Layer inflow into precipitating convection over the western tropical Pacific

Abstract A conceptual model of tropical convection frequently used in convective parametrization schemes is that of a parcel process in which boundary‐layer air, characterized by high equivalent potential temperature, ascends to great heights in convective updraughts, while air above the planetary boundary layer with lower equivalent potential temperature mixes into convective downdraughts and sinks. However, airborne Doppler‐radar data show that organized deep convective systems over the western tropical Pacific warm pool are often characterized by layers of ascending inflow ∼0.5–4 km in depth. These inflow layers do not consist merely of boundary‐layer air. In this study a high‐resolution numerical cloud model is employed to investigate these inflow layers. Input data are from the Tropical Ocean–Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE). Two time periods are selected in December 1992, which represent the onset and peak of a strong westerly phase of the intraseasonal oscillation. Model simulations for 14 December, representative of westerly onset conditions and the beginning of a convectively active period, and 23 December, representative of strong low‐level westerlies and extremely widespread convection, are conducted. To bridge the coarse resolution of the global‐model analysis fields and the fine resolution of the cloud model, hourly output from a mesoscale model is used to supply initial and lateral boundary conditions for the cloud model. Control simulations of 14 and 23 December reveal distinct convective organizations and, specifically, inflow characteristics for the two systems embedded in different large‐scale environmental conditions. In the case of 23 December, convection simulated under conditions of the strong westerly wind and near‐saturated low‐mid troposphere exhibits deep inflow layers. Trajectories computed from the simulation of 23 December under these conditions show a strong layer‐lifting inflow signal. In contrast, the control simulation of 14 December shows a parcel‐like inflow with only the air in the lower part of the inflow actually rising in the deep convective updraughts. One of the main differences between the two simulations is the lack of a deep environmental moist layer in the 14 December case. The control simulation did not capture well the extent of precipitating mesoscale stratiform clouds that developed from earlier convection in the vicinity of the deep convective cells as indicated in the COARE observations. Previous studies have shown that spatially extensive convection is correlated with enhanced mid‐level humidity. To isolate the effect of the mid‐level moist layer on the characteristics of inflow of convective systems, a numerical experiment based on the control simulation of 14 December was conducted. The relative humidity of environment air in the low‐mid troposphere (1.7–6 km layer) was increased to 95%. Trajectory statistics calculated for this sensitivity experiment show increased layer lifting, with a significant amount of air from the upper part of the inflow layer rising in the updraught along with air from just above the surface. Moistening the inflow layer in this sensitivity experiment allows it to saturate more quickly when it encounters the mesoscale cold pool. Once saturated, the relevant static stability is the moist rather than dry static stability, and the whole layer more easily rises over the cold pool. Moistening the inflow layer also modifies the nature of the simulated cold pool itself, which seems to promote layer lifting in the simulation. Possible mechanisms for moistening the mid‐levels are briefly discussed. Copyright © 2002 Royal Meteorological Society