{"success":true,"data":{"author":"AODN","bucketAccessType":"public","bucketListable":true,"bucketName":"aodn-cloud-optimised","bucketPrefix":"diver_photoquadrat_score_qc.parquet","category":"Life Sciences","cloudProvider":"aws","createdAt":"2026-10-04T00:00:00.000Z","description":"The National Reef Monitoring Network (NRMN) brings together shallow reef surveys conducted around Australia into a centralised database; the IMOS NRMN Sub-Facility collates, cleans, stores and makes this data available from Reef Life Survey, Parks Australia, the IMOS Understanding of Marine Imagery Sub-Facility, DBCA (WA), DEW (SA), DPIRD (NSW), Tasmanian Parks and Wildlife Service, Parks Victoria and IMAS. This dataset contains habitat composition scored from photoquadrats taken by RLS and ATRC divers along 50 m transects on shallow rocky and coral reefs: 20 images per transect at 2.5 m intervals, annotated with a point-overlay approach (usually 5 points per image) using standardised label schemes, either full-biodiversity or targeted (e.g. habitat-forming corals only). Values are percent cover per 50 m transect-analysis, keyed by dataset_id. Squidle+ is the image and annotation platform; this collection supersedes the retired RLS 'Habitat Quadrats' collection. This store is the quality-controlled percent-cover table as hive-partitioned parquet: 120,561 rows, 10,807 transect analyses, 2,892 sites in 47 countries, 2006-09 to 2026-03, partitioned by survey year and 20-degree cell.","displayTitle":"National Reef Monitoring Network - Global benthic cover data (photoquadrat image annotations)","fuseMountBrowsable":true,"id":"11a1891b-d0d0-5dcc-96de-ae6129582fc6","license":"http://creativecommons.org/licenses/by/4.0/","objectCountEstimate":21,"probeStatus":"ok","region":"ap-southeast-2","schemaCard":{"accessHints":[{"snippet":"# Task: root listing and the newest year directory (do not --recursive the root in a loop; 252 objects is fine once)\naws s3 ls --no-sign-request --region ap-southeast-2 s3://aodn-cloud-optimised/diver_photoquadrat_score_qc.parquet/\naws s3 ls --no-sign-request --region ap-southeast-2 s3://aodn-cloud-optimised/diver_photoquadrat_score_qc.parquet/timestamp=1767225600/ --recursive","language":"bash"},{"language":"python-anonymous","snippet":"# Task: read the schema and dataset identity without touching a data file (_common_metadata at the root)\nimport pyarrow.parquet as pq, s3fs, ast\nfs = s3fs.S3FileSystem(anon=True, client_kwargs={'region_name': 'ap-southeast-2'})\nfor store in ['diver_benthic_cover_in_situ_qc', 'diver_photoquadrat_score_qc', 'diver_photoquadrat_qc_DwC', 'diver_site_information_qc']:\n    schema = pq.read_schema(f'aodn-cloud-optimised/{store}.parquet/_common_metadata', filesystem=fs)\n    meta = {k.decode(): v.decode() for k, v in schema.metadata.items()}\n    print(store, '|', len(schema), 'columns |', ast.literal_eval(meta['global_attributes']))\n    print('   ', ast.literal_eval(meta.get('latitude') or meta.get('decimalLatitude')))"},{"snippet":"# Task: from the mount, read 2008 onward with a partition filter and get SA percent cover per transect\nimport pandas as pd\ndf = pd.read_parquet('/mnt/archive', filters=[('timestamp', '>=', 1199145600)])   # 2008-01-01 epoch; prunes files\nsa = df[df['area'] == 'South Australia']\nprint(len(df), 'rows; SA', len(sa), 'transects', sa['dataset_id'].nunique(), 'latest SA', sa['TIME'].max())\n# percent_cover sums to 100 per dataset_id; a label absent from a transect means 0 cover\nwide = sa.pivot_table(index='dataset_id', columns='rls_label', values='percent_cover', aggfunc='sum', fill_value=0)\nprint(wide.mean().nlargest(8).round(1))","language":"python-local"},{"language":"python-anonymous","snippet":"# Task: SA photoquadrat percent cover: mean cover per