Spatiotemporal monitoring of rabi wheat crop in Punjab, Pakistan
No hay miniatura disponible
Fecha
Título de la revista
ISSN de la revista
Título del volumen
Editor
FAO ;
Resumen
Descripción
Accurate and timely wheat crop area estimation and monitoring in Punjab, Pakistan, is essential for food security and agricultural planning and trade policy decisions, including import and export strategies. However, traditional field survey methods face limitations in terms of cost and scale. This study presents a comprehensive remote sensing-based methodology for mid-season projection of wheat cultivation areas using satellite imagery and machine learning techniques to address these challenges.The methodology integrates Sentinel-1 Synthetic Aperture Radar and Sentinel-2 optical satellite data with Random Forest classification models implemented on the Google Earth Engine platform. Ground truth data from Asian Development Bank field surveys during the 2022/23 rabi season was preprocessed and reclassified to create training and validation datasets. The approach achieved 89 percent overall classification accuracy, with wheat-specific user and producer accuracies of 92 percent and 97 percent respectively.Key findings reveal that Punjab's agricultural area spans over 12 million hectares, with wheat cultivation estimates of 6.12 million hectares in 2022/23 and 7.19 million hectares in 2023/24. Early projections for 2024/25 indicate approximately 6.46 million hectares, representing a 10 percent decrease from the previous year but maintaining 5.6 percent growth compared to 2022/23. The methodology enables mid-season projections of wheat cultivation before harvest completion, providing timely estimates that support proactive agricultural planning and policy interventions.This scalable, transparent approach using publicly available datasets offers significant potential for replication across other provinces and regions, contributing to enhanced food security monitoring and evidence-based agricultural decision-making throughout Pakistan.