CCAMLR

As a contribution to the joint effort on the large-scale international synoptic krill survey of Area 48 (2019 Area 48 survey), the Chinese krill fishing vessel F/V Fu Rong Hai conducted the acoustic survey around the South Shetland Islands and the Elephant Islands in the Subarea 48.1 from 5^th to 10^th February 2019. Implementation and preliminary results on the estimates of krill biomass density were presented to the WG-EMM 2019 meeting. After the meeting, the acoustic data were processed further using the swarm-based method to identify krill backscatter as recommended by CCAMLR. Updated analysis shown that mean krill biomass density was 80.7 g m^-2 with a CV of 13.6% in the survey area. However, the krill biomass estimates might be updated pending on the outcome of krill length clustering analysis and discussion on the methodology during this SG-ASAM meeting.

To address the potential effect of DVM behavior on krill biomass estimates, diel vertical distribution of krill swarms around the South Shetland Islands in Subarea 48.1 were analyzed based on data collected by the Chinese fishing vessel participating in the international 2019 Area 48 survey. Results show that more krill swarms were found in shallower depth at night than during daytime, while 16.5 % compared to 5.7 % of the corresponding total NASC value (m² n.mile^-2) integrated for layers bellow 15 m were found above 15 m at night and during daytime, respectively. The results implies that, when 15m is used for upper integration layer limit, both daytime survey and nighttime survey are likely underestimate the krill density due to the shallow distribution pattern of the animal, which may affect the nighttime estimate more than the daytime estimate.

To address the effect of various identification methods on krill biomass estimates, comparison on the swarm-based and dB-difference (Sv_120kHz-Sv_38kHz) identification method were made using the acoustic data collected around the South Shetland Islands in Subarea 48.1 by the Chinese fishing vessel F/V Fu Rong Hai during the 2019 Area 48 survey. Identification window of swarm-based template were set to four ranges including [-20 20], [2 16], [0.4 12] and without dB window (only 120kHz used). Identification window of dB-difference method were set to [2 16] and [0.4 12] respectively. Distributions of estimated krill densities were statistically compared using significance test and geometric linear regression. Pairwise comparison indicated that there were significant differences among biomass estimated from various identification methods, except for the estimates from [-20 20] identification window and without dB window (only 120kHz used). Geometric linear regression analysis also confirmed the same conclusion.

Biomass (B₀) of Antarctic krill for CCAMLR Division 58.4.1 in 2018/19 was preliminarily estimated as 4.349 million ton (CV=54.8 %) based on the swam based method using data obtained by Kaiyo-maru. The point estimate was comparable with the estimate in 1996 by BROKE (4.83 million ton with CV=17 %). However, they can not be compared directly because (1) biomass estimation methods were different, (2) timing of the surveys were different (Kaiyo-maru survey commenced about 40 days earlier than BROKE) and (3) areal coverages were different primary because of difference of positions of sea ice edge especially in the western part of the Division. The CV of biomass estimate of Kaiyo-maru survey was high primary because the spatial distribution of biomass densities was highly heterogeneous. The swarm based method potentially includes non-krill echoes but also excluding aggregations of krill which are smaller than size of swarm set in the Echoview template used in this analysis. Comparison with B₀ estimated by the dB difference method is necessary before producing finial B0 in Division 58.4.1 in 2018/19.

We compare supervised, using the interactive software Echoview™, and unsupervised, using python,  processing  of acoustic  data  collected  during  the  UK  cruise  DY098.  The  same  data processing procedures are applied in both methods, and we show that the unsupervised method comes  up  with  a  similar estimate  of krill  NASC  and  krill  density  as  the  supervised  method.  An estimate of krill density is supplied for the WCB survey using three methods. The multifrequency method produces a krill density of 8.60 gm^-2, compared with 21.76 gm^-2 (supervised) or 20.41 gm^-2 (unsupervised).

SG-ASAM are invited to consider provide further recommendations for improving this fishing vessel data collection program and provide advice on

• the timing and frequency for conducting acoustic transects;
• the methods for the data transmission to the Secretariat including the use of EchoExplore to provide a meta-data catalogue of the data and
• whether to have acoustic data transferred to the Secretariat as complete (raw) acoustic data files for processing and/or analysis products (e.g. from the EchoViewR protocol agreed at SG-ASAM).