3600 de la Commonwealth Scientific and Industrial Research Organisation. Les données brutes provenant des livres de pêche, ainsi que les détails sur les mises à terre et les programmes sont enregistrés sur bande magnétique, et tout le traitement ultérieur est effectué par étapes par l'ordinateur, au moyen de sa mémoire rapide, de ses bandes magnétiques et de sa mémoire à disques. Les résultats de chaque stade du traitement sont enregistrés sur bande et aisément disponibles aux fins de contrôle. Des récapitulations journalières de l'effort de pêche, des prises ajustées de chaque espèce, des captures par unité d'effort et par livre de pêche sont mises en tableaux pour chaque navire dans chaque zone exploitée. On travaille à l'élaboration de nouveaux programmes dans le cadre de ce système.

Le système, du fait de sa souplesse, pourrait aisément être adapté à d'autres programmes de recherche impliquant l'enregistrement et la manipulation de données très nombreuses.

ELABORACION AUTOMATICA DE LOS DATOS DE LOS REGISTROS DE LOS PESCADORES DE LANGOSTINOS DE AUSTRALIA OCCIDENTAL

Extracto

Se describe un sistema de recogida intensiva de datos de los registros que llevan los pescadores de langostinos de Australia occidental. Para acelerar la computación y tabulación de este considerable volumen de datos, se han establecido dos sistemas de elaboración con calculadoras electrónicas.

El primero, diseñado para la calculadora PDP-6 de la Universidad de Australia Occidental resume los datos brutos por grupos correspondientes a embarcaciones por zona y por mes dando un cuadro completo e inmediato de la industria. Los resúmenes mensuales se envían a los pescadores para que les sirvan para estimar el rendimiento relativo de sus barcos.

El segundo es un sistema de almacenamiento, análisis, manipulación y localización de datos detallados usando la calculadora 3600 CDC de la Organización para la Investigación Científica e Industrial (CSIRO) en el Commonwealth. Los datos brutos del libroregistro, y los detalles de los desembarques y programas se conservan en cintas magnéticas y toda la elaboración ulterior se lleva a cabo por fases en la memoria de acceso rápido de la calculadora, en las cintas magnéticas y en la memoria de las calculadoras de disco. Los resultados de cada fase de elaboración se conservan en cintas que se hallan dispuestas siempre para su comprobación. Se tabulan para cada embarcación en cada zona pescada resúmenes diarios de los esfuerzos, de la captura ajustada de cada especie, de la captura por unidad de esfuerzo (C/UE) y datos de registro C/UE. Se están ideando otros programas para este sistema.

El sistema es flexible y podría ser adaptado fácilmente a otros programas de investigación en los que hay que almacenar y manipular grandes cantidades de datos.

1 INTRODUCTION

Since 1964, detailed information on the operations of prawn trawlers in Shark Bay, Western Australia, has been collected by using log-books issued to skippers. As the volume of these data is considerable (up to 15,000 individual trawl records each year), a system of automatic processing has been devised so that this information can be safely stored, readily manipulated and detailed analyses made. This paper describes the system, the methods of data acquisition, and the methods of producing detailed summaries of these data.

Processing is carried out on two digital computers, the Digital Corporation PDP-6 housed in the Western Australian University and the Control Data Corporation CDC-3600 housed in the Commonwealth Scientific and Industrial Research Organisation Computing Research Section's (CRS) Centre at Canberra, A.C.T. The PDP-6 has approximately 28,000 words of high speed memory storage, using 36 bit words. This installation, at present, is somewhat limited for handling detailed analysis of large volumes of data as its ‘hardware’ does not include magnetic tape transports or a high speed line printer. Being near the site of data preparation, this machine is used for the initial editing and summarizing of information. The CDC-3600 has 32,768 words of high speed memory storage, using 48 bit words; 8 magnetic tape transports; 1,048,576 words of magnetic drum storage, and a full range of other peripheral equipment. The computer is controlled by the Drum-and-Display (DAD) monitor, which allows computer documents to be recorded on the magnetic drums and held pending execution. This allows series of operations, each having intermediate results, to be returned to the user. CARDIMED and SORT II are amongst the software provided; both are an integral part of the processing system described.

CARDIMED is a procedure devised for the establishment and maintenance of a ‘library’ of ‘data’. This may consist of computer programs or information of any kind capable of being read by the computer. The library consists of versions (‘editions’), and is stored on reels of magnetic tape. An edition has the facility of updating itself in response to control cards and information presented to it on input media. It does this by rewriting itself with appropriate amendments. To guard against accidental loss of the library, it is customary to hold the three or four most recent editions.

