The real-world monitoring system of air pollution ordinarily collects data about pollutant concentration levels at pollution sources and monitors stations in a high-frequency manner. Inspired atmospheric models, the meteorological conditions could play an important role in building up the data-driven model for dispersing atmospheric pollutants from pollution sources to monitor stations. We propose a varying-coefficient model to analyze how emissions of monitor stations are influenced by pollution sources with changing with the wind speed. To estimate the unknown coefficient curves, we use a spline basis to approximate the functions. The asymptotic properties of the proposed method are studied and show the consistency of the estimator. Inference procedures based on a resampling subject bootstrap is developed to construct the conservative confidence bands. A simulation study is carried out to demonstrate the performance of our method. Illustrated by a real-world dataset of environmental sensors collected in Shenyang, China, the proposed varying-coefficient model reveals that the wind speed changes the dispersion mechanism of atmospheric pollutants between monitor stations and pollution sources.