Urban greening: application of remote sensing and machine learning-based to assess the status of green roofs in district 1, Ho Chi Minh city
Abstract
The rapid urbanization process in Ho Chi Minh City has led to a significant decrease in green spaces, worsening problems such as the urban heat island effect and environmental degradation. However, conventional green space inventories often overlook green roofs — an emerging form of urban greenery. This study aims to evaluate the current status of green roofs in the central area of Ho Chi Minh City, helping to provide a more comprehensive assessment of urban green surfaces. The research utilizes a combination of satellite imagery (Sentinel-2), OpenStreetMap (OSM) vector data, and field survey information as input datasets. A semi-automatic classification process was conducted using the Semi-Automatic Classification Plugin (SCP) in QGIS. Roof features were extracted, and the Normalized Difference Vegetation Index (NDVI) was computed from Sentinel-2 bands. Random Forest (RF) machine learning algorithm was employed to classify rooftops as "green" or "non-green" based on spectral characteristics. The classification reached an overall accuracy of 78.26% with a Kappa coefficient of 0.56, indicating acceptable reliability. Statistical synthesis showed that the total green surface area accounts for 27.42% of the total urban area analyzed (equivalent to 211.74 ha), of which green roofs contribute 3.33% (equivalent to 25.67 ha). In contrast, non-green roofs occupy 25.49% of the total area, showing the potential for future green infrastructure development. This research successfully integrates remote sensing and machine learning to enhance the mapping of urban green spaces, particularly green roofs — a part often not included in existing databases. The results provide a scientific basis for planning green infrastructure and promoting urban sustainability. Future research may focus on improving classification accuracy by integrating higher-resolution imagery, and assessing the ecological benefits of rooftop greening at a finer scale.