Assessing the Accuracy of Random Forest in Mapping Urban Green Cover in Baguio City Using Sentinel-2 Imagery and Spectral Indices

Abstract

Urban green cover plays a pivotal role in sustainable urban development by providing environmental and socio-economic benefits. Accurate mapping of urban green cover is essential for developing urban greening strategies and managing urban green spaces in smart cities. Remote sensing, particularly Sentinel-2 imagery, offers high-resolution multispectral data suitable for vegetation analysis. Machine learning algorithms, such as Random Forest, have proven effective in classifying land cover, including urban green spaces. This research investigates the accuracy of Random Forest in mapping urban green cover in Baguio City, Philippines, utilizing Sentinel-2 imagery and spectral indices. The study utilized spectral indices, such as NDVI, SAVI, NDWI, and NDBI to train and validate the Random Forest model. The performance of the Random Forest classifier is evaluated using standard accuracy assessment metrics, such as overall accuracy, producer's accuracy, user's accuracy, F1-score, and Kappa coefficient. The Random Forest model proved to enhance the classification of urban green cover with 85.71%, 86%, and 86% for the overall accuracy, producer’s accuracy for urban green cover, and consumer’s accuracy for urban green cover, respectively. The F1-score of 0.92308 and Kappa coefficient of 0.7790927 showed that the Random Forest model accurately classified the urban green cover. By utilizing remote sensing and machine learning techniques, this research seeks to contribute to the development of accurate and up-to-date urban green cover maps. The findings shall provide valuable insights for urban planners and policymakers in Baguio City, enabling them to implement effective strategies for urban greening and sustainable urban development.