transect of the main habitat labels, by year (anonymous S3)\nimport time, pandas as pd, pyarrow.dataset as ds, s3fs\nfs = s3fs.S3FileSystem(anon=True, client_kwargs={'region_name': 'ap-southeast-2'})\nSA_CELL = ('010300000001000000050000000000000000005E4000000000000044C0000000000080614000000000000044C0'\n           '000000000080614000000000000034C00000000000005E4000000000000034C00000000000005E4000000000000044C0')\nt0 = time.time()\ndset = ds.dataset('aodn-cloud-optimised/diver_photoquadrat_score_qc.parquet', filesystem=fs, format='parquet', partitioning='hive')\ndf = dset.to_table(filter=ds.field('polygon') == SA_CELL).to_pandas()\nsa = df[df['area'] == 'South Australia'].copy()\nprint(f\"{len(df)} rows in the 120-140E cell in {time.time()-t0:.1f}s; SA rows {len(sa)}, transects (dataset_id) {sa['dataset_id'].nunique()}, \"\n      f\"sites {sa['site_code'].nunique()}, {sa['TIME'].min().date()} to {sa['TIME'].max().date()}, since 2025-03-01: {int((sa['TIME'] >= '2025-03-01').sum())}\")\n# one dataset_id = one 50 m transect analysis whose percent_cover rows sum to 100; absent label = 0 cover\nwide = sa.pivot_table(index='dataset_id', columns='rls_label', values='percent_cover', aggfunc='sum', fill_value=0)\nwide['year'] = sa.groupby('dataset_id')['TIME'].first().dt.year\ntop = wide.drop(columns='year').sum().nlargest(6).index\nprint(wide.groupby('year')[list(top)].mean().round(1).to_string())"},{"snippet":"# Task: decode a polygon partition key to its lon/lat box without shapely (little-endian WKB Polygon, 1 ring, 5 points)\nimport struct, s3fs\nfs = s3fs.S3FileSystem(anon=True, client_kwargs={'region_name': 'ap-southeast-2'})\nroot = 'aodn-cloud-optimised/diver_photoquadrat_score_qc.parquet'\ndef wkb_box(h):\n    b = bytes.fromhex(h); n = struct.unpack('<I', b[9:13])[0]\n    pts = [struct.unpack('<dd', b[13 + 16 * i: 29 + 16 * i]) for i in range(n)]\n    return min(p[0] for p in pts), min(p[1] for p in pts), max(p[0] for p in pts), max(p[1] for p in pts)\nyear = fs.ls(f'{root}/timestamp=1735689600')          # 2025: one directory per 20-degree cell\nfor p in year:\n    print(p.rsplit('/', 1)[1][:24] + '...', '-> (minlon, minlat, maxlon, maxlat)', wkb_box(p.split('polygon=')[1]))","language":"python-anonymous"}],"caveats":["Scale: the SA-gulfs-since-2025-03 subset is EMPTY. South Australia has 4,699 rows, 521 transect analyses, 76 sites (Encounter / Fleurieu 3,669 rows, Adelaide 570, Kangaroo Island 282, SA - Western 91, Yorke Peninsula 87) from 2008-02-05 to 2022-02-08, with the bulk in 2019-2020; the last SA scores are Wallaroo / Point Hughes (Yorke Peninsula, DEW RLS schema). diver_survey_metadata_qc lists 353 SA surveys since 2024 (latest 2026-05-29) with has_pq_scores_in_db == False for all of them: the photos may exist in Squidle+ but are not scored here yet.","DO NOT: average percent_cover across rows as if rows were samples. The sampling unit is dataset_id (one 50 m transect analysis); percent_cover sums to exactly 100 within each dataset_id and a label that was not seen has NO row. Pivot to dataset_id x rls_label with fill_value=0 first, then average transects. 1 to 12 dataset_ids share a site+date (median 2), typically several transects or several label schemes of the same dive.","DO NOT: mix annotation_method values. 