Content area
Abstract
Although accurate weather and climate prediction beyond one to two weeks is of great value to society, the skill of such extended prediction is limited in current operational global numerical models, whose coarse horizontal grid spacing necessitates the parameterization of atmospheric processes. Of particular concern is the parameterization of convection and specifically convection in the tropics, which impacts global weather at all time scales through atmospheric teleconnections. Global convection-permitting models (CPMs), which forego convective parameterization by explicitly resolving cumulus-scale motions using fine (1–4 km) horizontal grid spacing, can improve operational global weather prediction at extended time scales by more faithfully simulating tropical convection and associated teleconnections. Beyond this, global CPMs are also useful tools for studying processes important for simulating realistic tropical variability.
Month-long simulations targeting four Madden-Julian Oscillation events made with several global model configurations are verified against observations not only with the above operational and scientific goals in mind, but also to assess the roles of grid spacing and convective parameterization in global models. Specifically, the performance of a global CPM configuration with a uniform 3-km mesh is compared to that of a global 15-km configuration with and without convective parameterization, and of a variable-resolution “channel” simulation using 3-km grid spacing in the tropics and 15-km in the extratropics with a scale-aware convection scheme over the entire domain.
It is shown that global 3-km simulations produce realistic tropical precipitation statistics, except for an overall wet bias and delayed diurnal cycle. The channel simulation performs similarly, though with an unrealistically high frequency of heavy rain. The 15-km simulations with and without cumulus schemes produce too much light and heavy tropical precipitation, respectively. A deeper look at the environmental sensitivity of simulated precipitation reveals that the 15-km model with a cumulus parameterization triggers convection under unrealistically stringent environmental stability and moisture conditions; specifically, convection in the 15-km configuration is unrealistically dependent on environmental stability and lacks nonlinear sensitivity to moisture variations. These problems are alleviated in the global CPM configuration.
Only the global CPM configuration is able to capture eastward-propagating Madden-Julian Oscillation events, while the 15-km runs favor stationary or westward-propagating convection organized at the planetary scale. A case study of a convectively coupled Kelvin wave during the first simulation case is conducted to identify the instability/coupling mechanisms necessary for such waves and other important tropical convective features. Results suggest that the 15-km simulation lacks strong wave-convection coupling because of its aforementioned lack of sensitivity to moisture. The 3-km simulation realistically simulates the nonlinear relationship between moisture and precipitation—both in its mean state and throughout the life cycle a convectively coupled Kelvin wave.
The global 3-km CPM exhibits the highest extratropical forecast skill aloft and at the surface (particularly during week-3), consistent with its superior MJO representation. While more cases are needed to statistically verify these results, this study highlights the benefits of using global CPMs for identifying flaws in current-generation models, examining tropical convection and its relationship to environmental factors, and improving subseasonal forecasting. Furthermore, results show that alternatives to global convection-permitting resolution (using coarser resolution or a scale-aware channel configuration) exhibit drawbacks that reduce their predictive skill.
