🌲 WINDWURF – Storm Damage Detection in Swiss Forests

WINDWURF is an interactive tool to rapidly assess storm damage in Swiss forests using openly available Sentinel-2 satellite imagery provided via the swisstopo swissEO S2-SR Product. Developed during Data Hackdays Bern 2025, this project helps the Amt für Wald und Naturgefahren (AWN) get timely insight into storm-related forest disturbances over 180,000 hectares of woodland — an area too vast for field inspection alone.

-> Visit the WebApp : WINDWURF

🪵 Why WINDWURF?

With storms intensifying over the last 50 years, detecting forest damage efficiently is critical for mitigation, logging decisions, and ecological monitoring. WINDWURF empowers authorities with a scalable, transparent, and reproducible approach to disaster response.

🚀 What It Does

After a storm event, WINDWURF allows users to

• Select a date of a storm event

• Automatically get potential forest damage by compare satellite imagery before and after the event

• Highlight forest areas with significant vegetation loss (potential storm damage)

The result: a fast, data-driven overview of potentially affected forest regions.

🛰️ How It Works

WINDWURF is powered by Google Earth Engine (GEE) and utilizes Sentinel-2 imagery provided by swissEO S2-SR processed with a pixel-wise dNBR (delta Normalized Burn Ratio) analysis. Core Analysis Steps:

NBR Calculation:

    NBR = (B8A - B11) / (B8A + B11)

    Computed for pre- and post-storm periods

dNBR Calculation:

    dNBR = NBR_post - NBR_pre

    Significant vegetation loss → low dNBR values (≤ -0.15)

Filtering & Masking:

    [Swiss-harmonized Sentinel-2 data](https://www.swisstopo.admin.ch/de/satellitenbilder-swisseo-s2-sr)

    Forest-only mask (Swiss BAFU ecosystem data)[https://www.wsl.ch/de/projekte/lebensraumkarte-schweiz-1/]

    Custom 10m cloud (CloudScorePlus) and terrain (swissSURFACE3D) and shadow masks

    Snow masking using NDSI

Outputs:

 Median NBR before and after the event

 dNBR layer showing vegetation change

 Disturbance mask (dNBR ≤ -0.15) highlighting severe impact zones

Tech Stack

 Google Earth Engine (data analysis & visualization)

 Sentinel-2 imagery (via [swissEO S2-SR](https://www.swisstopo.admin.ch/de/satellitenbilder-swisseo-s2-sr))

 Forest & snow masking (Swiss federal data)

 React frontend (using Lovable)

App Usage

Visit the App and follow these steps:

Wählen Sie das Datum des Sturmereignisses .

Die Karte wird automatisch aktualisiert. Gelbe Flächen zeigen potentielle Sturmschäden.

Zoomen und verschieben Sie die Karte für eine genauere Ansicht.

📊 Legend

Color Meaning 🟡 Potential forest damage (dNBR ≤ -0.15)

📅 Notes

60-day comparison window post-storm

Compared to same window from previous year

Only forests are analyzed

Disturbance areas appear in yellow on delta map

Works best in Summer, not good during wintertime

📁 File Structure (Important Components)

gee_nbr_app.js — GEE script with NBR logic, cloud/snow masking

Next Steps & Future Development

The WINDWURF project has established a solid foundation for storm damage detection in Swiss forests, but several enhancements could further increase its value for forest management:

Quantitative Analysis Features

• Area Measurement Tool: Implement functionality to calculate the total affected area in hectares for selected regions
• Economic Impact Assessment: Develop an algorithm to estimate financial damage (CHF) based on:
  • Average timber value per hectare
  • Severity of damage (using# dNBR thresholds)
  • Forest type classification

Quantitative Analysis Features

Historical Analysis

• Storm Event Database: Create a pulldown menu of historical storm events in Switzerland
• Temporal Comparison: Enable comparison between current and past storm impacts on the same forest areas
• Long-term Trend Analysis: Visualize changing patterns of storm vulnerability across decades

Quantitative Analysis Features

Enhanced Visualization

• Damage Severity Classification: Implement a multi-level classification system beyond binary damage detection
• 3D Visualization: Integrate with swissSURFACE3D for topographic analysis of damage patterns
• Time-lapse Feature: Show forest recovery progress over time after storm events

Quantitative Analysis Features

Data Integration

• Field Validation Module: Allow foresters to upload ground-truth data to improve algorithm accuracy
• Integration with Weather Data: Correlate damage patterns with wind speed and direction data
• Automated Reporting: Generate PDF reports with statistics for affected forest management units

The implementation of these features would transform WINDWURF from a detection tool into a comprehensive forest storm damage management system, supporting the entire workflow from initial assessment to recovery planning.

👥 Team & Acknowledgments

Developed at Data Hackdays Bern with support from:

• Swisstopo – Harmonized Sentinel-2 data - Tschoun,davidoesch

• Amt für Wald und Naturgefahren (AWN) – Challenge owner nnjae, Dani Steinberger

• Swiss Federal Office for the Environment (BAFU) -expert guidance Rdataflow, Yannick Barton

• Eidg. Forschungsanstalt WSL and swisstopo – expert guidance and SAR Wrestling Marius Rüetschi

• BFH Institut Public Sector Transformation - FullStack! holdan-8