Quantifying noise complaint intensity via multi-LLM and revealing vertical greenery mitigation thresholds

Area of Science:

• Environmental Science
• Urban Planning
• Artificial Intelligence

Background:

• Urban noise pollution significantly impacts residents' quality of life.
• Existing research has limitations in assessing noise sensitivity and green space mitigation effects.
• Quantifying noise complaint intensity and understanding its relationship with urban green space is crucial.

Purpose of the Study:

• To develop an LLM-based method for quantifying noise complaint intensity.
• To investigate the impact of different types of green space on urban noise.
• To identify effective strategies for noise mitigation in urban environments.

Main Methods:

• Utilized large language models (LLMs) to create a benchmark dataset for noise complaint intensity.
• Employed a fine-tuned RoBERTa model, achieving 61.48% accuracy in classifying complaint intensity.
• Integrated NDVI from Sentinel-2 imagery and 3D Green View Index from street view images to measure green space exposure.

Main Results:

• The LLM approach enabled fine-grained analysis and large-scale distribution mapping of resident complaint intensity.
• Overhead greenspace (NDVI) and grass coverage showed a positive correlation, with low-lying vegetation intensifying noise complaints at low exposure.
• Tall vertical greenery (trees) demonstrated significant noise suppression, with diminishing marginal benefits above 0.12 coverage. Road density exhibited a "suppression then amplification" pattern.

Conclusions:

• Tall vertical green spaces are crucial for mitigating urban noise pollution and protecting vulnerable populations.
• Urban planning should prioritize the strategic implementation of high-level green infrastructure exceeding identified thresholds.
• The study provides a novel methodology for analyzing noise complaints and evaluating green space effectiveness.