In industrial applications, the large comprehensive wireless channel impulse response (CIR) reference dataset, measured by National Institute of Standards and Technology (NIST), has been a useful tool for understanding propagation within factory environments. The NIST CIR reference dataset is obtained using a precision channel sounder instrument where transmitter and receiver are time-synchronized by two rubidium clocks. While the accuracy of the NIST CIRs is much higher than the CIRs measured by general commercial digital receiver, two types of system errors have been discovered within the dataset from the perspective of signal processing. These errors are significant for wireless localization, physical layer security, and related applications. To calibrate the CIR, two channel sounder error calibration methods (CSEC) is proposed: the CSEC based on phase compensation and carrier frequency offset recovery. Our results reveal that the CSEC method can improve the accuracy of the CIR to the accuracy that precise instruments cannot achieve. To demonstrate the consequence of these systemic errors, a case study involving physical layer authentication is investigated showing a marked improvement in authentication accuracy after the systemic errors in the dataset are removed. Moreover, the CSEC method may be used to correct other CIR datasets with similar systemic errors.