Abstract: Radiosonde data assimilation can improve the simulation of boundary layer characteristics, but the effect of assimilation frequency on boundary layer simulations still needs further study. This paper uses 3-hourly high-resolution radiosonde data for assimilation experiments to explore how assimilation frequency affects the accuracy of urban heat island (UHI) simulations. It also evaluates improvements in the simulation of UHI, boundary layer meteorological features, and daily variations of meteorological elements. The results show that increasing assimilation frequency significantly improves the simulation of UHI spatial distribution, intensity, and circulation patterns. The 3-hourly assimilation achieves the best accuracy and detail. The initial field without assimilation contains systematic errors, which are corrected by radiosonde data assimilation. The 3-hourly assimilation frequency matches the forcing timescale of urban meteorological fields, improving boundary layer simulations and the accuracy of daily temperature variations. High-frequency assimilation of radiosonde data shows significant potential for improving urban weather forecasting and UHI