Enabling Systematic Modulation of Deep Convective Systems in Kilometer-scale Models using a Unified Cumulus Parameterization

The representation of deep moist convection and organized convective systems in numerical modeling of the atmosphere is pivotal to model fidelity. Although kilometer-scale global models better capture large-scale oscillations and convection features that traditional general circulation models struggle to represent, past studies have reported notable differences in their simulations of convective organization, especially over the tropics. However, untangling the controlling factors of convective organization remains challenging, since physical processes associated with deep moist convection are intrinsically multiscale and deeply intertwined. To address this challenge, this study develops a cumulus parameterization tailored for kilometer-scale models, aiming to enable the modulation and systematic testing of multiscale interactions associated with deep moist convection.

The cumulus parameterization developed in this study employs Arakawa's unified parameterization to represent the interactions of unresolved deep moist convection with its environmental flow, including the explicitly simulated convection. The underlying parameterizability of unresolved deep moist convection follows the notion of convective quasi-equilibrium, while a complementary closure is employed to predict the fractional area covered by cumulus updrafts and adjust the local vertical eddy transports accordingly. A representation of stochasticity and convection memory is introduced by coupling our parameterization with a cellular automaton. Empirical values and physical assumptions are used to establish our parameterization as a prototype subject to designing systematic experiments of specific process representation in the future.

Idealized experiments of tropical maritime deep convection at a horizontal grid spacing of 3 km demonstrate that employing our parameterization in convection-permitting simulations modulates deep convective system features while generally retaining the profile of total vertical energy transport associated with deep convection. Unresolved deep convection dominates the vertical energy transport in the early stages of deep convective systems, and its contribution gradually weakens as systems develop. Meanwhile, detrainment from unresolved deep convection leads to the early occurrence of cumulus congestus and cumulonimbus. In comparison with the ordinary convection-permitting simulation, our parameterization exhibits a pronounced congestus mode in the vertical velocity profile of mature convective systems, partially due to the enhanced dry static stability. Overall, simulations with our parameterization exhibit fewer and larger short-lived convective systems. Further investigation into the source of the uncertainty in convective organization is warranted.