WP-9 performs Sentinel-derived crop classification and longitudinal phenology mapping. It provides the high-resolution 'eyes' of the project, monitoring crop health and land use from space.
Scientific Objectives
Develop high-resolution Earth Observation (EO) derived crop maps for pilot regions.
Implement automated classification algorithms using multi-sensor data (Optical + SAR).
Characterize longitudinal crop phenology and growing season dynamics from space.
Integrate real-time EO indicators into the WebAIS Digital Twin platform.
Execution Framework
T1. Multi-Sensor Data Fusion
Combining optical and radar data for cloud-penetrating monitoring.
- Harmonization of Sentinel-1 (RADAR) and Sentinel-2 (OPTICAL) datasets.
- Development of automated preprocessing pipelines for SAR backscatter and optical reflectances.
- Creation of cloud-masked image composites for continuous spatial monitoring.
- Integration of MODIS and Landsat data for long-term historical context.
T2. Machine Learning for Crop Classification
Deep learning based mapping of agricultural diversity.
- Delineation and mapping of major rice sowing seasons (Boro, Aman, Aus).
- Classification of high-value non-rice crops (Potato, Maize, Mustard) at plot scale.
- Development and training of Random Forest and CNN models for land use mapping.
- Direct validation of EO products using WP-5 ground-truth field records.
T3. Phenology and Growth Health Monitoring
Tracking the heartbeat of the growing season.
- Calculation of longitudinal Vegetation Indices (e.g., NDVI, EVI) for crop health tracking.
- Detection of key phenological stages (Sowing, Booting, Maturity) from time series.
- Identification of crop stress events using multi-temporal backscatter dynamics.
- Visualization of dynamic cropping intensities in the WebAIS map dashboard.
Project Milestones
Annual high-resolution crop type maps (10m) finalized for all pilot regions.
Automated SAR/Optical classification pipeline operational for the Boro season.
10-year longitudinal cropping phenology database established for trend analysis.
Real-time EO-based crop health alerts integrated into the WebAIS platform.
