High-resolution Flood Susceptibility Mapping and Exposure Assessment in Pakistan

---

Open Source Data Available

The high-resolution (30m) flood susceptibility maps for Pakistan are now publicly available for download!

Download Option	Link	Format
Zenodo Repository	doi.org/10.5281/zenodo.18513601	Cloud Optimized GeoTIFF
LGBM Model (Direct)	Download fsm_lgbm_pakistan.tif	GeoTIFF (~150 MB)
XGBoost Model (Direct)	Download fsm_xgboost_pakistan.tif	GeoTIFF (~150 MB)
Google Earth Engine	See GEE Assets section	EE Asset

---

Overview

This repository provides the first national-scale, high-resolution (30m) flood susceptibility maps for Pakistan, developed using an integrated artificial intelligence, machine learning, and geospatial framework.

Key Highlights

• High-Resolution Mapping: National-scale flood susceptibility maps at 30m spatial resolution
• Best Performing Model: LightGBM (LGBM) with 0.85 accuracy
• Five Susceptibility Classes: Very Low, Low, Moderate, High, Very High
• Cloud-Native Access: Available as Cloud Optimized GeoTIFFs and Google Earth Engine assets
• Interactive Web App: Explore data without coding via GEE App

---

Publication

Waleed, M., & Sajjad, M. (2025). High-resolution flood susceptibility mapping and exposure assessment in Pakistan: An integrated artificial intelligence, machine learning and geospatial framework. International Journal of Disaster Risk Reduction, 121, 105442.

Resource	Link
Paper DOI	doi.org/10.1016/j.ijdrr.2025.105442
Paper PDF	Download PDF

Note: Figure 8 has been corrected. See Corrigendum DOI: 10.1016/j.ijdrr.2025.105842 (Published: November 14, 2025)

---

Interactive Notebooks

We provide two Jupyter notebooks for working with the flood susceptibility data:

Notebook	Description	Launch
01_zenodo_cog_analysis.ipynb	Cloud-Native COG Workflow Work with Zenodo-hosted Cloud Optimized GeoTIFFs directly without downloading. Demonstrates windowed reads for regional analysis (e.g., Karachi), visualization with custom colormaps, and efficient data extraction.
02_gee_interactive_map.ipynb	Google Earth Engine Workflow Server-side processing using GEE Python API. Features interactive split-panel comparison with satellite imagery, spatial analysis, and area calculations using geemap.

See docs/environment_setup.md for setup instructions.

---

Data Access

Zenodo Repository (Cloud Optimized GeoTIFFs)

The recommended method for downloading and analyzing data locally:

• DOI: 10.5281/zenodo.18513601
• Format: Cloud Optimized GeoTIFF (COG) - supports windowed reads without full download
• LGBM Model: fsm_lgbm_pakistan.tif (Best performing, 0.85 accuracy)
• XGBoost Model: fsm_xgboost_pakistan.tif (0.82 accuracy)

Google Earth Engine Assets

For server-side processing without downloading:

Model	Asset ID
LGBM	projects/waleedgeo/assets/fsm_pk_lgbm
XGBoost	projects/waleedgeo/assets/fsm_pk_xgboost

Interactive Web Application

Launch FSM-PK Interactive App

Data Specifications

Property	Value
Spatial Resolution	30 meters
Coordinate System	EPSG:4326 (WGS84)
Extent	Pakistan national boundary
Value Range	1-5 (Very Low to Very High)
Data Type	Unsigned 8-bit integer

For detailed data access instructions, see data/README.md.

---

Repository Structure

FSM-PK/
├── data/                    # Data documentation and paper PDF
├── notebooks/               # Jupyter notebooks (COG & GEE workflows)
├── codes/                   # GEE App source code
├── docs/                    # Setup documentation
├── img/                     # Images and figures
├── other/                   # Citation files and demo GIF
├── requirements.txt         # Python dependencies
└── LICENSE                  # CC BY-NC-SA 4.0

---

Citation

@article{WALEED2025105442,
  title = {High-resolution flood susceptibility mapping and exposure assessment in Pakistan: An integrated artificial intelligence, machine learning and geospatial framework},
  journal = {International Journal of Disaster Risk Reduction},
  volume = {121},
  pages = {105442},
  year = {2025},
  doi = {https://doi.org/10.1016/j.ijdrr.2025.105442},
  author = {Mirza Waleed and Muhammad Sajjad}
}

Export citation: BibTeX | RIS

---

Authors

Mirza Waleed	Primary Author
Email	waleedgeo@outlook.com
LinkedIn	linkedin.com/in/waleedgeo
Website	waleedgeo.com
Dr. Muhammad Sajjad	Co-author
Google Scholar	Profile

---

License

Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)

---

Keywords

flood · Pakistan · flood susceptibility · machine learning · LGBM · XGBoost · Google Earth Engine · geospatial · remote sensing · disaster risk management

---

If you find this work useful, please star this repository!