Very recent work shows that a third significant contributor to reflectivity biases is likely the parameterization of mixed phase microphysical processes. For a given updraft vertical velocity and condensate content, aircraft observations during the High Altitude Ice Crystals - High Ice Water Content (HAIC-HIWC) campaign in Darwin, Australia during January-February 2014 show that not a single simulated updraft can reproduce observed low reflectivities in updraft cores sampled at -12°C. Despite significant ice multiplication rates through rime splintering in simulated updrafts, many of these ice splinters end up intersecting raindrops being lofted, leading to graupel production. Insufficient ice number concentrations and large ice sizes in simulations result in vapor deposition rates that are incapable of depleting supersaturations over liquid in updrafts. This process results in the formation of cloud droplets that cause further riming and graupel formation at the expense of snow mass in the model. Based on recent field campaign images of particles between -5 and -15°C in tropical maritime updrafts, microphysics parameterizations may be missing representation of an important mechanism, such as raindrop fracturing or large cloud droplet splintering/ejection, that enhances ice crystal concentrations and limits concentrations of moderate to large graupel.
Differences in observed and simulated ice properties impact ice sedimentation and detrainment rates from convective cores, which go on to impact precipitation efficiency, distribution of rain rates, and likely the life cycle of the entire convective system. Decreasing model grid spacing to 333 m does not eliminate any of the three mentioned contributors to convective precipitation biases. Furthermore, it is likely that feedbacks between biased dynamics and microphysics at the convective scale and mesoscale in simulations also contribute to tropical convective system precipitation biases. To reduce these biases without unrealistically tuning a model requires further laboratory and field experiment observations to collectively constrain the aforementioned factors that cause consistent model biases.