Evaluation of convective vertical transport properties and their sensitivity to parameterization characteristics

Résumé fourni par la source

Abstract Deep convection parameterizations remain a major source of uncertainty in global circulation models (GCMs). Key aspects such as the convective trigger function and parameterized updraft structure have been widely studied to improve their representation and, subsequently, reduce downstream GCM biases. Parameterization verification is challenging, however, due to limited observations of in-storm convective properties. This work evaluates the ability of several versions of the Kain-Fritsch (KF) convective parameterization to realistically reproduce vertical and temporal structures of convective transport generated via a large eddy simulation (LES). Four KF modifications were used to quantify the sensitivity of convective transport to perturbations across multiple spatial scales. Modifications included: an unmodified control (origKF), use of lightning data assimilation (LDA) to force convective initiation where lightning was observed, use of LDA coupled with cloud top height assimilation (CTHDA), and LDA with halved entrainment (ENT). Results showed that the LDA significantly strengthened convection, increased updraft mass flux by an order of magnitude compared to origKF, and improved agreement with the LES in vertical and temporal structure, though the overall entrainment remained overestimated. Halving entrainment (ENT) reduced updraft mass flux but had limited impact on precipitation, cloud-top height, or entrainment. The impact of the ENT test on updraft mass flux, despite its extreme nature, was of roughly the same magnitude as the impact of using the LDA. The results from this case study suggest that convective trigger properties, represented by LDA and larger-scale processes, exert greater influence on vertical convective transport than storm-scale properties such as entrainment or cloud top height.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.