Crop Monitoring & Change Detection System

Project Overview

The Crop Monitoring & Change Detection System is a satellite-based geospatial intelligence platform designed to monitor vegetation health, identify potential land-cover changes, and assess agricultural conditions over time. By integrating multi-temporal Sentinel-2 imagery, the Prithvi-EO-2.0 Earth observation foundation model, vegetation indices, and Analytic Hierarchy Process (AHP)-based assessment, the platform provides spatial and temporal insights through an interactive GIS dashboard.

Problem Statement

Agricultural monitoring often depends on periodic field surveys and fragmented satellite-analysis workflows. This makes it challenging to assess vegetation health consistently, identify seasonal changes, and locate areas requiring further investigation. A unified geospatial platform can integrate satellite observations, vegetation indicators, and temporal analysis to improve agricultural monitoring.

Our Approach

Acquire → Preprocess → Analyze → Detect Change → Visualize

The system processes multi-temporal Sentinel-2 imagery using cloud and shadow masking, reflectance preprocessing, and vegetation-index calculation. Frozen Prithvi-EO-2.0 embeddings support crop classification and latent change detection. An AHP-based methodology combines seven vegetation indices into a composite Vegetation Health Score (VHS), while annual comparison overlays and interactive visualizations support temporal analysis.

Core Functionality

• Multi-temporal Sentinel-2 imagery processing
• Prithvi-based geospatial feature extraction
• Latent change detection using cosine distance
• Crop and non-crop patch classification
• Vegetation-index calculation and analysis
• AHP-based vegetation health assessment
• Annual vegetation change overlays
• Patch-level inspection
• Multi-year monitoring across five study areas

Technologies Used

Python, Sentinel-2 Satellite Imagery, Prithvi-EO-2.0, PyTorch, Google Earth Engine, Rasterio, NumPy, NDVI, EVI, SAVI, MSAVI, DVI, RVI, ARVI, and Analytic Hierarchy Process (AHP).

Results & Impact

The platform provides an integrated workflow for monitoring vegetation conditions and investigating potential agricultural changes across Tirupati, West Godavari, Anantapur, Krishna, and Guntur in Andhra Pradesh. Satellite-derived indicators, foundation-model analysis, and temporal visualization support crop monitoring, vegetation-health assessment, and data-driven land-condition analysis.

Target Users

• Agricultural monitoring departments
• Remote sensing and GIS analysts
• Government agriculture departments
• Crop and land-use researchers
• Environmental monitoring teams
• Geospatial intelligence teams
• Agricultural planning and assessment personnel

Technical Challenges

Maintaining consistent multi-temporal satellite preprocessing, handling cloud-contaminated and missing pixels, integrating foundation-model embeddings with downstream classifiers, calibrating change-detection thresholds, and ensuring reliable spatial and temporal comparisons.

Operational Challenges

Managing differences in satellite acquisition dates and seasonal conditions, validating crop classifications against independent reference data, interpreting vegetation-index variations, and distinguishing actual land-cover changes from seasonal or atmospheric effects.