Estimation of wheat crop production using multispectral information fusion

Area of Science:

• Agricultural Remote Sensing
• Geospatial Analysis
• Crop Monitoring

Background:

• Estimating wheat crop area and production is crucial for food security in regions like Uttar Pradesh, India.
• Multispectral satellite imagery from Sentinel-2, Landsat-8, and Landsat-9 was acquired during the pre-harvest season (March) for 2021 and 2022.
• A multispectral information fusion approach was developed for accurate crop assessment.

Purpose of the Study:

• To estimate wheat crop area and production in the Etah region of Uttar Pradesh, India.
• To evaluate the effectiveness of a multispectral information fusion approach for crop monitoring.
• To compare the performance of different classification algorithms and vegetation indices.

Main Methods:

• Image classification using Support Vector Machine (SVM), Artificial Neural Network (ANN), and Maximum Likelihood (ML) algorithms.
• Vegetation index-based extraction using Normalized Difference Vegetation Index (NDVI) and Fractional Vegetation Cover (FVC).
• Separability analysis to identify optimal image bands for land cover classification and wheat crop extraction.

Main Results:

• Maximum classification accuracy of 98.34% achieved with Sentinel-2 data using ANN.
• Estimated wheat area for Sentinel-2 in 2021 (260,540 ha) closely matched reference data (238,600 ha).
• SVM classifier demonstrated good performance, calculating 8,408,490 tons of wheat for 2022 using Sentinel-2 data.

Conclusions:

• The proposed method offers a novel approach for crop production monitoring by utilizing a single image per year with ground truth data.
• The fusion of multispectral data and advanced classification techniques enhances the accuracy of crop area and yield estimation.
• This methodology provides a reliable tool for agricultural planning and food security assessments.