A card image, or character within a card image, may be altered, or whole images deleted or replaced, by feeding appropriate control cards to the computer. The old edition itself is not altered, but a new reel is prepared incorporating the changes requested. The contents of amended record blocks and a complete inventory of a new edition is output on the line printer.

SORT II accepts a file of records on magnetic tape, arranged in any order, and rearranges these records into a prescribed order, according to sequencing keys contained within each record. The result is identical with that from mechanical sorting.

All routines have been written in FORTRAN IV by the authors.

Processing is carried out in three phases:-

(i) data preparation and monthly summaries; (ii) final editing and transfer to the data library; and (iii) production of detailed summaries.

2 PHASE I

2.1 Data collection and preparation

Log-book information is supplied voluntarily by individual skippers who record the circumstances and outcome of each trawl hauled during the course of their operations. The Western Australian Department of Fisheries and Fauna provides log-books specially designed for this purpose. (See Fig. 1 for a typical page from a book). The design of the log is simple; it records adequate research data but is also suitable for fishermen to keep their personal records. The books have duplicate pages impregnated with chemicals, so that the originals remain for the fishermen's use after duplicates have been collected by Departmental personnel. These duplicates are edited to eliminate indistinct or obviously incorrect entries, coded where necessary, and landed catch weights are recorded in the spaces provided. The latter data are obtained independently from the processing sites.

Fig. 1 A page of a prawn fishery log-book from Shark Bay.

Commonly, fishermen make fishing trips of several days' duration and weigh in their catches on return. At the end of each day's fishing all hauls are summed to give total catches of each species and time fished for that day. These totals are coded for automatic processing by placing 98 in the “Shot No.” column. Landed weights are identified by placing 99 in the “Shot No.” column.

A card is punched for each completed line, including “98” and “99” lines. After punching, the cards are assembled in the same order as lines on the logs. Fields corresponding to data on the log-sheet are printed on each card to facilitate editing (Fig. 2). Each card is identifiable by its vessel, by skipper number, date, and shot serial number. Vessel and skipper numbers, each between 001 and 499, are assigned arbitrarily. These are kept constant so that a fishing unit can always be identified.

2.2 Elimination of gross errors from the card deck

Invariably, large numbers of incorrect entries occur in card decks punched from the raw log-sheets. Difficulties experienced by card-punch operators in reading these hand-written entries cause most errors. These errors are detected by editing the deck on the PDP-6 using the program SBAYED. This program adds the catches reported for individual shots and compares the sum of reported catches with the processor's records of landed catch. The computer output (Fig. 3) is examined for anomalies and the deck is revised by replacing defective cards manually.

The data deck is again edited using SBAYED to ensure that all errors have been eliminated.

2.3 Compiling of monthly summaries

The data is next run under LOGSUM, a program written to produce monthly summaries of each vessel's operations (Fig. 4). These provide research and administrative personnel of the Fisheries and Fauna Department with up-to-date resumés of the fishery as the season proceeds. Each skipper receives a copy of the summary computed from his log.

Pooled monthly summaries from all vessels fishing in each month can also be obtained if required.

3 PHASE II - FILING NEW INFORMATION ON THE LIBRARY

Before transmission to the CDC 3600 CARDIMED library, the deck is interleaved with appropriate CARDIMED control cards. These divide it into data blocks classified by fishing area, year, and fishing unit number, (i.e., vessel x skipper code). Each block is named by an eight character identifier: the first character is alphabetic (S for Shark Bay, E for Exmouth Gulf, etc.), the second numeric (6 for 1966, etc.), and the remaining 6 are the unit number. The system requires that each block of data does not exceed 1,000 card images; if a natural block exceeds 1,000 images it is manually subdivided into two blocks by adding 500 to the vessel number.

Fig. 2 Card layout for prawn fishery log-book data.

WEST AUST PRAWN FISHERY * LOG BOOK ANALYSIS * BIOLOGIST IC R.J. SLACK-SMITH

Fig. 3 A segment of output from the prawn log-book editing program SBAYED.

DEPARTMENT OF FISHERIES AND FAUNA
MONTHLY LOG BOOK SUMMARY - 1966
SHARK BAY

Fig.4 Typical monthly summary computed from Shark Bay prawn log-books. These summaries are produced by the program LOGSUM.

