electricsheepafrica/africa-mdg-ibtracs-tropical-storm-tracks

Madagascar: IBTrACS Storm Tracks

Publisher: HDX · Source: HDX · License: cc-by-igo · Updated: 2026-03-03

---

Abstract

The International Best Track Archive for Climate Stewardship (IBTrACS) project is the most complete global collection of tropical cyclones available. It merges recent and historical tropical cyclone data from multiple agencies to create a unified, publicly available, best-track dataset that improves inter-agency comparisons.

Fields available:
SID: A unique storm identifier (SID) assigned by IBTrACS algorithm.
ISO_TIME: Time of the observation in ISO format (YYYY-MM-DD hh:mm:ss)
BASIN: Basin of the current storm position
SUBBASIN: Sub-basin of the current storm position
NATURE: Type of storm (a combination of the various types from the available sources)
NUMBER: Number of the storm for the year (restarts at 1 for each year
LAT: Mean position - latitude (a combination of the available positions)
LON: Mean position - longitude (a combination of the available positions)
WMO_WIND: Maximum sustained wind speed assigned by the responsible WMO agency
WMO_PRES: Minimum central pressure assigned by the responsible WMO agency.

Each row in this dataset represents geolocated point observations. Temporal coverage is indicated by the iso_time column(s). Geographic scope: MDG.

Curated into ML-ready Parquet format by Electric Sheep Africa.

---

Dataset Characteristics

Domain	Climate and environment
Unit of observation	Geolocated point observations
Rows (total)	101,124
Columns	12 (3 numeric, 8 categorical, 1 datetime)
Train split	80,899 rows
Test split	20,224 rows
Geographic scope	MDG
Publisher	HDX
HDX last updated	2026-03-03

---

Variables

Geographic — iso_time, lat (range -47.4–8.7), lon (range 11.3–133.0).

Outcome / Measurement — number (range 1.0–145.0).

Identifier / Metadata — sid (1997013S08101, 2023036S12117, 2000032S11116), esa_source (HDX), esa_processed (2026-04-05).

Other — basin (South Indian, North India, ), subbasin (Missing, Western Australia, Arabian Sea), nature (Tropical, Not reported, Mixture(contradicting report from different agencies)), wmo_wind ( , 25, 20), wmo_pres ( , 1000, 997).

---

Quick Start

from datasets import load_dataset

ds    = load_dataset("electricsheepafrica/africa-mdg-ibtracs-tropical-storm-tracks")
train = ds["train"].to_pandas()
test  = ds["test"].to_pandas()

print(train.shape)
train.head()

---

Schema

Column	Type	Null %	Range / Sample Values
sid	object	0.0%	1997013S08101, 2023036S12117, 2000032S11116
number	float64	0.0%	1.0 – 145.0 (mean 27.7787)
basin	object	0.0%	South Indian, North India,
subbasin	object	0.0%	Missing, Western Australia, Arabian Sea
iso_time	datetime64[ns]	0.0%
nature	object	0.0%	Tropical, Not reported, Mixture(contradicting report from different agencies)
lat	float64	0.0%	-47.4 – 8.7 (mean -18.2439)
lon	float64	0.0%	11.3 – 133.0 (mean 59.6153)
wmo_wind	object	0.0%	, 25, 20
wmo_pres	object	0.0%	, 1000, 997
esa_source	object	0.0%	HDX
esa_processed	object	0.0%	2026-04-05

---

Numeric Summary

Column	Min	Max	Mean	Median
number	1.0	145.0	27.7787	12.0
lat	-47.4	8.7	-18.2439	-17.7
lon	11.3	133.0	59.6153	58.9

---

Curation

Raw data was downloaded from HDX via the CKAN API and converted to Parquet. Column names were lowercased and standardised to snake_case. Common missing-value markers (N/A, null, none, -, unknown, no data, #N/A) were unified to NaN. 4 column(s) were cast from string to numeric or datetime based on parse-success rate (>85% threshold). The dataset was split 80/20 into train and test partitions using a fixed random seed (42) and saved as Snappy-compressed Parquet.

---

Limitations

• Data originates from HDX and has not been independently validated by ESA.
• Automated cleaning cannot correct for misreported values, definitional inconsistencies, or sampling bias in the original collection.
• Refer to the original HDX dataset page for the publisher's own methodology notes and caveats.

---

Citation

@dataset{hdx_africa_mdg_ibtracs_tropical_storm_tracks,
  title     = {Madagascar: IBTrACS Storm Tracks},
  author    = {HDX},
  year      = {2026},
  url       = {https://data.humdata.org/dataset/mdg-ibtracs-tropical-storm-tracks},
  note      = {Repackaged for machine learning by Electric Sheep Africa (https://huggingface.co/electricsheepafrica)}
}

---

Electric Sheep Africa — Africa's ML dataset infrastructure. Lagos, Nigeria.