Assessment of Agricultural Land Changes using Machine Learning: A Case Study of Babil, Central Iraq

Agriculture is a very important sector in the life of human beings as it provides food, bio-based energy, and fiber products. Remote sensing and other geospatial technologies have helped to study agricultural activities comprehensively. This research aimed to mapping of agricultural lands and assess their changes in Al-Hillah, Babylon province, Iraq during 2000-2021. The main datasets used for this research were Landsat TM, ETM, and OLI images. The classification models were based on Random Forest (RF), K Nearest Neighbor (KNN), Artificial Neural Network (ANN), and Support Vector Machine (SVM). These models were optimized using the random search optimization algorithm. The accuracy of the models was assessed using Overall Accuracy (OA), F1-socre, and Kappa index. The results indicated that KNN is the best classification as it performed better than the three models. KNN achieved an OA of 0.954, 0.956, and 0.966 for the image data 2000, 2008, and 2021, respectively. Optimizing models’ hyperparameters yielded better classification accuracies in many occasions except SVM for the image data 2000 and KNN for the image data 2021. The assessment of spatial distribution of urban and agricultural lands showed that urban area growth was centric outwards from the city center and the latter expanded to encompass the surrounding areas from 2000 to 2021.

Assessment of Agricultural Land Changes using Machine Learning: A Case Study of Babil, Central Iraq Keywords:

KNN , Hyper parameters , Landsat , Iraq