Google Earth Engine (GEE) Data Collection Project

Complete Python project for collecting, processing, and exporting satellite imagery from Google Earth Engine. Optimized for GAN and VLM training datasets.

Features

• ✓ GEE Authentication - Secure OAuth2 authentication
• ✓ Image Collection - Sentinel-2 satellite imagery retrieval
• ✓ Cloud Masking - Automatic cloud detection and removal
• ✓ Image Processing - NDVI, NDBI, NDWI calculation
• ✓ Visualization - Interactive maps with geemap
• ✓ Export to Drive - Batch export to Google Drive
• ✓ Configurable - Environment-based settings

Project Structure

gee_data_collection/
├── scripts/                      # Main Python scripts
│   ├── 01_authenticate.py       # GEE authentication
│   ├── 02_initialize.py         # GEE initialization
│   ├── 03_collect_images.py     # Image collection
│   ├── 04_process_images.py     # Image processing
│   ├── 05_visualize.py          # Visualization
│   ├── 06_export_to_drive.py    # Export to Google Drive
│   └── gee_utils.py             # Utility functions
├── config/
│   └── settings.py              # Configuration settings
├── notebooks/
│   └── visualization.ipynb      # Jupyter notebook for interactive visualization
├── data/                        # Local data storage
├── requirements.txt             # Python dependencies
├── .env.example                 # Environment variables template
├── .gitignore                   # Git ignore rules
└── README.md                    # This file

Prerequisites

Item	Version	Notes
Python	3.6 - 3.12	Python 3.13 not supported yet
Google Account	Any	Gmail or institutional email
GEE Account	-	Free at signup.earthengine.google.com
Internet Connection	-	Required for GEE access

Installation

1. Clone or Download Project

# Navigate to your projects directory
cd your_projects_folder

2. Create Virtual Environment

# Windows
python -m venv gee_env
gee_env\Scripts\activate

# macOS/Linux
python3 -m venv gee_env
source gee_env/bin/activate

3. Install Dependencies

pip install -r requirements.txt

4. Configure Settings

# Copy environment template
cp .env.example .env

# Edit .env with your settings
# - Set GEE_PROJECT_ID (from GEE Code Editor)
# - Adjust AOI bounds as needed
# - Update other parameters if desired

Quick Start

Step 1: Authenticate (First Time Only)

python scripts/01_authenticate.py

This opens your browser to authorize GEE access.

Step 2: Initialize GEE

python scripts/02_initialize.py

Verifies connection with your GEE project.

Step 3: Collect Images

python scripts/03_collect_images.py

Retrieves Sentinel-2 imagery for your area of interest.

Step 4: Process Images

python scripts/04_process_images.py

Creates composites and calculates vegetation indices.

Step 5: Visualize (Optional)

# Interactive map (if in Jupyter)
python scripts/05_visualize.py

# Or use Jupyter notebook
jupyter notebook notebooks/visualization.ipynb

Step 6: Export to Google Drive

python scripts/06_export_to_drive.py

Exports processed images to your Google Drive folder.

Configuration

Environment Variables (.env)

# GEE Project
GEE_PROJECT_ID=ee-yourusername

# Area of Interest (Latitude/Longitude bounds)
AOI_MIN_LONGITUDE=88.0
AOI_MIN_LATITUDE=20.0
AOI_MAX_LONGITUDE=92.5
AOI_MAX_LATITUDE=26.5

# Data Collection
START_DATE=2023-01-01
END_DATE=2023-12-31
CLOUD_COVER_THRESHOLD=20
MAX_IMAGES=50

# Export Settings
EXPORT_SCALE=10              # meters (Sentinel-2 native)
EXPORT_CRS=EPSG:4326        # Lat/Lon
EXPORT_FORMAT=GeoTIFF

Custom AOI (Area of Interest)

Edit config/settings.py or .env to change:

• Latitude/Longitude bounds
• Date range
• Cloud cover threshold
• Number of images

Usage Examples

Collect Images for Bangladesh

from scripts.gee_utils import *
from config.settings import *

initialize_gee()
aoi = create_aoi_from_bounds(AOI_BOUNDS)
collection = ee.ImageCollection(SATELLITE_DATASET) \
    .filterBounds(aoi) \
    .filterDate('2023-01-01', '2023-12-31') \
    .filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 20))

Calculate Vegetation Index

from scripts.gee_utils import calculate_ndvi, initialize_gee
import ee

initialize_gee()
image = ee.Image('COPERNICUS/S2_SR_HARMONIZED/20230615T040559_20230615T041557_T45RUI')
ndvi = calculate_ndvi(image)

Export Custom Image

from scripts.gee_utils import create_export_task, start_export_task
from config.settings import *

task = create_export_task(
    image=your_image,
    description='my_export',
    folder='GEE_Exports',
    aoi=aoi,
    scale=10
)
start_export_task(task)

Troubleshooting

Issue	Solution
ModuleNotFoundError: earthengine-api	Run: pip install -r requirements.txt
Authentication fails	Run: python scripts/01_authenticate.py --force
No images found	Check date range, cloud threshold, and AOI bounds
Map not displaying	Install: pip install geemap
Python 3.13 error	Use Python 3.12 or earlier
Google Drive folder not found	Create the folder in Google Drive first

For GAN+VLM Training

After downloading images from Google Drive:

1. Organize by category:

   data/
   ├── flood/
   ├── urban/
   ├── forest/
   └── agricultural/
   

2. Resize images:

   # To 256x256
   convert image.tif -resize 256x256 image_256.tif
   
   # To 128x128
   convert image.tif -resize 128x128 image_128.tif

3. Create captions:

   • Basic: flood, urban, forest
   • Enhanced: Satellite view of flooded agricultural area

4. Use for training:

   • GAN training: Pairs of image→label or style transfer
   • VLM fine-tuning: Image+caption pairs for vision-language models

Supported Datasets

Dataset	ID	Resolution	Bands
Sentinel-2	COPERNICUS/S2_SR_HARMONIZED	10m	11 bands
Landsat 8	USGS/LANDSAT_8_SR	30m	11 bands
Landsat 9	USGS/LANDSAT_9_SR	30m	11 bands

Common Sentinel-2 Bands

Band	Name	Wavelength	Resolution
B2	Blue	490 nm	10 m
B3	Green	560 nm	10 m
B4	Red	665 nm	10 m
B5	Vegetation Edge	705 nm	20 m
B8	NIR	842 nm	10 m
B11	SWIR	1610 nm	20 m
B12	SWIR	2190 nm	20 m

Useful Resources

• GEE Documentation
• Sentinel-2 Info
• geemap Documentation
• GEE Tutorials

License

MIT License - See LICENSE file for details

Support

For issues, errors, or questions:

1. Check the Troubleshooting section
2. Review GEE documentation
3. Check error messages carefully
4. Verify your project ID and authentication

Credits

Built for GAN+VLM training pipeline with satellite imagery from Google Earth Engine.

---

Last Updated: May 2026 Python Version: 3.6 - 3.12 GEE API Version: Latest