The Dawe and Austin (2011, 2013) direct entrainment and detrainment analysis code was used to calculate entrainment and detrainment rates for clouds within the simulation. The simulation itself was initialized using a non-idealized GATE sounding, with a three dimensional spatial resolution of 50 m and a domain size of 1728 x 1728 x 320. The stored temporal resolution of the data was one minute, over a running simulation time of three hours. The entrainment and detrainment rates were calculated for both the core (positive vertical velocity, buoyancy and condensed water) and cloud (condensed water, without the other criteria) regions. For the purposes of this presentation, calculations using entrainment or detrainment were primarily performed on the core regions.
In order to visualize the high resolution data from the LES, the Python programming language was used to post-process and convert the data into appropriate forms for the VAPOR and Met.3D visualization platforms. Assorted variables (e.g. liquid water specific humidity, tetrahedral core mass entrainment or detrainment, vertical velocity) were animated over convective plume life cycles in a 3D Eulerian perspective to analyze their characteristics over time and look for patterns. In addition to the three dimensional visualizations, cross sections at varying heights were analyzed to determine features such as low-level cold pool generation via precipitation-loaded downdrafts (which can provide impetus for new convective development on gust fronts). Over 200 individual convective plumes from various times during the simulation were then aggregated and statistically analyzed using Python to determine if there were significant correlations between entrainment or detrainment and cloud properties.
We found that contrary to conventional assumptions, the fractional entrainment rate of a cloud core shows only a weak correlation with its radius; while it is true that larger cores appear to entrain relatively less than the smaller ones, the rate at which the entrainment rate changes with increasing cloud area tends to be too small to assert such a relationship. We also looked at the relationship between virtual buoyancy and vertical velocity for individual updrafts over their life cycles. Plotting the means of these two variables for the core regions against each other showed a nearly linear relationship in the lower levels, becoming increasingly complex in the higher altitudes, especially during the mature stages of the updraft life cycle. This would be expected as vertical velocities or buoyancies change, sometimes erratically, with altitude due to momentum, vertical advection and other physical processes. The mean values are also dependent on the size of the convective plume itself.
In summary, visualizations of high resolution, three dimensional LES variables were used to analyze the life cycles of convective plumes. Aggregates of these plumes were then used to calculate statistics relating to entrainment and detrainment and look for relationships in the interest of better understanding tropical convection and subsequent parameterization in numerical models.
Dawe, J. T., and P. H. Austin, 2011: Mon. Wea. Rev., 139, 444-456.
Dawe, J. T. and Austin, P. H., 2013: Atmos. Chem. Phys., 13, 7795-7811.