a Modelling and Observation Study of Convective Interaction with Large-Scale Dynamics in the Tropics.

Synoptic and planetary scale waves in the tropical atmosphere are observed to modulate deep moist convection, and theorists have long attempted to explain this interaction. Two of the foremost theories developed, wave-CISK and evaporative wind feedback (EWF), have been previously tested using numerical models in which deep convection is parameterized, but the parameterizations did not directly account for modification of the boundary layer by convective downdrafts of low theta_{e} air. The wave-CISK and EWF instabilities are examined in this study using a two dimensional (x-p), hydrostatic, nonrotational model linearized about a basic state in radiative -convective equilibrium, with a convective parameterization by Emanuel (1991) that incorporates the effects of evaporatively driven unsaturated downdrafts. All wave-CISK modes are stable in this model, but long wavelength EWF modes are unstable. The unsaturated downdrafts are shown to have a major damping effect on both short wave EWF and wave -CISK modes. The model's zonal wavenumber 1 and 2 EWF modes have phase speeds 3-4 times larger than that observed for the Madden and Julian Oscillation (MJO). Conceptual studies of EWF modes have parameterized convection by assuming the atmosphere is in strict quasiequilibrium (SQE), and have obtained more realistic phase speeds for the MJO. In SQE, boundary layer theta_{e } and tropospheric temperature changes remain locked together so that CAPE remains constant. It is shown that at long wavelengths, the Emanuel convective parameterization behaves qualitatively in accordance with SQE, and that the quantitative deviations from SQE are partly responsible for the excessively high phase speeds. The validity of the SQE hypothesis is tested on long time and space scales using 11 years of COADS and MSU data, and on time scales of a day to a month using TOGA-COARE sounding data in conjunction with MSU and SSMI precipitation data. It is shown that on the time and space scales studied, SQE is qualitatively valid in that theta_{eb} and < T> are generally positively correlated. However, changes in theta_ {eb} are too large relative to changes in (T) to maintain constant CAPE as predicted by SQE.