11:00 AM - 1:00 PM
[PEM09-P04] Evaluation of vertical ozone profile derived from mm-wave observations at Syowa Station and development of observation method and data analysis program for the new spectrometer for multi-line observation
Keywords:mm-wave spectroscopy, Ozone, Antarctica
Since 2011, we have been continuously observing trace molecules such as nitric oxide (NO) and ozone (O3) using a superconductive millimeter-wave spectrometer at Syowa Station, Antarctica, in order to clarify the influence of solar activity on the middle atmosphere. In addition, a 2nd-generation millimeter-wave spectrometer with a multi-channel superconducting receiver and a broadband digital spectrometer were installed in 2020 for simultaneous observations of multiple emission lines including nitric oxide and ozone.
On the other hand, the data processing, particularly the retrieval analysis of vertical ozone profile, has been somewhat delayed. The ozone volume mixing ratio (VMR) derived from the observed data at Syowa were more variable than those at Rikubetsu, where ozone observations have been carried out for more than 20 years. In this study, (1) we reviewed the data analysis procedure for 1st-generation millimeter-wave spectrometer to improve the data for the past ~10 years, and (2) we developed a new analysis program for the data obtained by a new frequency-switching method with the 2nd-generation millimeter-wave spectrometer.
As for (1), we found that the main reason of large fluctuation of VMR is related with the atmospheric correction for the spectral intensity by the optical thickness of lower atmosphere. There was a correlation between the optical thickness and the fluctuation of VMR suggesting that the present procedure does not properly compensate for the atmospheric absorption. In addition, in case that spatial non-uniformity of optical thickness (i.e., condition that the cloud thickness varies greatly in the sky direction due to bad weather), it was difficult to compensate for the large difference in optical thickness between the two directions of the elevation-switching. In order to choose appropriate dataset of ozone spectra, we used two types of scaling factors. One is called as spectral scaling factor (SPSF) corresponding to the intensity ratio between the model spectrum calculated from the typical vertical ozone profile used as an a-priori in the retrieval analysis and the actual observed spectrum. The other scaling factor is ozone-sonde scaling factor (OSSF) calculated from the vertical ozone profile obtained from the ozone sondes conducted by the Japan Meteorological Agency at Syowa Station. By combining SPSF and OSSF, we empirically derived the screening conditions to select properly calibrated spectral data.
As for (2), we had previously performed a procedure called "reform" (e.g., Nagahama et al. 1998) to combine the two data before (=on) and after (=off) switching the frequency to obtain a single spectral data, and then applied the retrieval analysis to the reformed spectra. However, since two ozone spectra were adjacent to each other (about 180 MHz apart) in the 2nd-generation spectrometer, we did not reform from the on- and off data. Instead of reform, the forward model was modified to reproduce the spectra for frequency switching observations, and the retrieval analysis was applied to the non-reformed observation data. In frequency switching, unlike altitude-angle switching, we are observing the sky in the same direction, and thus are less affected by the non-uniformity of the sky. In the study of the effects of energetic particle precipitations associated with solar activity, it is important to reduce the short-term variability (i.e., standard deviation) because ozone variations within a few days must be evaluated with sufficient accuracy.
At present, based on the comparison with the satellite instrument AURA/MLS, the standard deviation of the 1st-generation ozone data in (1) has been improved from less than 22% to less than 17% in the range of 21 - 55 km by using new screening conditions. The analysis of the frequency switching data in (2) shows that the standard deviation has been reduced to less than 10% in the range of 33 - 60 km. As the next step, we plan to further improve the accuracy by further improving the correction method of optical thickness using empirical relationship. In this presentation, we will report and discuss the accuracy of the analysis results achieved at that time.
On the other hand, the data processing, particularly the retrieval analysis of vertical ozone profile, has been somewhat delayed. The ozone volume mixing ratio (VMR) derived from the observed data at Syowa were more variable than those at Rikubetsu, where ozone observations have been carried out for more than 20 years. In this study, (1) we reviewed the data analysis procedure for 1st-generation millimeter-wave spectrometer to improve the data for the past ~10 years, and (2) we developed a new analysis program for the data obtained by a new frequency-switching method with the 2nd-generation millimeter-wave spectrometer.
As for (1), we found that the main reason of large fluctuation of VMR is related with the atmospheric correction for the spectral intensity by the optical thickness of lower atmosphere. There was a correlation between the optical thickness and the fluctuation of VMR suggesting that the present procedure does not properly compensate for the atmospheric absorption. In addition, in case that spatial non-uniformity of optical thickness (i.e., condition that the cloud thickness varies greatly in the sky direction due to bad weather), it was difficult to compensate for the large difference in optical thickness between the two directions of the elevation-switching. In order to choose appropriate dataset of ozone spectra, we used two types of scaling factors. One is called as spectral scaling factor (SPSF) corresponding to the intensity ratio between the model spectrum calculated from the typical vertical ozone profile used as an a-priori in the retrieval analysis and the actual observed spectrum. The other scaling factor is ozone-sonde scaling factor (OSSF) calculated from the vertical ozone profile obtained from the ozone sondes conducted by the Japan Meteorological Agency at Syowa Station. By combining SPSF and OSSF, we empirically derived the screening conditions to select properly calibrated spectral data.
As for (2), we had previously performed a procedure called "reform" (e.g., Nagahama et al. 1998) to combine the two data before (=on) and after (=off) switching the frequency to obtain a single spectral data, and then applied the retrieval analysis to the reformed spectra. However, since two ozone spectra were adjacent to each other (about 180 MHz apart) in the 2nd-generation spectrometer, we did not reform from the on- and off data. Instead of reform, the forward model was modified to reproduce the spectra for frequency switching observations, and the retrieval analysis was applied to the non-reformed observation data. In frequency switching, unlike altitude-angle switching, we are observing the sky in the same direction, and thus are less affected by the non-uniformity of the sky. In the study of the effects of energetic particle precipitations associated with solar activity, it is important to reduce the short-term variability (i.e., standard deviation) because ozone variations within a few days must be evaluated with sufficient accuracy.
At present, based on the comparison with the satellite instrument AURA/MLS, the standard deviation of the 1st-generation ozone data in (1) has been improved from less than 22% to less than 17% in the range of 21 - 55 km by using new screening conditions. The analysis of the frequency switching data in (2) shows that the standard deviation has been reduced to less than 10% in the range of 33 - 60 km. As the next step, we plan to further improve the accuracy by further improving the correction method of optical thickness using empirical relationship. In this presentation, we will report and discuss the accuracy of the analysis results achieved at that time.