'targeted coral' (13,170 rows, 11%) labels only target taxa, so its 100% is 100% of coral points, not of the seabed; for whole-reef composition keep annotation_method == 'full biodiversity' (all 4,699 SA rows are).","Quirk: 15 label_scheme values coexist and the same reef is labelled at different granularity ('RLS Basic' vs 'RLS Catalogue' vs 'DEW RLS schema' for SA DEW surveys). rls_label is the harmonised layer and catami_label / catami_lineage the CATAMI crosswalk; compare across schemes at the rls_lineage or catami level, not by raw label strings.","Quirk: label_count, dataset_total_label_count and visibility are STRINGS; cast with pd.to_numeric(errors='coerce'). label_count/dataset_total_label_count are null for the 'RLS Basic (Aggregate Obs)' scheme (7,056 rows). database == 'NRMN' rows (46,281) have no sq_annotation_set_id and cannot be traced back to Squidle+ images.","Quirk: this is a GLOBAL table (47 countries, 76 polygon cells, lon -179 to 177); area == 'South Australia' is the state label and 91 of its rows (SA - Western, lon 134.1-134.4) fall outside the gulfs bbox; the bbox contains no non-SA rows. All SA rows are in the 120E-140E / 40S-20S polygon partition.","Quirk: the companion diver_photoquadrat_qc_DwC store is a Darwin Core re-expression of a SUBSET of this table (97,922 of 120,561 rows: every row whose dataset_id is an eventID there; western-hemisphere rows and some schemes are excluded). eventID == dataset_id, RLSIdentification == rls_label, individualCount == label_count; percent_cover is not carried there.","Scale: 251 files of ~12 KB; latency-bound. Full table 31 s anonymous into 60 MB pandas; one year 1.2 s; the SA cell 16-20 s. On the FUSE mount the same per-file cost applies.","Quirk: hive keys are strings in the path. timestamp=<epoch seconds> is the 1 January 00:00 UTC of the row's survey year (34/34 benthic, 21/21 score, 19/19 DwC partitions decoded to Jan-1; TIME.year == partition year for every row), so it is a YEAR partition, not a month. polygon=<WKB hex> is a little-endian WKB Polygon (0103..., 1 ring, 5 points) of a 20 x 20 degree lon/lat cell; decode with struct (or shapely.wkb.loads(bytes.fromhex(h)) where shapely exists).","Quirk: pyarrow's hive discovery types `timestamp` as int64 and `polygon` as string; pandas.read_parquet returns both as category columns. Filter on them with ds.field('timestamp') >= <epoch> and ds.field('polygon') == '<hex>' to prune files; filters on data columns (area, latitude) read every file."],"columns":[{"name":"TIME","type":"timestamp[ns]; 'Derived timestamp' = date plus hour when known; 2006-09-02 to 2026-03-12; no nulls"},{"name":"area","type":"string; state/territory or overseas region, 120 distinct (Queensland 27,374 | NSW 21,498 | WA 19,473 | Victoria 9,554 | South Australia 4,699 | Tasmania 4,673 | ... Galapagos, West Papua, Canary Islands)"},{"type":"string YYYY-MM-DD local date","name":"date"},{"type":"string HH:mm:ss local; null 8.4%","name":"hour"},{"type":"string; annotation tags: '' (most), 'dead', 'bleached', 'bleached|dead'; never null","name":"tags"},{"type":"float32 m","name":"depth"},{"type":"string; 47 countries, Australia 95,131 rows","name":"country"},{"name":"polygon","type":"string (hive key); WKB hex of a 20 x 20 degree cell; 76 cells in use worldwide; all SA rows in the 120E-140E / 40S-20S cell"},{"type":"string; single value 'RLS'","name":"program"},{"name":"database","type":"string; 'Squidle+' 74,280 (scored in Squidle+, sq_annotation_set_id set) | 'NRMN' 46,281 (legacy scores held in the NRMN db, sq_annotation_set_id null)"},{"name":"filename","type":"string; single value 'ep_m13_pq_scores.csv'"},{"type":"float32 degrees_north; site coordinate, 2 dp; -55.05 to 78.45","name":"latitude"},{"type":"string; 179 regions ('Encounter', 'Adelaide', 'Kangaroo Island', 'Yorke Peninsula', 'SA - Western' for SA)","name":"location"},{"name":"longitude","type":"float32 degrees_east; -179.14 to 177.18"},{"name":"rls_label","type":"string; RLS standardised habitat label, 77 distinct (Turfing algae (<2 cm high algal/sediment mat on rock), Crustose coralline algae, Sand, Encrusting corals, Massive corals, Ecklonia radiata, Other fucoids ...)"