# Seagrass - Australian seagrass occurrence - Aggregated data product (1967 - ongoing)

## Facts

- Provider: AODN
- License: http://creativecommons.org/licenses/by-nc/4.0/
- Size: 43 objects, 247.5 MiB
- Records: 3,286,273 rows (pyarrow count_rows over all 42 partitions, 2026-10-04); 26 columns in the schema (24 in the files + hive keys timestamp and polygon)
- Format: Parquet (in sampled files)
- Stored at: Amazon S3, s3://aodn-cloud-optimised/aggregated_seagrass_nonqc.parquet (ap-southeast-2)
- Access: public, readable without an account
- Category: Life Sciences
- Source: https://catalogue-imos.aodn.org.au/geonetwork/srv/eng/catalog.search#/metadata/009a1131-efc1-4a61-8f90-cf289e7c043d
- Page: https://scigantic.com/archive/ed1ee487-90e3-5bce-a077-8f3323837918
- Keywords: Life Sciences, satellite imagery, remote sensing, time series, Parquet, AWS Open Data

## About

The Australian Seagrass Aggregated Data Product harmonises disparate seagrass datasets from providers across Australia into a single, standardised resource suitable for ecological assessment, long-term monitoring and habitat modelling. Repeat aerial and satellite remote sensing surveys provide comprehensive spatial coverage for regional and long-term mapping programs; many legacy datasets, including the NSW estuarine macrophytes and the Cockburn Sound historical time series, were derived from purpose-flown aerial photography digitised into habitat polygons with field validation, while recent datasets add UAV orthomosaics, high-resolution multispectral satellite imagery (WorldView-2, IKONOS, QuickBird-2) and PlanetScope imagery to model seagrass distribution, species composition and biomass. Source data were collated from Seamap Australia (IMAS, TropWATER/JCU, CSIRO, AIMS and state agencies such as SA DEWNR, NSW DPE, Vic DEECA and WA DWER) under NESP Marine and Coastal Hub Project 5.9. This store is the cloud-optimised Parquet copy: 3,286,273 long-format presence/absence rows for 29 seagrass taxa from 43 source datasets, 1967 to February 2025, one hive partition per year, 247.5 MiB, each row carrying a WKT footprint, its centroid, a WoRMS name and the source dataset name and metadata link; licensed CC-BY-NC 4.0 because one contributing dataset is non-commercial.

## Load the data

```python
import os
import pyarrow.parquet as pq
pf = pq.ParquetFile(os.path.join(os.environ['SCIGANTIC_MOUNT_PATH'], 'timestamp=978307200/polygon=01030000000100000005000000000000000000000000000000008056C0000000000080664000000000008056C00000000000806640000000000080564000000000000000000000000000805640000000000000000000000000008056C0/seagrass.parquet-0.parquet'))
print(pf.metadata)
df = pf.read_row_group(0).to_pandas()
df.head()
```

Run in a Scigantic notebook on 2026-10-05 against `timestamp=978307200/polygon=01030000000100000005000000000000000000000000000000008056C0000000000080664000000000008056C00000000000806640000000000080564000000000000000000000000000805640000000000000000000000000008056C0/seagrass.parquet-0.parquet`: Parquet 27,050 rows, 23 columns, 17 row groups; columns ['endDay:int8', 'endYear:int16', 'h3Index:string', 'endMonth:int8', 'sourceID:string', 'startDay:int8', 'startYear:int16', 'taxonRank:string', 'startMonth:int8', 'datasetName:string'].

From any machine:

```python
# Task: schema and GeoNetwork uuid from the sidecar, then an exact row count from the file footers (no column data read)
import pyarrow.parquet as pq, pyarrow.dataset as ds, s3fs
fs = s3fs.S3FileSystem(anon=True, client_kwargs={'region_name': 'ap-southeast-2'})
ROOT = 'aodn-cloud-optimised/aggregated_seagrass_nonqc.parquet'
with fs.open(f'{ROOT}/_common_metadata', 'rb') as f:
    schema = pq.read_schema(f)
print(schema.names)
print(schema.metadata[b'global_attributes'])
dset = ds.dataset(ROOT, filesystem=fs, format='parquet', partitioning='hive')
print('rows:', dset.count_rows(), 'files:', len(dset.files))   # 3,286,273 rows, 42 files; footers only
```

Lists the files with anonymous access. Runs on any machine, with no cloud account.

