Abstract

In recent years, remote sensing has developed agriculture by leveraging satellite imagery to achieve various goals, including land classification, crop area identification, growth monitoring, pattern recognition, and agricultural surveys. This technology has been extensively used for major crops like wheat, cotton, and rice. Cotton plays a pivotal role in the global agricultural sector and economy, providing fiber and substantial income while yielding valuable by-products like seeds and banola. Researchers worldwide have developed various remote sensing methods to monitor crop production, yet there remains a research gap in Pakistan concerning the utilization of Sentinel images for monitoring cotton growth. Therefore, it is crucial to employ remote sensing technology for more precise identification of cotton growth areas in Pakistan to optimize production and resource management. In our study, we propose a remote sensing-based system using efficient machine learning models to predict cotton crop areas from satellite images. This system utilizes Sentinel-2 MSI surface reflectance images from the years 2020-2021, filtered to include less than 20% cloud cover over Punjab. It focuses on the red (B4), green (B3), and blue (B2) bands for true-colour visualization, with reflectance values ranging from 0 to 2500 and a gamma correction of 1.1, operating at a 10-meter resolution and exporting a maximum of 7,699,027 pixels. Our system feeds satellite images into a machine learning model to accurately identify and calculate the total cotton cultivation area in Punjab. It has demonstrated an accuracy rate of 90-95% in predicting cotton regions, validated against statistical data from the authoritative publication "Crop Area and Production" by the Ministry of National Food Security and Research, Government of Pakistan. By providing detailed insights into growth patterns, this technology empowers farmers to make informed decisions on optimal cotton cultivation practices. This information is essential for enhancing agricultural productivity, sustainability, and economic outcomes.