GIS-Based Crop Mapping

Project Overview

This project developed the first systematic satellite-based crop inventory for the full Gezira-Managil complex (2.2 million hectares), replacing previous manual survey methods. A multi-scale approach combined MODIS 250 m time-series phenology for large-area crop extent detection with Sentinel-2 10 m imagery for field-level classification. Object-Based Image Analysis (OBIA) using eCognition algorithms segmented field boundaries and classified 8 crop types across 3 growing seasons.

Methodology

1. 01MODIS MOD13Q1 NDVI time-series smoothing (Savitzky-Golay) for phenology extraction
2. 02Multi-resolution segmentation in eCognition for field boundary delineation
3. 03Random Forest classification on 35 spectral, textural, and temporal features
4. 04NDVI change detection for inter-season crop rotation mapping
5. 05GPS-RTK ground control network establishment (120 points across 6 blocks)
6. 06ArcGIS Pro geodatabase design for annual inventory management

Key Outputs

• 2.2 million hectare multi-class crop inventory at 10 m resolution
• 30% improvement in classification accuracy over previous manual method
• Annual crop monitoring protocol adopted by HRC Sudan management
• Geodatabase archive of 3 growing seasons for time-series analysis

2.2M ha mapped, 30% classification accuracy improvement