Integrated Framework for Light Pollution Assessment and Mitigation: A Multifaceted Approach with TOPSIS and Machine Learning

Abstract

In this study, we propose the development of an evaluation system to quantify the degree of light pollution and offer three strategies for mitigating its negative effects. Our research identifies 4 distinct types of cities and employs a 16 key indicators to assess the level of light pollution risk, forming the basis for the construction of the Light Pollution Assessment (LPA) Model. Utilizing the TOPSIS and EWM methods, we ranked each city and determined a composite score interval, categorizing pollution levels from I to IV. Our evaluation system designates level I with an overall score of 0.7466, establishing it as a benchmark for assessing the risk level of light pollution. Recognizing that certain factors influencing LPA levels may not directly contribute to light pollution, we introduce the ML-based Strategy Evaluation (MLE) method, this approach classifies 11 out of 16 indices based on their relevance to light pollution. Our analysis reveals that the number of streetlights, the range of ULOR (Upward Light Output Ratio), and the presence of natural protected areas are the top three factors affecting light pollution. With an accuracy of 79.1%, the MLE method demonstrates a strong alignment with the data. Based on our findings, we propose three targeted strategies focusing on luminance regulation, greenery implementation, and legal enhancements to address and alleviate light pollution challenges in urban environments. This research contributes a comprehensive assessment framework and actionable strategies for policymakers and urban planners to effectively manage and reduce light pollution.