The deck, together with control cards, is then submitted to the computer.

After the deck has been copied on to the library, the new contents are edited using SBAYEDIT (the CDC 3600 version of SBAYED) to detect any errors which may have passed unnoticed at the first editing or have been introduced in manipulating the cards or copying them on to the library.

4 PHASE III - THE WESPRAWN SYSTEM

4.1 Outline of the WESPRAWN System

WESPRAWN has been designed so that it will accept widely-varying amounts of data without modifications to the routines. Furthermore, because of the flexibility of CARDIMED, amendments can be made easily to any data on the library, including high-refined summaries. This capability, together with the fact that WESPRAWN itself is carried out in a number of distinct phases, ensures that an unnoticed deficiency in the data, detected at any processing phase, can be remedied without going through either the whole set of data or the whole processing procedure. Data may be processed piecemeal or as a whole for any given area. WESPRAWN is capable of easy modification to accept data of any sort which is substantially in the same format as described above.

Routines stored on the library can be invoked by presenting appropriate calling (control) cards to the computer. Similarly, data of any order of refinement stored on the library may be transcribed to temporary storage media, such as ‘scratch’ magnetic tape or in the form of ‘documents’ to magnetic drums. The routines can then be commanded to process the data by presenting the necessary control cards. Thus, once the routines and data are stored on the library, it is possible to commence processing by presenting small decks of control cards to the computer.

4.2 Haul adjustment and separation

Using the routine CADJREC, hauls are adjusted so that the sum of catches taken in each haul agree with the total landed catches of each trip. In the same operation new lines of information, one for each haul, are written in data blocks on magnetic tape. These blocks are interspersed by end-of-file marks. These lines are in the form of card images; the first part is the original haul data, the second is a series of adjusted catches, and the final part an area code (e.g. T for Shark Bay) with the month of the year (e.g. T6). Simultaneously, control cards are punched on-line by the computer. These are used to add the new blocks of adjusted data on the library. A segment of this output is shown in Fig. 5.

4.3 Sorting blocks of adjusted data

In this step, blocks of adjusted data are copied from the library on magnetic tape, re-arranged into day x fishing x area order with SORT II, and are then output on another magnetic tape. Hauls are sorted so that those belonging to the first day come first, followed by all the second day hauls, etc. Within each day-group the hauls are sorted in order of area.

4.4 Computing effort, catch and catch per unit of effort

The sorted file described above is now processed by SBAYCOMP, which summarizes the data within each day x block subclass. When it detects a change in date or block a new subclass summary is commenced. Each haul is recognized by its unit number. SBAYCOMP must be provided with a list of fishing units appropriate to the particular run; at present the number of these units is limited to 50. The sequence of data and blocks within dates need not follow any special pattern, thus making this routine quite flexible with respect to the structure of the data. In addition to the day x block summaries, tables combining all data for blocks fished during each day are compiled. A specimen of the output is given in Fig. 6.

Fig. 5 A segment output from the trawl adjusting program ADJREC.

WA PRAWN FISHERIES * LOG BOOK ANALYSIS * SHARK BAY 1964
VESSEL X DAY X BLOCK SUMMARY

Fig. 6 A summary or prawn fishing operations in a particular area (block) for a particular day computed by the WESPRAWN system.

4.5 Proposed extensions to WESPRAWN

Proposed extensions and refinements to this processing system include:

1. Provisions for addition of day x block summaries to the CARDIMED library.

2. A routine (T4HR), to convert all “start times” recorded on the log sheets to 24 hour time, has been written. This routine will allow time recorded on log sheets to be punched directly, thus eliminating a large amount of coding.

3. Maps of the fishing areas overlaid with selected catch-rate data will be produced on the graph-plotter.

4. Improved data-scanning facilities and more comprehensive diagnostic output will be written into existing routines, so that data deficiences will be more effictively eliminated.

5 DISCUSSION

The Western Australian prawn fishery extends several hundred miles north of Shark Bay and further expansion is now occurring. Its detailed growth is difficult for existing research staff to monitor manually. Intensive log book promotion has resulted in a high proportion of fishermen accepting log books from the Fisheries and Fauna Department, primarily to keep their own records. Most skippers willingly post in duplicates from areas not visited by research staff. The data obtained can now be stored and summarized using the processing systems described above, thus providing adequate up-to-date figures for administrative purposes and storing detailed data in a form readily available for later analysis.

Copies of the routines, and further details of these systems may be obtained from either author.