},{"type":"string; 2,892 distinct; 99.1% present in diver_site_information_qc","name":"site_code"},{"type":"string; 2,875 distinct","name":"site_name"},{"type":"int64 (hive key); epoch seconds of 1 Jan of the survey year, 2006..2026","name":"timestamp"},{"type":"string; 10,807 distinct; one 50 m transect-analysis, e.g. '912352572_53_9535' = <survey_id>_<label_scheme id>_<annotation set>; the unit over which percent_cover sums to 100; 1 to 38 rows each, median 10; one site per dataset_id, 1 to 12 dataset_ids per site+date (median 2)","name":"dataset_id"},{"name":"visibility","type":"string m; null 22.7%"},{"type":"string (not int); annotation points carrying this label in the dataset; null 5.9% (RLS Basic (Aggregate Obs) scheme)","name":"label_count"},{"type":"string; RLS hierarchy path 'Biota > Macroalgae > Macroalgae_canopy forming > Other fucoids'; null 1.1%","name":"rls_lineage"},{"name":"catami_label","type":"string; CATAMI leaf label, 65 distinct"},{"name":"label_scheme","type":"string; 15 schemes: RLS Australian Coral Species List 41,509 | RLS Catalogue 25,510 | RLS/CATAMI 17,280 | RLS Basic (Aggregate Obs) 12,686 | RLS Basic (Transparency) 6,572 | RLS Basic 4,300 | Reef Repair Hub 3,106 | RLS/Lord Howe Island 2,334 | RLS/Squidle 2,163 | DEW RLS schema 1,909 (all SA) | ..."},{"name":"percent_cover","type":"float32 %; label's share of the transect's annotation points; sums to exactly 100.0 within every dataset_id (10,807/10,807)"},{"name":"catami_lineage","type":"string; CATAMI path with code prefix, e.g. '1.1 Biota > Cnidaria > Corals > Stony corals > Encrusting'"},{"name":"annotation_method","type":"string; 'full biodiversity' 107,391 | 'targeted coral' 13,170 (target taxa only: the 100% is of coral points, not of the reef)"},{"type":"string; Squidle+ annotation set id; null for all database=='NRMN' rows (46,281)","name":"sq_annotation_set_id"},{"type":"string (not int); all annotation points in the dataset_id; null 5.9%","name":"dataset_total_label_count"}],"format":"v2-hybrid","generatedAt":"2026-10-04T00:00:00Z","layout":{"pattern":"<prefix>/_common_metadata (17.6 KB) and <prefix>/timestamp=<epoch s of 1 Jan>/polygon=<WKB hex 20x20 deg cell>/ep_m13_pq_scores.csv-0.parquet. 21 year partitions (2006..2026) x 76 cells worldwide = 251 files, one per directory, 8,356 B min / 12,450 B median / 49,031 B max. No _metadata."},"mountPath":"/mnt/archive","overview":"Reef Life Survey (RLS) photoquadrat habitat scores as served by the IMOS National Reef Monitoring Network: the global RLS 'method 13' benthic cover table, exported from the NRMN endpoint ep_m13_pq_scores.csv. Each row is one habitat label's percent cover on one 50 m transect analysis (dataset_id), derived from point annotations of 20 photoquadrats in Squidle+ (or, for 38% of rows, legacy scores held in the NRMN database). Measured 2026-10-04: 120,561 rows x 29 columns, 10,807 dataset_ids, 2,892 sites, 47 countries (Australia 79% of rows), 2006-09-02 to 2026-03-12; 15 label schemes and two annotation methods. On disk: 252 objects, 3.81 MB, _common_metadata plus 251 small parquet files (8-49 KB, median 12 KB) in timestamp=<1 Jan epoch>/polygon=<20-degree cell>/ directories, 21 years x 76 cells. Because files are tiny and numerous, cost is per-request latency, not bytes: one file 0.9 s, one year 1.2 s, the whole table 31 s into 60 MB of pandas, the South Australian cell (all SA rows) 4,699 rows in 16-20 s. The natural access direction is one polygon cell or one year, then group by dataset_id. SA holds 76 sites but data end 2022-02-08, so this is pre-bloom kelp/fucoid baseline, not bloom monitoring.","rowCount":"120,561 rows (exact, full read 