Flood Mapping and Impact Assessment Using Google Earth Engine

Project Overview

This project was developed as part of the SES 598 Cloud-Based Remote Sensing course at Habib University, led by Instructor Jiwei Li. It leverages Google Earth Engine (GEE) to map flood events and assess their impacts on urban and agricultural land. Using satellite data from Sentinel-1 SAR and MODIS, the application enables near-real-time flood detection and impact analysis, empowering decision-makers in flood-prone areas to take informed actions.

REPORT: Final_Report.pdf
PRESENTATION: Final_Presentation.pptx

---

Project Motivation

The frequency of extreme weather events, such as flooding, has increased due to climate change. A striking example was the 2022 Pakistan floods, which submerged one-third of the Sindh province. This project aims to provide a scalable solution to quickly map and assess flood impacts, helping communities prepare and respond to future events.

---

Project Features

• Flood Mapping:

  • Utilizes Sentinel-1 SAR (Synthetic Aperture Radar) data with high spatial resolution (10m), capable of capturing flood events even under cloud cover.
  • Supports before-and-after analysis using user-defined date ranges.

• Impact Assessment:

  • Integrates MODIS land cover data to analyze flood effects on urban and crop land.
  • Generates statistics such as total flooded area and percentage of affected regions.

• Speckle Noise Reduction:

  • Applies a Refined Lee Speckle Filter to enhance image clarity by reducing radar noise.

• User-Friendly Interface:

  • Dynamic input panels for region selection, date inputs, and analysis parameters.
  • Customizable thresholds for flood and terrain filtering.

• Visualization and Outputs:

  • Displays flood extent, impacted urban/crop areas, and a detailed legend.
  • Outputs key statistics for decision-making.

---

How It Works

1. Select Region: Choose a country and state using administrative boundaries from the FAO GAUL dataset.
2. Define Timeframes: Specify "before" and "after" flood event date windows.
3. Run Analysis: The app processes Sentinel-1 and MODIS satellite data to detect flooded regions.
4. View Results: Visual maps and statistical outputs appear on the interface, highlighting affected areas.

---

Datasets Used

Dataset	Description	Resolution	Source
Sentinel-1	C-band SAR for flood detection	10m	ESA Copernicus
MODIS	Land cover data for impact assessment	500m	NASA EOSDIS
Global Surface Water	Seasonal water body data	Various	European Commission
HydroSHEDS	Elevation model for slope filtering	3 arc-seconds	WWF

---

Code Structure

File	Description
main.js	Core analysis and flood mapping logic
ui.js	User interface code
fetchregion.js	Fetches regional boundaries from FAO GAUL
specklefilter.js	Implements speckle noise reduction

---

Design Challenges & Future Work

• Asynchronous Data Processing: Managing Google Earth Engine's asynchronous tasks within a user interface posed challenges.
• Edge Case Testing: Improving the robustness of flood detection in varying geographical and climatic conditions is ongoing.
• Enhanced Parameter Control: Future versions aim to offer more dynamic user-defined parameters and visualization features.

---

Cost Efficiency

As a cloud-based solution utilizing open-access satellite data, this project incurs minimal operational costs while offering scalability and global applicability.

---

How to Use

1. Install Google Earth Engine and ensure access to relevant datasets.
2. Clone this repository.
3. Run the app through GEE's code editor by uploading the provided scripts (main.js, ui.js, etc.).
4. Follow on-screen prompts to configure and execute the flood analysis.

---