资源环境科技发展态势分析平台

Global climate models fail to represent the annual cycle of tropical-extratropical cloud bands and produce too much summer rainfall over subtropical southern Africa. This study demonstrates that running a regional convective-permitting climate simulation alleviates these biases, counteracting biases that are present in the parent model. The improvement emerges from stronger vertical mass flux in the tropics, which forces a stronger local Hadley overturning into the summer hemisphere. This enhanced overturning increases upper-level subsidence in the subtropics and amplifies the forcing of the local subtropical jet. Together, these improvements halve the wet subtropical rainfall bias and are associated with a 50% increase to an 80% match between the simulated and observed annual cycle of crucial tropical-extratropical cloud band rainfall systems. The results advocate for the increased use of convective-permitting climate models with domains that include regional tropical convection hot spots, in order to fully benefit from the explicit representation of deep convection.

Plain Language Summary Climate models, both global and regional, approximate the development and effect of thunderstorms on the simulated circulation. The inadequacy of these approximations is widely thought to produce inaccuracies in simulated regional climates. These errors are particularly severe over southern Africa where model simulations produce too much rain and poorly represent the annual cycle of the dominant summer rainfall systems. In this study, a pan-African regional climate simulation is run for 10 years at a grid spacing that allows direct simulation of widespread outbreaks of thunderstorms, breaking the model dependence on inadequate approximations. This convection-permitting model produces a substantially more realistic regional climate over subtropical southern Africa with the main rainfall systems occurring at the right time of year and only a moderate overestimate of monthly summer rainfall. This result is particularly striking because this regional model overcomes errors that are present in the global model which provides the boundary conditions to its pan-African domain.