HERAS survey indices: automation, TAF and testing

Fisheries acoustic surveys are routinely conducted around the world, and particularly within the ICES community. Following pre-defined transects, these surveys make use of downward active acoustic systems (i.e. so called scientific echosounders) coupled with biological sampling through fishing operations to estimate abundance and distribution of marine species. Of interest here is the HERAS survey that takes place yearly for ~1 month and is a dedicated international survey effort. It is performed across the North Sea, West of Scotland and the Malin Shelf, more specifically in the context of the Western Baltic Spring Spawning (WBSS) and North Sea Autumn Spawning (NSAS) herring stocks. Recently, the calculation of the WBSS and NSAS indices underwent a major update with the introduction of the ICES acoustic trawl database and the use of the StoX software. In the context of these recent changes, the aim of the current study is three-fold: 1. Develop R code for the automatic calculation of the NSAS and WBSS indices with the aim of minimizing manual user input. 2. Run sensitivity tests against various assumption on NSAS and WBSS indices. 3. Generate NSAS and WBSS indices under ICES TAF (Transparent Assessment Framework). Overall, good agreement was found between previously derived indices and indices calculated using automatic routines. The only large discrepancies found was for 2017 in strata 11 and 141 which is accountable to discrepancies in WBSS/NSAS stock separation in these strata. This should be investigated further but the automatic routines as derived through this project will be used in the future for the derivation of the index alongside rigorous checking of outputs. Yearly, this process will also be running on the ICES TAF framework, providing improved transparency and robustness. A secondary aim for this study was to run sensitivity tests programmatically. This was done for a range of assumptions around: calibration of acoustic instruments, stock splitting in strata 11 and 141, alternative strata definition and alternative haul allocation strategy. Firstly, it was found that calibration error was very influential on the index. Considering this, it is recommended to run thorough checks of calibration results prior to every survey (e.g. yearly comparing results historically). Secondly, the stock splitting in strata 11 and 141 mostly influences WBSS and the introduction of novel identification methods will be beneficial, rotating out the currently used method based on vertebrae count. Thirdly, the change in strata tested here did not exemplify large differences, suggesting that the current stratification is appropriate. Lastly, the use of automatic haul allocation to transect was tested and was found influential, highlighting the importance of expert input during for the haul allocation process.

Saved in:
Bibliographic Details
Main Authors: Bergès, B.J.P., Sakinan, S., Berg, F., Lusseau, S.M., Schaber, M., O’Connnell, S.
Format: External research report biblioteca
Language:English
Published: Centre for Fisheries Research (CVO)
Subjects:Life Science,
Online Access:https://research.wur.nl/en/publications/heras-survey-indices-automation-taf-and-testing
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Fisheries acoustic surveys are routinely conducted around the world, and particularly within the ICES community. Following pre-defined transects, these surveys make use of downward active acoustic systems (i.e. so called scientific echosounders) coupled with biological sampling through fishing operations to estimate abundance and distribution of marine species. Of interest here is the HERAS survey that takes place yearly for ~1 month and is a dedicated international survey effort. It is performed across the North Sea, West of Scotland and the Malin Shelf, more specifically in the context of the Western Baltic Spring Spawning (WBSS) and North Sea Autumn Spawning (NSAS) herring stocks. Recently, the calculation of the WBSS and NSAS indices underwent a major update with the introduction of the ICES acoustic trawl database and the use of the StoX software. In the context of these recent changes, the aim of the current study is three-fold: 1. Develop R code for the automatic calculation of the NSAS and WBSS indices with the aim of minimizing manual user input. 2. Run sensitivity tests against various assumption on NSAS and WBSS indices. 3. Generate NSAS and WBSS indices under ICES TAF (Transparent Assessment Framework). Overall, good agreement was found between previously derived indices and indices calculated using automatic routines. The only large discrepancies found was for 2017 in strata 11 and 141 which is accountable to discrepancies in WBSS/NSAS stock separation in these strata. This should be investigated further but the automatic routines as derived through this project will be used in the future for the derivation of the index alongside rigorous checking of outputs. Yearly, this process will also be running on the ICES TAF framework, providing improved transparency and robustness. A secondary aim for this study was to run sensitivity tests programmatically. This was done for a range of assumptions around: calibration of acoustic instruments, stock splitting in strata 11 and 141, alternative strata definition and alternative haul allocation strategy. Firstly, it was found that calibration error was very influential on the index. Considering this, it is recommended to run thorough checks of calibration results prior to every survey (e.g. yearly comparing results historically). Secondly, the stock splitting in strata 11 and 141 mostly influences WBSS and the introduction of novel identification methods will be beneficial, rotating out the currently used method based on vertebrae count. Thirdly, the change in strata tested here did not exemplify large differences, suggesting that the current stratification is appropriate. Lastly, the use of automatic haul allocation to transect was tested and was found influential, highlighting the importance of expert input during for the haul allocation process.