Abstract
A three-dimensional thermodynamic retrieval method has been developed and tested for application to deep convective clouds. To test the accuracy of the method and for sensitivity studies, output from a three-dimensional numerical cloud model has been utilized in place of observations. Input to the method are wind component and liquid water fields and basic output variables within the same volume are the deviation of potential temperature and perturbed pressure from their respective horizontal averages. The derivation of the retrieval equations from the momentum equations and the programming of these equations is shown to be correct by comparison of the retrieved fields with output from the numerical model. Other cases test the sensitivity of the retrieved result to inadequacies potentially present in observed (Doppler radar) wind and water fields. Tests are carried out examining the problem of time resolution in the observed data, possible inadequacies in observation and parameterization of turbulence, and accuracy in the measurement of liquid water fields. In other experiments velocity perturbations are added to the input velocity fields to simulate very crudely the errors which might be present on various scales in observed fields, and the sensitivity of the retrieved fields to these errors is assessed. Filtering the result is shown to be effective if the predominant scale within which input errors occur is known.
