ImpactMesh-Fire

[](https://arxiv.org/abs/todo)
[](https://github.com/IBM/ImpactMesh)
[](https://research.ibm.com/blog/todo)

# ImpactMesh-Fire

ImpactMesh is a large-scale multimodal, multitemporal dataset for flood and wildfire mapping, released by IBM, DLR, and the ESA Φ-lab.
It integrates **Sentinel-1 SAR**, **Sentinel-2 optical**, **Copernicus DEM**, and high-quality annotations from Copernicus EMS.
The technical report is released soon. You find the flood subset here: https://huggingface.co/datasets/ibm-esa-geospatial/ImpactMesh-Flood.


---
## Features
- Multimodal: SAR, optical, DEM
- Multitemporal: Four time steps (pre-month, pre-event, event, post-event)
- Global coverage: 200+ fire events
- Scale: 80K samples
- License: CC-BY 4.0

## Quick Start

Download the dataset with:
```shell
hf download ibm-esa-geospatial/ImpactMesh-Fire --repo-type dataset --local-dir data/ImpactMesh-Fire

# Only download a single modality (e.g., S2L2A)
hf download ibm-esa-geospatial/ImpactMesh-Fire --repo-type dataset --include "data/S2L2A.tar" --local-dir data/ImpactMesh-Fire
```

Untar the samples with:
```shell
tar -xf data/ImpactMesh-Fire/data/S1RTC.tar -C data/ImpactMesh-Fire/data
tar -xf data/ImpactMesh-Fire/data/S2L2A.tar -C data/ImpactMesh-Fire/data
tar -xf data/ImpactMesh-Fire/data/DEM.tar -C data/ImpactMesh-Fire/data
tar -xf data/ImpactMesh-Fire/data/MASK.tar -C data/ImpactMesh-Fire/data
```

We use [TerraTorch](https://terrastackai.github.io/terratorch/stable/) for the model fine-tuning and provide data modules for ImpactMesh. You can download the code and configs for the fine-tuning from https://github.com/IBM/ImpactMesh.

Alternatively, you can install the data loading code with:

```shell
pip install impactmesh
```

```shell
terratorch fit --config configs/terramind_v1_tiny_impactmesh_fire.yaml
```

## Citation

Our technical report is released soon!

## Acknowledgement

ImpactMesh was developed as part of the FAST‑EO project funded by the European Space Agency Φ‑Lab (contract #4000143501/23/I‑DT).

Sentinel-2 Level-2A data were downloaded from Microsoft Planetary Computer and are provided under Copernicus Sentinel license conditions (© European Union 2015–2025, ESA) (https://planetarycomputer.microsoft.com/dataset/sentinel-2-l2a).

Sentinel-1 Radiometrically Terrain Corrected (RTC) SAR data were retrieved from Microsoft Planetary Computer (calibrated to GRD and terrain-corrected using PlanetDEM) under Copernicus Sentinel license terms (© European Union 2014–2025) (https://planetarycomputer.microsoft.com/dataset/sentinel-1-rtc).

The DEM data is produced using Copernicus WorldDEM-30 © DLR e.V. 2010-2014 and © Airbus Defence and Space GmbH 2014-2018 provided under COPERNICUS by the European Union and ESA; all rights reserved.

Annotations were sourced from the Copernicus Emergency Management Service (© European Union, 2012–2025), available at https://emergency.copernicus.eu/.

[](https://arxiv.org/abs/todo)
[](https://github.com/IBM/ImpactMesh)
[](https://research.ibm.com/blog/todo)

# ImpactMesh-Fire

ImpactMesh是由IBM、德国宇航中心（DLR）以及欧洲空间局（ESA）Φ实验室联合发布的大规模多模态多时间序列数据集，用于洪涝与野火制图。该数据集整合了**Sentinel-1合成孔径雷达（Sentinel-1 SAR）**、**Sentinel-2光学影像**、**哥白尼数字高程模型（Copernicus DEM）**，以及来自哥白尼应急管理服务（Copernicus EMS）的高质量标注数据。相关技术报告即将发布。您可通过以下链接获取洪涝子集：https://huggingface.co/datasets/ibm-esa-geospatial/ImpactMesh-Flood。


---
## 数据集特性
- 多模态：支持合成孔径雷达、光学影像与数字高程模型
- 多时间序列：包含四个时间节点（灾前一月、灾前、灾时、灾后）
- 全球覆盖：涵盖200余起野火事件
- 数据规模：包含8万余个样本
- 授权协议：CC-BY 4.0

## 快速上手
通过以下命令下载数据集：
shell
hf download ibm-esa-geospatial/ImpactMesh-Fire --repo-type dataset --local-dir data/ImpactMesh-Fire

# 仅下载单一模态数据（例如S2L2A）
hf download ibm-esa-geospatial/ImpactMesh-Fire --repo-type dataset --include "data/S2L2A.tar" --local-dir data/ImpactMesh-Fire

通过以下命令解压样本：
shell
tar -xf data/ImpactMesh-Fire/data/S1RTC.tar -C data/ImpactMesh-Fire/data
tar -xf data/ImpactMesh-Fire/data/S2L2A.tar -C data/ImpactMesh-Fire/data
tar -xf data/ImpactMesh-Fire/data/DEM.tar -C data/ImpactMesh-Fire/data
tar -xf data/ImpactMesh-Fire/data/MASK.tar -C data/ImpactMesh-Fire/data

我们使用[TerraTorch](https://terrastackai.github.io/terratorch/stable/)进行模型微调，并为ImpactMesh提供数据处理模块。您可通过https://github.com/IBM/ImpactMesh 下载微调所需的代码与配置文件。

此外，您也可通过以下命令安装数据加载代码：
shell
pip install impactmesh

shell
terratorch fit --config configs/terramind_v1_tiny_impactmesh_fire.yaml

## 引用声明
相关技术报告即将发布！

## 致谢
ImpactMesh是欧洲空间局Φ实验室资助的FAST‑EO项目的研究成果（项目编号：4000143501/23/I‑DT）。

Sentinel-2 Level-2A数据源自微软行星计算机（Microsoft Planetary Computer），依据哥白尼哨兵数据许可协议发布（© 欧盟2015–2025，欧洲空间局ESA），链接：https://planetarycomputer.microsoft.com/dataset/sentinel-2-l2a。

Sentinel-1辐射地形校正（RTC）合成孔径雷达数据源自微软行星计算机（已校准至GRD格式并使用PlanetDEM完成地形校正），依据哥白尼哨兵数据许可协议发布（© 欧盟2014–2025），链接：https://planetarycomputer.microsoft.com/dataset/sentinel-1-rtc。

本数据集所用的数字高程模型数据基于哥白尼WorldDEM-30制作，© 德国宇航中心DLR e.V. 2010-2014、© 空客防务与航天公司2014-2018，由欧盟与欧洲空间局ESA依据哥白尼计划提供，保留所有权利。

标注数据源自哥白尼应急管理服务（© 欧盟2012–2025），获取链接：https://emergency.copernicus.eu/。