Ice13Ball/biodiversity-hotspots

🌊 Deep-Sea Biodiversity Hotspots Dataset

Curated dataset of 3,253,527 georeferenced deep-sea species occurrence records aggregated from OBIS, GBIF, and NOAA Deep-Sea Coral. Focused on chemosynthetic ecosystems (hydrothermal vents, cold seeps) and abyssal biodiversity below 200m depth.

Updated daily via automated pipeline. Attribution: something-rare.com

---

📊 Dataset Stats

Field	Value
Total records	3,253,527
Unique species	86,080
Depth range	200m – 11,000m+
Date range	1767 – 2026
Coverage	Global ocean
Format	GeoJSON + CSV
License	CC BY 4.0

Ecosystem Breakdown

Ecosystem	Records	%
Slope (200–2000m)	2,870,539	88.2%
Abyssal plain (2000–6000m)	322,851	9.9%
Seamount	48,918	1.5%
Trench (>6000m)	6,522	0.2%
Hydrothermal vent	4,158	0.1%
Cold seep	539	<0.1%

---

📁 Files

File	Description	Size
biodiversity_hotspots_1.0.0_part001.geojson … part066.geojson	GeoJSON FeatureCollection chunks (100k records each)	~2–5MB each
biodiversity_hotspots_1.0.0.csv	Full flat CSV, all records	524MB
biodiversity_hotspots_sample_1.0.0.csv	Random 1,000-record sample for preview	165KB
README_1.0.0.md	Release notes for v1.0.0	1.3KB

---

🗂️ Schema

Field	Type	Description
scientificName	string	Accepted scientific name (genus + species)
latitude	float	WGS84 decimal degrees
longitude	float	WGS84 decimal degrees
depth_m	float	Depth in meters (positive = below surface)
eventDate	string	ISO 8601 observation date (YYYY-MM-DD)
ecosystem	string	vent / seep / abyssal_plain / seamount / trench / slope
is_chemosynthetic	bool	True for vent and seep ecosystems
data_quality_score	int	0–100 quality indicator
datasetSource	string	obis / gbif / noaa
taxonRank	string	Taxonomic rank of the record
basisOfRecord	string	How the record was collected
occurrenceID	string	Unique ID from source database
locality	string	Textual location description
countryCode	string	ISO 3166-1 alpha-2 country code
kingdom	string	Taxonomic kingdom

---

🚀 Usage

Load sample (fast preview)

import pandas as pd
df = pd.read_csv("hf://datasets/Ice13Ball/biodiversity-hotspots/biodiversity_hotspots_sample_1.0.0.csv")
print(df[["scientificName", "depth_m", "ecosystem", "latitude", "longitude"]].head())

Load full dataset with HuggingFace datasets

from datasets import load_dataset
ds = load_dataset("Ice13Ball/biodiversity-hotspots", data_files="biodiversity_hotspots_1.0.0.csv")

Load GeoJSON chunks

import json, glob
features = []
for path in sorted(glob.glob("biodiversity_hotspots_1.0.0_part*.geojson")):
    with open(path) as f:
        features.extend(json.load(f)["features"])
print(f"Loaded {len(features):,} records")

Filter by ecosystem

import pandas as pd
df = pd.read_csv("biodiversity_hotspots_1.0.0.csv")
vents = df[df["ecosystem"] == "vent"]
print(f"Hydrothermal vent records: {len(vents):,}")
print(vents["scientificName"].value_counts().head(10))

---

🔬 Data Sources

Source	Records	Notes
GBIF	3,253,527	Global Biodiversity Information Facility
OBIS	Coming daily	Ocean Biodiversity Information System
NOAA Deep-Sea Coral	Coming daily	NOAA coral and sponge occurrence data

---

🧹 Data Processing

Records are processed through a 4-stage automated pipeline:

1. Collector — fetches new records from APIs daily (OBIS, GBIF, NOAA)
2. Cleaner — deduplicates using spatial (0.01° tolerance) + temporal (monthly) + species matching, normalizes field names across sources
3. Enricher — classifies ecosystem type (keyword + depth-based), adds quality score, flags ISA contract zones
4. Publisher — packages versioned release, uploads to HuggingFace + Kaggle

Pipeline source: github.com/ice13ball/dataCollector

---

📜 License & Citation

License: CC BY 4.0 — free to use with attribution.

Citation:

something-rare.com Biodiversity Hotspots Dataset v1.0.0 (2026).
Aggregated from GBIF (https://gbif.org), OBIS (https://obis.org), NOAA Deep-Sea Coral.
https://huggingface.co/datasets/Ice13Ball/biodiversity-hotspots

---

🔄 Updates

This dataset is updated daily via automated pipeline. New records are added incrementally from all three sources. Version history is maintained — older releases remain available.

Premium version with ISA mining zone overlays, enriched taxonomy, and API access available at something-rare.com.