Evaluation of Doppler Velocity in a GSRM Using the EarthCARE Satellite with Implications for Improving Model Cloud Microphysics

Conventional climate and numerical weather prediction models have long relied on empirical parameterizations of hydrometeor fall speeds, which have not been comprehensively validated on the global scale due to a lack of their global observations. Nevertheless, fall-speed parameters strongly influence model performance and are often subject to tuning. For example, Takasuka et al. (2024) showed that modifying the fall speeds of snow and rain improves the representation of both climate-scale statistics and intraseasonal variability. However, such tuning is not directly constrained by observations; instead, parameter values are selected to best reproduce large-scale climate fields and disturbances.

Notable in this regard is the recent emergence of the EarthCARE satellite, launched in late May of 2024, which provides the first-ever global observations of the vertical motion of hydrometeors from space. In this study, we compare representative fall-speed parameter settings in the Nonhydrostatic Icosahedral Atmospheric Model (NICAM) against EarthCARE observations. We use a single-moment cloud microphysics scheme (Tomita, 2008; Roh and Satoh, 2014) with two configurations. One employs the tuned fall-speed parameters proposed by Takasuka et al. (2024), while the other follows the original parameterization used in Kodama et al. (2021). The Takasuka et al. (2024) configuration prescribes slower fall speeds for both snow and rain than the Kodama et al. (2021) setting. To enable a consistent comparison with EarthCARE, EarthCARE-like observables are generated using the Joint Simulator for Satellite Sensors (Hashino et al., 2013) and evaluated against satellite measurements.

The results show that the Takasuka et al. (2024) configuration produces snow and rainfall fall speeds that are closer to EarthCARE observations than those obtained with the Kodama et al. (2021) setting, although it tends to overestimate radar reflectivity. In addition, the Takasuka configuration is confirmed to better reproduce deep convective characteristics. Our analysis also identifies several issues that require further refinement of the cloud microphysics scheme, including the representation of weak precipitation and the temperature dependence of snowfall terminal velocity. These results highlight an added value of unprecedented measurement information from EarthCARE Doppler capability that points to a possible area of further improvement of model microphysics in GSRMs at a process level.