ANNOUNCEMENTS
Applications of remote sensing are focused on monitoring and observing agriculture. Accurate agricultural crop mapping and yield prediction is a fundamental tool for sustainable agricultural planning and to ensure food security in regions critically affected by climate change and extreme weather events. In this study, sugarcane field identification and yield estimation have been attempted using two popular machine learning models i.e. Random Forest (RF) and Support Vector Machine (SVM) in Yamunanagar district of Haryana, using the publicly available remote sensing Sentinel-2. The Sugarcane maps produced in this study offers insights into spatial variations in Sugarcane crop extent, growth stages, and the leaf area index (LAI), serving as essential components for precise yield assessment. A two-step procedure was used to accomplish the goal, with RF and SVM classifiers being used to identify the sugarcane fields first. As a crop mask, the most exact outcome which were identified using various band combinations both in a single date image as well as in a time-series mode and states statistical data were employed. Normalised Difference Vegetation Index (NDVI), soil moisture, surface temperature, and rainfall were valuable sources of information for the estimation and prediction of sugarcane crop yield. The yield model was built using the monthly average of standard Moderate Resolution Imaging Spectroradiometer (MODIS) data products, such as the LAI, NDVI, potential evapotranspiration (PET), land surface temperature (LST), enhanced vegetation index (EVI), and evapotranspiration (ET), which was obtained for a five-year period (2020–2024).
The SVM classifier applied on November 2024 imagery, produced the sugarcane crop fields with an overall accuracy of 86% and a kappa coefficient of 0.65 showing accurate estimation of acreage covering 32.8% (99323 acres) of total agriculture land in the district. The highest R2 of 0.70 was obtained for the estimation of sugarcane yield using the random forest regression (RFR) algorithm. The findings demonstrated that Sentinel-2 has a lot of potential and can be a good tool for classifying sugarcane crop in small farms. The study presents the results of classifying, mapping, and estimating acreage of sugarcane through the process of satellite remote sensing following various classification algorithms and identifying most suitable one that could be applicable for a global scale.
Keywords: Remote Sensing, Image Classification, Machine Learning, Sugarcane Mapping, Yield Estimation, RF.