2026-10-04)","sampleHeaders":[{"key":"diver_photoquadrat_score_qc.parquet/timestamp=1136073600/polygon=0103000000010000000500000000000000008056C000000000000024C000000000008051C000000000000024C000000000008051C0000000000000244000000000008056C0000000000000244000000000008056C000000000000024C0/ep_m13_pq_scores.csv-0.parquet","truncatedAt":512,"bytes":"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","encoding":"hex"}],"sampleObjects":["_common_metadata","timestamp=1199145600/polygon=010300000001000000050000000000000000005E4000000000000044C0000000000080614000000000000044C0000000000080614000000000000034C00000000000005E4000000000000034C00000000000005E4000000000000044C0/ep_m13_pq_scores.csv-0.parquet","timestamp=1735689600/polygon=010300000001000000050000000000000000005E4000000000000044C0000000000080614000000000000044C0000000000080614000000000000034C00000000000005E4000000000000034C00000000000005E4000000000000044C0/ep_m13_pq_scores.csv-0.parquet","timestamp=1293840000/polygon=0103000000010000000500000000000000000034C000000000000024400000000000000000000000000000244000000000000000000000000000003E4000000000000034C00000000000003E4000000000000034C00000000000002440/ep_m13_pq_scores.csv-0.parquet"],"sampleRows":[{"depth":4,"date":"2008-02-05","catami_label":"Rock","tags":"","TIME":"2008-02-05 00:00:00","location":"Encounter","latitude":-35.51,"site_code":"GSV95","annotation_method":"full biodiversity","database":"Squidle+","dataset_id":"5000349_8_3901","rls_lineage":"Physical > Substrate > Bare Rock","sq_annotation_set_id":"3901","site_name":"Lasseters Reef","label_scheme":"RLS Catalogue","catami_lineage":"2 Physical > Substrate > Consolidated (hard) > Rock","percent_cover":0.6849315,"area":"South Australia","longitude":138.22,"program":"RLS","timestamp":1199145600,"dataset_total_label_count":"146","rls_label":"Bare Rock","label_count":"1"},{"rls_label":"Other fucoids","timestamp":1546300800,"catami_label":"Large canopy-forming","date":"2019-02-18","dataset_id":"912352572_53_9535","database":"Squidle+","label_count":"22","TIME":"2019-02-18 11:00:00","catami_lineage":"1.1 Biota > Macroalgae > Large canopy-forming","label_scheme":"DEW RLS schema","area":"South Australia","rls_lineage":"Biota > Macroalgae > Macroalgae_canopy forming > Other fucoids","location":"Encounter","hour":"11:00:00","annotation_method":"full biodiversity","sq_annotation_set_id":"9535","site_name":"2nd Valley Boat Shed","dataset_total_label_count":"115","percent_cover":19.130434,"visibility":"10","latitude":-35.51,"longitude":138.21,"site_code":"GSV26","depth":5.2,"program":"RLS"},{"location":"Yorke Peninsula","TIME":"2022-02-08 14:00:00","hour":"14:00:00","database":"Squidle+","site_code":"SSG58","date":"2022-02-08","site_name":"Point Hughes","area":"South Australia","label_scheme":"DEW RLS schema","program":"RLS","annotation_method":"full biodiversity","timestamp":1640995200}],"sourceObjectKey":"diver_photoquadrat_score_qc.parquet/timestamp=1136073600/polygon=0103000000010000000500000000000000008056C000000000000024C000000000008051C000000000000024C000000000008051C0000000000000244000000000008056C0000000000000244000000000008056C000000000000024C0/ep_m13_pq_scores.csv-0.parquet","starterCell":"import time, pandas as pd, pyarrow.dataset as ds, s3fs\n# NRMN photoquadrat scores: percent cover per RLS habitat label per 50 m transect (dataset_id), sums to 100 per transect\nfs = s3fs.S3FileSystem(anon=True, client_kwargs={'region_name': 'ap-southeast-2'})\nroot = 'aodn-cloud-optimised/diver_photoquadrat_score_qc.parquet'   # in the pod: '/mnt/archive' with no filesystem=\nSA_CELL = ('010300000001000000050000000000000000005E4000000000000044C0000000000080614000000000000044C0'\n           '000000000080614000000000000034C00000000000005E4000000000000034C00000000000005E4000000000000044C0')\nt0 = time.time()\ndf = ds.dataset(root, filesystem=fs, format='parquet', partitioning='hive').to_table(filter=ds.field('polygon') == SA_CELL).to_pandas()\nsa = df[df['area'] == 'South Australia']\nprint(f\"SA rows {len(sa)} ({sa['dataset_id'].nunique()} transects, {sa['site_code'].nunique()} sites) in {time.time()-t0:.1f}s; \"\n      f\"{sa['TIME'].min().date()} to {sa['TIME'].max().date()}; rows since 2025-03-01: {int((sa['TIME'] >= '2025-03-01').sum())}\")\nlatest = sa[sa['TIME'] >= '2020-01-01']\nkelp = latest[latest['rls_label'].isin(['Ecklonia radiata', 'Other fucoids'])]\nprint(kelp.groupby(['location', 'site_name', 'rls_label'])['percent_cover'].mean().round(1).unstack(fill_value=0).to_string())","starterSamplePath":"timestamp=1136073600/polygon=0103000000010000000500000000000000008056C000000000000024C000000000008051C000000000000024C000000000008051C0000000000000244000000000008056C0000000000000244000000000008056C000000000000024C0/ep_m13_pq_scores.csv-0.parquet","totalObjects":"252 objects, 3,813,959 bytes (aws s3 ls --summarize 2026-10-04); 251 data files + _common_metadata","verifiedRead":{"attempts":[{"family":"parquet","ok":true,"path":"timestamp=1767225600/polygon=01030000000100000005000000000000000040604000000000000049C00000000000C0624000000000000049C00000000000C062400000000000003EC000000000004060400000000000003EC0000000000040604000000000000049C0/ep_m13_pq_scores.csv-0.parquet","seconds":1.5}],"bytes":15308,"facts":["under 'diver_photoquadrat_score_qc.parquet/': 21 directories (e.g. timestamp=1136073600, timestamp=1167609600, timestamp=1199145600, timestamp=1230768000); 1 objects named like _common_metadata, 17 KB to 17 KB","under 'diver_photoquadrat_score_qc.parquet/timestamp=1767225600/': 1 directories (e.g. polygon=01030000000100000005000000000000000040604000000000000049C00000000000C0624000000000000049C00000000000C062400000000000003EC000000000004060400000000000003EC0000000000040604000000000000049C0)","under 'diver_photoquadrat_score_qc.parquet/timestamp=1767225600/polygon=01030000000100000005000000000000000040604000000000000049C00000000000C0624000000000000049C00000000000C062400000000000003EC000000000004060400000000000003EC0000000000040604000000000000049C0/': 1 objects named like ep_m##_pq_scores.csv-#.parquet, 15 KB to 15 KB"],"family":"parquet","image":"531775725781.dkr.ecr.us-east-1.amazonaws.com/scigantic-notebook-genomics:latest","library":"pyarrow","path":"timestamp=1767225600/polygon=01030000000100000005000000000000000040604000000000000049C00000000000C0624000000000000049C00000000000C062400000000000003EC000000000004060400000000000003EC0000000000040604000000000000049C0/ep_m13_pq_scores.csv-0.parquet","rssDeltaMb":0.4,"seconds":1.5,"snippet":"import os\nimport pyarrow.parquet as pq\npf = pq.ParquetFile(os.path.join(os.environ['SCIGANTIC_MOUNT_PATH'], 'timestamp=1767225600/polygon=01030000000100000005000000000000000040604000000000000049C00000000000C0624000000000000049C00000000000C062400000000000003EC000000000004060400000000000003EC0000000000040604000000000000049C0/ep_m13_pq_scores.csv-0.parquet'))\nprint(pf.metadata)\ndf = pf.read_row_group(0).to_pandas()\ndf.head()","starterCell":{"ok":true,"seconds":4.58},"summary":"Parquet 454 rows, 27 columns, 1 row groups; columns ['TIME:timestamp[ns]', 'area:string', 'date:string', 'hour:string', 'tags:string', 'depth:float', 'country:string', 'program:string', 'database:string', 'filename:string']","verifiedAt":"2026-11-04T06:45:53.911Z"}},"summary":"The National Reef Monitoring Network (NRMN) brings together shallow reef surveys conducted around Australia into a centralised database; the IMOS NRMN Sub-Facility collates,…","title":"National Reef Monitoring Network - Global benthic cover data (photoquadrat image annotations)","updatedAt":"2026-11-04T06:45:53.911Z","visibility":"public","websiteUrl":"https://catalogue-imos.aodn.org.au/geonetwork/srv/eng/catalog.search#/metadata/0a65be6d-1c76-49ac-a151-80acf123612c"}}