Sowing Information Extraction for Groundnut Crop with Temporal optical data using Fuzzy Machine Learning Model

Suhas M, Anil Kumar*

Abstract

This interdisciplinary research integrates advanced technology, remote sensing, and AI to enhance precision farming in the Dungargarh Tehsil of Bikaner District, Rajasthan, India. The study focuses on two main objectives: groundnut crop classification (mapping) and the extraction of sowing information. Utilizing temporal optical data preprocessing, optimized temporal indices, and a contextual fuzzy model, the Modified Possibilistic c-Means (MPCM) algorithm with both conventional mean and Individual Sample as Mean (ISM) training approaches were employed. Quantitative results demonstrate that CBSI MSAVI2 achieves a significantly lower mean membership difference (MMD) of 0.00196 and a variance of 0.5 compared to conventional MSAVI2. Further experimentation identifies ADMPLICM with a 3x3 window size and the ISM training approach as the optimal algorithm based on MMD and Variance values. By integrating vegetation indices, training approaches, and fuzzy-based algorithms, this study offers a novel approach for extracting groundnut sowing information. The results provide valuable insights into the temporal dynamics of groundnut sowing, offering reliable tools for farmers, agencies, and researchers, ultimately contributing to sustainable agricultural practices.

Keywords: Groundnut crop, Modified possibilistic c-means (MPCM), Individual samples as mean (ISM), Class based sensor independent (CBSI), Modified soil adjusted vegetation index (MSAVI)

Full Text: