
OUR
TECHNOLOGY
Satellite Data Sources
Sentinel-1 Satellite

Synthetic Aperture Radar (SAR) satellite
Frequency: 6-12 days
Data Points
- Soil moisture levels
- Surface water extent
- Ground movement/flooding
- Crop structure
Landsat-8 Satellite

Optical imaging satellite
Frequency: 16-day
Key Data Bands Used:
- Band 4 (Red): For vegetation analysis
- Band 5 (Near-Infrared): For water detection
- Band 6 (SWIR 1): For moisture content
- Band 7 (SWIR 2): For cloud detection
Machine Learning Components
CNN (Convolutional Neural Network)

Purpose: Analyze satellite images to detect features like water bodies and field boundaries.
Key Layers: -
Convolution Layers: Detect intricate features within the images.
Pooling Layers: Reduce image size while retaining important information.
Dense Layers: Make final flood probability predictions.- Input: 256x256 pixel satellite image tiles.
Output: Flood probability (0-1).
LSTM (Long Short-Term Memory)
Purpose: Analyze time-series data to predict flood events.
- Input Data Includes:
- Daily rainfall
- River levels
- Soil moisture
- Previous flood events
Sequence Length: 30 days
Output: Flood probability for the next 72 hours.
