TANAKA Taichu Y.
- Meteorological Society of Japan
- 気象集誌. 第2輯 (ISSN:00261165)
This study evaluated the impact of a future space-borne Doppler wind lidar (DWL) on a super-low-altitude orbit using an observing system simulation experiment (OSSE) based on a sensitivity observing system experiment (SOSE) approach. Realistic atmospheric data, including wind and temperature, was provided as “pseudo-truth” (PT) to simulate DWL observations. Hourly aerosols and clouds that are consistent with PT winds were also created for the simulation. A full-scale lidar simulator, which is described in detail in the companion paper, simulated realistic line-of-sight wind measurements and observation quality information, such as signal-to-noise-ratio (SNR) and measurement error. Quality control (QC) procedures in the data assimilation system were developed to select high-quality DWL observations based on the averaged SNR from strong backscattering in the presence of aerosols or clouds. Also, DWL observation errors used in the assimilation were calculated using the measurement error estimated by the lidar simulator.
The forecast impacts of DWL onboard polar- and tropical-orbiting satellites were assessed using the operational global data assimilation system. Data assimilation experiments were conducted in January and August in 2010 to assess overall impact and seasonal dependence. It is found that DWL on either polar- or tropical-orbiting satellites is overall beneficial for wind and temperature forecasts, with greater impacts for the January experiments. The relative forecast error reduction reaches almost 2 % in the tropics. An exception is a degradation in the southern hemisphere in August, suggesting a need to further refine observation error assignment and QC. A decisive conclusion cannot be drawn of the superiority of polar- or tropical-orbiting satellites due to their mixed impacts. This is probably related to the characteristics of error growth in the tropics. The limitations and possible underestimation of the DWL impacts, for example due to a simple observation error inflation setting, in the SOSE-OSSE are also discussed.