The nighttime urban environment is increasingly affected by various artificial light at night (ALAN), and color temperature, as a significant characteristic of light, has important applications in many fields and industries. Particularly, in recent years, the widespread adoption of Light-Emitting Diode (LED) light, a low-carbon technology, has led to the extensive utilization with varying color temperatures across diverse settings. However, it is important to note that different color temperatures have various impacts on human health and ecological systems. Thus, information regarding spatial distribution and composition of nighttime light (NTL) with different color temperatures is essential for formulating sustainable strategies that balance nighttime public security, energy consumption, and ecosystem conservation. To address this challenge and meet the demand, we propose a color temperature based light source classification system based on UAV NTL images, employing an object-oriented classification method to classify lights into High Pressure Sodium (HPS), Warm LEDs, Cool LEDs, and Colored LEDs. Additionally, considering the impact of flight altitude on classification accuracy, we classify lights from seven distinct altitudes and evaluate their accuracy at each level. Results showed the following. (1) Optimal classification accuracy was attained at a flight altitude of 350 meters, boasting an overall accuracy of 95.7% and a Kappa coefficient of 0.947. (2) Spectral features were identified as the most influential classification attributes, contributing over 70% at all altitudes, followed by the textural and the geometric features, which had the least impact. (3) The methodology demonstrated strong performance and adaptability in varying urban contexts, as indicated by an off-site application accuracy of 89.8% and a kappa coefficient of 0.873. This study represents the first attempt to identify NTL types by color temperature, providing a new perspective for urban lighting planning and light pollution management.

Nighttime light (NTL); Color temperature; Unmanned aerial vehicle (UAV); Machine learning; Light types classification