## Data structure

File types in sampled files: Parquet

| Field | Type |
| --- | --- |
| `datasetName` | string, not null; 43 source dataset titles, e.g. 'South Australia State Marine Benthic Habitats (DEWNR)', 'NSW… |
| `sourceID` | string, not null; '<source table>.<row>' e.g. 'SeamapAus_SA_state_benthic_habitats_2013.5646'; the primary sort key of… |
| `metadataLink` | string, not null; source record URL (IMAS/eAtlas GeoNetwork, data.wa.gov.au, doi.org) or the literal 'unknown' (6… |
| `startYear` | int16, not null; 1967 to 2025; equals the partition year on 100% of rows |
| `startMonth` | int8 (float64 in pandas because of nulls), NULL on 297,300 rows; 1-12 |
| `startDay` | int8, NULL on 2,437,001 rows (74%); most sources only know the month or year |
| `endYear / endMonth / endDay` | int16 / int8 / int8, filled on only 11,715 / 11,715 / 11,047 rows (multi-year surveys), 1997 to 2023 |
| `_temporal_extent` | date32[day] (pandas object of datetime.date), not null; startYear-(startMonth or 1)-(startDay or 1); 1967-01-01 to… |
| `footprintWKT` | string, not null; WKT in EPSG:4326: POINT (1,863,283 rows), POLYGON (1,389,443), MULTIPOLYGON (11,453); median 27… |
| `decimalLatitude` | float32 (NOT double), EPSG:4326, not null; centroid of footprintWKT; -40.5809 to -9.13895 |
| `decimalLongitude` | float32, not null; centroid; 114.99439 to 153.60187 |
| `h3Index` | string, not null; H3 cell at resolution 5 from the centroid, e.g. '85b914c7fffffff'; 1,746 distinct (the technical… |
| `australianMarineRegionsTags` | string, NULL on 358,535 rows; '\|'-separated IMCRA / Marine Park / World Heritage Area tags; SA rows carry… |
| `scientificName` | string, not null; 29 WoRMS/CAAB-aligned names: Tracheophyta (phylum-level 'seagrass'), 7 genus-level (Ruppia, Zostera… |
| `scientificNameID` | int64, not null; WoRMS AphiaID (e.g. 374714 Posidonia sinuosa), NOT an LSID string like the kelp sibling |
| `taxonRank` | string, not null; Species 2,027,699 / Genus 642,493 / 'Phylum (Division)' 616,081 |
| `taxonomicStatus` | string, not null; accepted 3,092,802 / unaccepted 193,471 (names such as Zostera capricorni and Heterozostera tasmanica… |
| `occurrenceStatus` | string, not null; 'absent' 2,862,901 / 'present' 423,372: every taxon the source scheme could report is written out per… |
| `organismQuantity` | float32 gdw/m^2 biomass, filled on 1,533 rows only (all from 'Seagrass meadows of Hervey Bay and the Great Sandy Strait… |
| `organismQuantityType` | string; the literal 'biomass' on ALL 3,286,273 rows even where organismQuantity is NULL |

and 3 more fields

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/ed1ee487-90e3-5bce-a077-8f3323837918
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/ed1ee487-90e3-5bce-a077-8f3323837918
- Over MCP: `get_schema_card` with id `ed1ee487-90e3-5bce-a077-8f3323837918` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Satellite imagery datasets](https://scigantic.com/topics/satellite-imagery)
- [Remote sensing datasets](https://scigantic.com/topics/remote-sensing)
- [Time series datasets](https://scigantic.com/topics/time-series)
- [Life Sciences datasets](https://scigantic.com/topics/life-sciences)
