Abstract: To better meet the field radiometric calibration requirements of the thermal infrared band satellite remote sensor, and to improve the calibration frequency, timeliness, and measurement accuracy of the field radiometric calibration, a Multi-channel Self-calibration Thermal Infrared Radiometer (MSTIR) with automatic observation capabilities has been developed for measuring the atmospheric downward radiance and ground surface radiance in the field. In this paper, the structure, working principle, and test results of the MSTIR were introduced, and field experiments are carried out on this basis. Taking the observation data of the Golmud calibration field in Qinghai Province as an example, combined with the temperature and emissivity separation algorithm of the multi-channel data of the site, the land surface temperature, and channel emissivity results of the calibration site were obtained, and the uncertainty was evaluated. The data results show that the maximum relative standard deviation of channel emissivity obtained by the four spectral channels of the MSTIR is 0.012 at maximum, and the field temperature deviation is within 0.11 K. The developed MSTIR can meet the application requirements of thermal infrared band field calibration, which lays a foundation for the application of site automatic radiometric calibration of thermal infrared band remote sensor.