Pantnagar Journal of Research

Abstract

One crucial use of remote sensing in agriculture is crop type mapping and area estimation. Recently, crop categorization has seen a rise in the use of machine-learning classification approaches. Google Earth Engine (GEE), a platform where users can explore a variety of satellite data sets without even downloading the satellite data and provides several powerful classification techniques. The main objective of this study is to explore the ability of different machine learning classification techniques like Random Forest (RF), K-Nearest Neighbour (KNN), Classification And Regression Trees (CART), Support Vector Machine (SVM), and Minimum Distance (MD) for crop classification using Google Earth Engine (GEE). Sentinel-2, MSI (10 m) Level-2A_SR dataset was used for crop type mapping and sugarcane area estimation for Udham Singh Nagar district for the year 2022-2023. Normalized difference vegetation index (NDVI) image composite were created at less than 10% cloud cover and using this we can easily identify the sugarcane area from other areas like other crops, forest, water bodies and buildings. Post classification, accuracy assessment analysis was done through the generation of the confusion matrix (producer and user accuracy), and F1 score. The results indicate that using GEE through a cloud platform, satellite data accessing, filtering, and pre-processing of satellite data can be performed very efficiently. In terms of overall classification accuracy and F1 score, RF (97.88%, 97.31%) classifier performed better than SVM (95.61%, 94.42%), MD (94.96, 93.60%), KNN (94.47%, 92.98%), and CART (90.57%, 88.01%) classifiers in the respective growing season of sugarcane.