| description abstract | Six different convective?stratiform separation techniques are compared and evaluated using 2D numerical simulations of a tropical and a midlatitude continental squall line. The techniques used include a texture algorithm applied to surface rainfall, a similar algorithm but with additional criteria applied to vertical velocity and cloud, a texture algorithm applied to vertical velocities below the melting layer, a simple approach that assumes a constant characteristic width for the convective region, a more sophisticated texture algorithm applied to radar reflectivities below the melting layer, and a new technique based on the premise that the fall speed of precipitation particles is large relative to air velocity in regions of stratiform precipitation. Comparisons are made in terms of rainfall, mass fluxes, apparent heating and moistening, hydrometeor contents, reflectivity and vertical-velocity contoured-frequency-with-altitude diagrams (CFAD), microphysics, and latent heating retrieval. Overall, it was found that the different separation techniques produced results that qualitatively agreed. However, the quantitative differences were significant. The texture algorithm applied to surface rain consistently produced the most stratiform rain while the texture algorithm applied to radar reflectivities below the melting layer and the new method comparing air velocities to terminal velocities consistently produced the most convective rain. The simple constant-area method performed comparably to the others in this squall line setting. Observational comparisons within the context of the model were unable to identify a superior technique. However, all of the methods were able to generate CFADs that were consistent with observations. Latent heating retrieval was shown to be sensitive to the use of separation technique mainly as a result of differences in the stratiform region. Methods that found very little stratiform rain resulted in exaggerated rain-normalized stratiform heating profiles. | |
