This paper proposes a self-developed multi-channel ground-based microwave radiometer that uses a brightness temperature and data inversion method. We compared the radiometer data and the radiosonde observations

and then analyzed the radiometer observations in terms of brightness temperature

inversion temperature

and water vapor profile accuracy. Results show that the microwave radiometer has a small observation error of brightness temperature and the neural network inversion profiles of atmospheric temperature

water vapor

and other parameters are accurate and reliable. The proposed radiometer has practical applications.