09:00 〜 09:15
[AAS09-01] 気象庁地上マイクロ波放射計観測網による気象監視・予測技術高度化の研究
★招待講演
キーワード：地上マイクロ波放射計、水蒸気、大雨
In Japan, landslides and flash floods are often caused by quasi-stationary convective bands (QSCBs) during the rainy season, and many previous studies have pointed out the importance of large amounts of water vapor supply in the lower atmosphere in the formation of QSCBs. Therefore, the Japan Meteorological Agency (JMA) has been enhancing water vapor observations in order to improve the monitoring and forecasting techniques for QSCBs and to elucidate the mechanism of QSCBs. As part of this effort, we have established a network of ground-based microwave radiometers (MWRs) and developed a system for real-time monitoring using data assimilation methods.
First, RPG-HATPRO-G5 was adopted for this MWR observation network. This MWR observes brightness temperature at intervals of 1 second to several minutes with 7 channels in the 22 to 31 GHz band sensitive to water vapor and cloud water, and 7 channels in the 51 to 58 GHz band sensitive to oxygen. We have installed MWRs at 17 stations, mainly in western Japan where heavy rainfall due to QSCBs and typhoons is likely to occur. Observation sites were chosen where the JMA had already installed wind profilers. This allows us to obtain atmospheric thermodynamic profiles of air temperature and water vapor density and dynamic profiles of wind at the same time.
We have developed a vertical one-dimensional variational data assimilation method (1DVAR) that combines numerical weather model results with MWR data and verified that its accuracy is better than that of neural network, which is widely used in the world, and model results (Araki et al., 2015). In this study, a new 1DVAR method was developed at JMA based on MRI's 1DVAR. In addition to MWR data, this method assimilates JMA surface meteorological observations of temperature and water vapor and wind data from wind profilers. The results of the JMA operational mesoscale model were used as the first estimate, and the 1DVAR analysis was performed every 10 minutes in real time at all MWR stations. We compared it with the results of the July-August 2022 sonde observations at Naze (Kagoshima pref.). From the result, it is found that the 1DVAR estimation overperform the model output for both temperature and water vapor profiles. This confirms that 1DVAR with MWR data can be used to obtain highly accurate temperature and water vapor profiles.
A case study was performed on the QSCB event on 10 July 2023 by using 1DVAR analysis. Water vapor flux increased below the altitude of 1km about 12 hours before the formation of the QSCB, and water vapor mixing ratio reached over 20 g/kg, which was significantly large value compared with the recent QSCB events. The MWR data was also applied into the data assimilation and numerical prediction for the heavy rainfall events. The result showed that the brightness temperature, 1DVAR profiles, precipitable water vapor obtained by the MWR contributed to improving the accuracy of heavy rainfall forecasts, including the events caused by QSCBs. The MWR observations will be applied to phenomena other than heavy rainfall to improve monitoring and forecasting techniques
First, RPG-HATPRO-G5 was adopted for this MWR observation network. This MWR observes brightness temperature at intervals of 1 second to several minutes with 7 channels in the 22 to 31 GHz band sensitive to water vapor and cloud water, and 7 channels in the 51 to 58 GHz band sensitive to oxygen. We have installed MWRs at 17 stations, mainly in western Japan where heavy rainfall due to QSCBs and typhoons is likely to occur. Observation sites were chosen where the JMA had already installed wind profilers. This allows us to obtain atmospheric thermodynamic profiles of air temperature and water vapor density and dynamic profiles of wind at the same time.
We have developed a vertical one-dimensional variational data assimilation method (1DVAR) that combines numerical weather model results with MWR data and verified that its accuracy is better than that of neural network, which is widely used in the world, and model results (Araki et al., 2015). In this study, a new 1DVAR method was developed at JMA based on MRI's 1DVAR. In addition to MWR data, this method assimilates JMA surface meteorological observations of temperature and water vapor and wind data from wind profilers. The results of the JMA operational mesoscale model were used as the first estimate, and the 1DVAR analysis was performed every 10 minutes in real time at all MWR stations. We compared it with the results of the July-August 2022 sonde observations at Naze (Kagoshima pref.). From the result, it is found that the 1DVAR estimation overperform the model output for both temperature and water vapor profiles. This confirms that 1DVAR with MWR data can be used to obtain highly accurate temperature and water vapor profiles.
A case study was performed on the QSCB event on 10 July 2023 by using 1DVAR analysis. Water vapor flux increased below the altitude of 1km about 12 hours before the formation of the QSCB, and water vapor mixing ratio reached over 20 g/kg, which was significantly large value compared with the recent QSCB events. The MWR data was also applied into the data assimilation and numerical prediction for the heavy rainfall events. The result showed that the brightness temperature, 1DVAR profiles, precipitable water vapor obtained by the MWR contributed to improving the accuracy of heavy rainfall forecasts, including the events caused by QSCBs. The MWR observations will be applied to phenomena other than heavy rainfall to improve monitoring and forecasting techniques