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

---
annotations_creators:
- no-annotation
language_creators:
- found
language:
- en
license: cc-by-4.0
multilinguality:
- monolingual
size_categories:
- 1K<n<10K
source_datasets:
- original
task_categories:
- other
task_ids: []
tags:
- africa
- humanitarian
- hdx
- electric-sheep-africa
- cyclones-hurricanes-typhoons
- hxl
- dji
pretty_name: "Djibouti: IBTrACS Storm Tracks"
dataset_info:
splits:
- name: train
num_examples: 6140
- name: test
num_examples: 1535
---

# Djibouti: IBTrACS Storm Tracks

**Publisher:** HDX · **Source:** [HDX](https://data.humdata.org/dataset/dji-ibtracs-tropical-storm-tracks) · **License:** `cc-by-igo` · **Updated:** 2025-10-14

---

## 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: **DJI**.

*Curated into ML-ready Parquet format by [Electric Sheep Africa](https://huggingface.co/electricsheepafrica).*

---

## Dataset Characteristics

| | |
|---|---|
| **Domain** | Climate and environment |
| **Unit of observation** | Geolocated point observations |
| **Rows (total)** | 7,676 |
| **Columns** | 12 (3 numeric, 8 categorical, 1 datetime) |
| **Train split** | 6,140 rows |
| **Test split** | 1,535 rows |
| **Geographic scope** | DJI |
| **Publisher** | HDX |
| **HDX last updated** | 2025-10-14 |

---

## Variables

**Geographic** — `iso_time`, `lat` (range -13.2–31.9), `lon` (range 41.3–141.0).

**Outcome / Measurement** — `number` (range 1.0–142.0).

**Identifier / Metadata** — `sid` (2013305N07141, 1996288N09092, 2016102S12074), `esa_source` (HDX), `esa_processed` (2026-04-06).

**Other** — `basin` (North India, South Indian, Western North Pacific), `subbasin` (Arabian Sea, Bay of Bengal, Missing), `nature` (Tropical, Not reported, Disturbance), `wmo_wind` ( , 25, 30), `wmo_pres` ( , 1000, 1002).

---

## Quick Start

```python
from datasets import load_dataset

ds = load_dataset("electricsheepafrica/africa-dji-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% | 2013305N07141, 1996288N09092, 2016102S12074 |
| `number` | float64 | 0.0% | 1.0 – 142.0 (mean 65.6345) |
| `basin` | object | 0.0% | North India, South Indian, Western North Pacific |
| `subbasin` | object | 0.0% | Arabian Sea, Bay of Bengal, Missing |
| `iso_time` | datetime64[ns] | 0.0% | |
| `nature` | object | 0.0% | Tropical, Not reported, Disturbance |
| `lat` | float64 | 0.0% | -13.2 – 31.9 (mean 13.2014) |
| `lon` | float64 | 0.0% | 41.3 – 141.0 (mean 65.6669) |
| `wmo_wind` | object | 0.0% | , 25, 30 |
| `wmo_pres` | object | 0.0% | , 1000, 1002 |
| `esa_source` | object | 0.0% | HDX |
| `esa_processed` | object | 0.0% | 2026-04-06 |

---

## Numeric Summary

| Column | Min | Max | Mean | Median |
|---|---|---|---|---|
| `number` | 1.0 | 142.0 | 65.6345 | 74.0 |
| `lat` | -13.2 | 31.9 | 13.2014 | 13.2 |
| `lon` | 41.3 | 141.0 | 65.6669 | 64.5 |

---

## 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](https://data.humdata.org/dataset/dji-ibtracs-tropical-storm-tracks) for the publisher's own methodology notes and caveats.

---

## Citation

