Pobie Bank Reef Still Imagery Analysis 0923S

Project Aims JNCC wishes to commission a contract to undertake the analysis of seabed still imagery collected on the 0923S survey from a drop-frame camera. Seabed imagery will be analysed using the image annotation software BIIGLE. During 0923S imagery data was collected for both MPA monitoring, and ground truthing purposes. The total number of sampled drop camera monitoring stations was 97 and the total number of images after filtering to ensure 40 good quality images per station is 3,880. Analysis of 10 groundtruthing stations with 383 images is to be included as a costed extra only, however the total cost including costed extras must not exceed the stated value of the contract. More detailed metadata will be provided including information from logs and the survey report. Representative images from each station from the 0923S survey are shown in Annex C, and the successful contractor will be provided with access to the 2023 BIIGLE project, the annotated 2023 PBR video imagery and the 2020 PBR seabed imagery to aid with identification. 3. Project Background JNCC and Marine Directorate (MD) conducted a survey aboard MRV Scotia (0923S) in 2023 to gather evidence to monitor Pobie Bank Reef SAC (PBR) (Figure 1) and inform assessment of the extent and condition of the designated features of the site. This is the second monitoring survey in a repeat timeseries at PBR, three years after the first survey in 2020, and aims to generate comparable results to make long-term conclusions about the condition of the site. PBR is located in the North Sea approximately 20 km east of the Shetland Islands. The site has an area of 966 km2 and a depth ranging of 70 m to 100 m. The designated features of PBR are listed in Table 1. More information on PBR can be found in the JNCC SIC https://jncc.gov.uk/our-work/pobie-bank-reef-mpa/. Table 1: Designated features of Pobie Bank Reef SAC. Feature Feature Type 1170 Reefs Annex I Habitat • Produce a final analysis report including, at a minimum, sections detailing the methodology, results , and details of all QA work undertaken with any remedial action deemed necessary. The report should be no longer than 15,000 words including all tables and appendices and be provided electronically via email as a Microsoft Word document and PDF. • Provide substrate and taxonomic image reference collections for each substrate type and taxon identified from imagery. • (Costed extra) Undertake the analysis as set out for the monitoring stations for an additional 383 still images from the 10 groundtruthing stations. • (Costed extra) Create a subset of 100 still images annotated in a way suitable to act as training data for artificial intelligence model training as described below Some information, where specified, may be recorded directly into the proformas provided. The majority will be recorded first into BIIGLE and then used to populate the proforma. No analysis additional to what is described in this document is required. Any deviation from this methodology should be approved in writing by the project officer. 4.1 Stills Processing Stills should be analysed in BIIGLE using the label trees shown in Table 2. Annotations should be added to stills as either tier 1 or tier 2 labels depending on the label tree used. Optionally tier 1 labels may be added directly into the proforma instead of being applied to still images, where this is done it must be consistent (i.e. there should not be a mixture of BIIGLE annotations and proforma entries for JNCC image quality). More details on the still annotation types and how they should be applied are provided in Appendix A. Stills will be analysed to extract the following information: 1. Each still will be assigned an Image quality using labels from two label trees. If a still image is given a score of zero (NMBAQC equivalent), no further analysis should be carried out for that image and the contractor should alert JNCC to replace the image with one of suitable quality: a. JNCC image quality, a summary of these labels is shown in Table 3. b. NMBAQC image quality, a summary of these scores is shown in Table 4 and described in more detail in section 3.1 of Turner et al (2016). 2. Annotate each still using the Biota label tree. This tree is pre-populated with a list of taxa identified in 2020 and new species should be added. For each still image, taxa should be identified and quantified using the standardised approach detailed in Appendix B. Where a previously unidentified taxon is observed, the following approaches should be used: a. Solitary and/or erect epifaunal species present (to the lowest taxonomic level for which identification can be guaranteed) using counts (achieved by using the point annotation tool within BIIGLE). b. Colonial and/or encrusting epifaunal species present as far as possible (to the lowest taxonomic level for which identification can be guaranteed) using percentage cover (achieved by using the polygon, magic wand, magic SAM or brush tools within BIIGLE). c. Special cases, such as where a taxon is present in high densities (i.e. Ophiothrix fragilis, Caryophyllia smithii, faunal turf), may be recorded directly into the tier 1 spreadsheet as an estimate of count, or estimate of percentage cover to the nearest 0.5%. These cases are identified in Appendix B. 3. Use laser points label tree to annotate lasers and calculate image area in BIIGLE. Analysts are not to use the automated laser detection tool in BIIGLE. 4. Identify evidence of anthropogenic impacts on the seabed: a. Use the litter label tree to record the presence of litter using the categories listed in Annex 5.1 of the Joint Research Centres Guidance on Monitoring of Marine Litter in European Seas . b. Use the Anthropogenic label tree to annotate trawl marks or anthropogenic impacts other than litter. This will not be a complete label tree and new labels may need to be added to the label tree. 5. Use the substrate label tree to record the substrate types present to the nearest 0.5 percent directly to the proforma a. 10 x 10 superimposed grids should be used to aid estimates of percentage cover 6. Any comments from the analyst which are considered of importance can be recorded in the spreadsheet A reference collection of representative images must be provided for each Biota label and Substrate identified. Table 3. JNCC image quality categories • Imagery quality level • Description • Fauna • Most fauna can be identified (e.g. including smaller taxa such as brittlestars etc.) • Conspicuous fauna • Large and conspicuous fauna can be identified (e.g. sponges, soft corals etc.) • Substrate • The substrate type can be identified, but the fauna cannot (e.g. the water column is obscured / the camera is too high off the seabed) • Zero • No visibility of the seabed, substrate cannot be reliably identified. 4.2 Quality Assurance Following award of the contract, and dependent on the evaluation scores they achieve, unsuccessful bidders may be invited to undertake the independent external QA aspects of this project. Please indicate in your tender if you do not wish to be considered for this work. Note external QA analysis may be undertaken at same time as primary analysis if preferred by successful contractor and suitable process is agreed with JNCC at start-up meeting. 4.2.1 Internal Quality Assurance process for image analysis A minimum of 10% of still images analysed are to be subject to an internal QA process. 4.2.2 External Quality Assurance process for image analysis A separate contract will be run by JNCC for the external QA process. It will be conducted as described in this section. 10% of still images analysed are to be subject to an independent external QA process. The following steps will be adhered to: • Still images to be quality assured will be selected by JNCC; • The auditing analysts receive the imagery data from JNCC; • The auditing analysts complete the still image analysis for selected images as described in section 4.1 and complete the auditing analyst section of the comparison proforma; • The comparison proforma is sent to JNCC and original analysts; • The original analysts fully complete the partially filled JNCC image analysis comparison proforma; • JNCC, the original and the auditing analysts meet to resolve any discrepancies in taxa identification and enumeration to the satisfaction of all parties. Any remedial action required following the QA process must be completed by the original analysts within an agreed timeframe at no extra cost to the project. It is the responsibility of the successful contractor to ensure this process is completed before the final delivery of the dataset. All remedial action taken, and associated rationale, are to be agreed with JNCC before being implemented, recorded and provided as outputs of this contract. All remedial action taken and associated rationale are to be recorded and provided as outputs of this contract. The contract shall not be deemed complete until this external Quality Assurance step has been completed successfully.