Noise Characterization and Mitigation for the Microwave Sounding Instruments

Observations from microwave sounding instruments are important data source for NWP applications. Study of the SNPP ATMS data in ECMWF shows the increased channel correlation in ATMS when compare with the legacy AMSU instrument, which is linked to the striping effects in the dataFor NWP users, presenting of the striping noise will increase the channel correlation, lead to a detrimental assimilation of observations affected by such error correlations. The presence of notable error correlations are need to be properly characterized for a successful assimilation of the observations. To make the microwave sounding data become more useful, the noise will need to be addressed in ground data processing. In this paper, the root cause of the different noise in the instrument is investigated. Different methods are used to quantitatively characterize the noise feature of the instrument. Based on the study for the noise power spectral density (PSD) of the measurements, impacts of different calibration parameters on instrument noise are investigated. A mitigation algorithm for noise reduction is proposed for on-orbit calibration processing. Preliminary results show that by improving the performance of the calibration for 1/f noise rejection, the striping noise in the calibrated measurements can be largely mitigated. In future, study work will focus on validating the results for future operational applications.