```bibtex
@dataset{hdx_africa_dji_ibtracs_tropical_storm_tracks,
title = {Djibouti: IBTrACS Storm Tracks},
author = {HDX},
year = {2025},
url = {https://data.humdata.org/dataset/dji-ibtracs-tropical-storm-tracks},
note = {Repackaged for machine learning by Electric Sheep Africa (https://huggingface.co/electricsheepafrica)}
}
```

---

*[Electric Sheep Africa](https://huggingface.co/electricsheepafrica) — Africa's ML dataset infrastructure. Lagos, Nigeria.*

annotations_creators: 无标注
language_creators: 抓取获取
language: 英语
license: CC BY 4.0
multilinguality: 单语言
size_categories: 1000 < n < 10000
source_datasets: 原始数据集
task_categories: 其他
task_ids: 无
tags: 非洲、人道主义、HDX、electric-sheep-africa、气旋-飓风-台风、HXL、DJI
pretty_name: "吉布提：IBTrACS风暴路径"
dataset_info:
splits:
- name: train
num_examples: 6140
- name: test
num_examples: 1535

# 吉布提：IBTrACS风暴路径

**发布方：** 人道主义数据交换（Humanitarian Data Exchange, HDX） · **来源：** [HDX](https://data.humdata.org/dataset/dji-ibtracs-tropical-storm-tracks) · **授权协议：** `cc-by-4.0` · **更新时间：** 2025-10-14

---

## 摘要
国际最佳路径档案气候管理项目（International Best Track Archive for Climate Stewardship, IBTrACS）是目前可用的最完整的全球热带气旋数据集集合。该项目整合了多机构的最新及历史热带气旋数据，构建了统一的公开可用最佳路径数据集，可提升不同机构间的比对效率。

可用字段：
- SID：IBTrACS算法分配的唯一风暴标识符（SID）
- ISO_TIME：采用ISO格式的观测时间（YYYY-MM-DD hh:mm:ss）
- BASIN：当前风暴所在的洋盆
- SUBBASIN：当前风暴所在的子洋盆
- NATURE：风暴类型（整合了各来源提供的多种类型）
- NUMBER：当年的风暴编号（每年从1重新开始计数）
- LAT：平均位置纬度（整合了各来源提供的位置数据）
- LON：平均位置经度（整合了各来源提供的位置数据）
- WMO_WIND：负责世界气象组织（World Meteorological Organization, WMO）机构指定的最大持续风速
- WMO_PRES：负责WMO机构指定的最低中心气压

该数据集的每一行均代表一个地理定位的点观测数据。时间覆盖范围由`iso_time`列标注。地理覆盖范围：**DJI（吉布提）**。
本数据集已由[Electric Sheep Africa](https://huggingface.co/electricsheepafrica)整理为适配机器学习的Parquet格式。

---

## 数据集特征
| | |
|---|---|
| **领域** | 气候与环境 |
| **观测单元** | 地理定位点观测数据 |
| **总行数** | 7,676 |
| **列数** | 12（3个数值型、8个分类型、1个日期时间型） |
| **训练集划分** | 6,140行 |
| **测试集划分** | 1,535行 |
| **地理覆盖范围** | DJI（吉布提） |
| **发布方** | HDX |
| **HDX最后更新时间** | 2025-10-14 |

---

## 变量
**地理相关变量**：`iso_time`、`lat`（取值范围：-13.2~31.9）、`lon`（取值范围：41.3~141.0）。
**结果/测量变量**：`number`（取值范围：1.0~142.0）。
**标识符/元数据变量**：`sid`（示例值：2013305N07141、1996288N09092、2016102S12074）、`esa_source`（HDX）、`esa_processed`（2026-04-06）。
**其他变量**：`basin`（北印度洋、南印度洋、西北太平洋）、`subbasin`（阿拉伯海、孟加拉湾、缺失值）、`nature`（热带气旋、未报告、扰动）、`wmo_wind`（空值、25、30）、`wmo_pres`（空值、1000、1002）。

---

## 快速上手
python
from datasets import load_dataset

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

print(train.shape)
train.head()

---

## 数据Schema
| 列名 | 数据类型 | 空值占比 | 取值范围/示例值 |
|---|---|---|---|
| `sid` | object | 0.0% | 2013305N07141、1996288N09092、2016102S12074 |
| `number` | float64 | 0.0% | 1.0 – 142.0（均值65.6345） |
| `basin` | object | 0.0% | 北印度洋、南印度洋、西北太平洋 |
| `subbasin` | object | 0.0% | 阿拉伯海、孟加拉湾、缺失值 |
| `iso_time` | datetime64[ns] | 0.0% | 无 |
| `nature` | object | 0.0% | 热带气旋、未报告、扰动 |
| `lat` | float64 | 0.0% | -13.2 – 31.9（均值13.2014） |
| `lon` | float64 | 0.0% | 41.3 – 141.0（均值65.6669） |
| `wmo_wind` | object | 0.0% | 空值、25、30 |
| `wmo_pres` | object | 0.0% | 空值、1000、1002 |
| `esa_source` | object | 0.0% | HDX |
| `esa_processed` | object | 0.0% | 2026-04-06 |

---

## 数值型变量统计摘要
| 列名 | 最小值 | 最大值 | 均值 | 中位数 |
|---|---|---|---|---|
| `number` | 1.0 | 142.0 | 65.6345 | 74.0 |
| `lat` | -13.2 | 31.9 | 13.2014 | 13.2 |
| `lon` | 41.3 | 141.0 | 65.6669 | 64.5 |

---

## 数据整理流程
原始数据通过CKAN应用程序编程接口（CKAN API）从HDX下载，并转换为Parquet格式。列名统一转换为小写蛇形命名法。常见的缺失值标记（`N/A`、`null`、`none`、`-`、`unknown`、`no data`、`#N/A`）被统一替换为`NaN`。根据解析成功率（阈值>85%），将4列从字符串类型转换为数值型或日期时间型。本数据集使用固定随机种子（42）按照80/20的比例划分为训练集和测试集，并保存为Snappy压缩的Parquet格式。

---

## 数据集局限性
- 数据源自HDX，尚未经Electric Sheep Africa独立验证。
- 自动化清洗无法修正原始数据集中的错误报告值、定义不一致或采样偏差问题。
- 请参阅[原始HDX数据集页面](https://data.humdata.org/dataset/dji-ibtracs-tropical-storm-tracks)获取发布方提供的方法说明与注意事项。

---

## 引用格式
bibtex
@dataset{hdx_africa_dji_ibtracs_tropical_storm_tracks,
title = {Djibouti: IBTrACS Storm Tracks},
author = {HDX},
year = {2025},
url = {https://data.humdata.org/dataset/dji-ibtracs-tropical-storm-tracks},
note = {由Electric Sheep Africa（https://huggingface.co/electricsheepafrica）重新打包以适配机器学习任务}
}

---

*[Electric Sheep Africa](https://huggingface.co/electricsheepafrica) — 非洲机器学习数据集基础设施提供商，总部位于尼日利亚拉各斯。*