Influence Of Instrument Noise On The Retrieval Accuracy Of Atmospheric Temperature Profiles From Ultra-Spectral Thermal Infrared Data

The high instrument noise of the ultra-spectral thermal infrared (Ultra-TIR) spectrometers greatly undermines the accuracy of the retrieved results. As such, the effects of instrument error on the retrieved atmospheric temperature profiles from different spectral-resolution data by a statistical regression model are studied in this paper, aiming to guide the development of Ultra-TIR spectrometers. It is unsurprisingly given that the results from Ultra-TIR data yield the best accuracy with same noise equivalent temperature (NE$\Delta$T). Comparisons between the results from Ultra-TIR data ($.05cm^{-1}$) and High-TIR data (0.5 $cm^{-1}$) is further conducted. When the NE $\Delta$T is up to 1.0 K, the RMSE of retrieval profile from the Ultra-TIR data is higher than that from High-TIR data with 0.2 K NE$\Delta$T in the region of middle troposphere, revealing that the advantage of Ultra-TIR data is impaired. When the NE$\Delta$T is lower (0.5 K), the result from Ultra-TIR data becomes better than the one from High-TIR data with 0.2 K NE$\Delta$T.