5:15 PM - 7:15 PM
[AAS02-P02] Observing system simulation experiment for tropical cyclone prediction using SAMRAI
Keywords:Data assimilation, Numerical weather prediction, Observing system simulation experiment, Tropical cyclone
The improvement of numerical weather prediction is essential for management of weather disaster. In the numerical weather prediction, obtaining better initial values is an important issue. Although satellites are powerful tool for observing atmospheric condition, conventional sensor has missing values due to the radio frequency interference (RFI). The new microwave radiometer called SAMRAI, which is under development by JAXA, has higher spatial, temporal and frequency resolution compared to conventional systems and technologies. SAMRAI will provide observation without missing values from the RFI because of high frequency resolution. This study investigates the potential of SAMRAI for tropical cyclone (TC) forecast by the observing system simulation experiment (OSSE). We conducted a forecast experiment which assimilates sea surface winds and column-integrated water vapor over the sea retrieved by SAMRAI and compared with forecast without SAMRAI observation to assess the impacts of SAMRAI.
In this OSSE, we chose 18-km grid spacings although it was coarser than observation. We applied local ensemble transform Kalman filter (LETKF) as a data assimilation system. The SAMRAI-observed variables were assimilated under a horizontal localization length of 400 km and a vertical localization length of 5 km. We performed forecast experiments using mean of analysis from 30-member data assimilation. Results suggested that SAMRAI improved a 5-day forecast of track and intensity though it depended on the initial date. The future issues are to increase horizontal resolution and the number of data assimilation cycles.
In this OSSE, we chose 18-km grid spacings although it was coarser than observation. We applied local ensemble transform Kalman filter (LETKF) as a data assimilation system. The SAMRAI-observed variables were assimilated under a horizontal localization length of 400 km and a vertical localization length of 5 km. We performed forecast experiments using mean of analysis from 30-member data assimilation. Results suggested that SAMRAI improved a 5-day forecast of track and intensity though it depended on the initial date. The future issues are to increase horizontal resolution and the number of data assimilation cycles.