A long-standing question regarding tropical cumuli is why, with cloud tops no colder than say -12 °C, they are sometimes observed to produce hundreds of ice crystals per liter, despite their pristine environment (where orders of magnitude fewer ice nuclei are detected and expected). Past studies have implicated faulty ice crystal measurements, undiscovered sources of ice nuclei over the oceans, and/or unknown ice multiplication mechanisms.

The Ice in Clouds Experiment -Tropical (ICE-T) field campaign was held in July 2011 in the Caribbean to investigate this issue further. The microphysical evolution of cumulus congestus was documented using observations from the NSF/NCAR C130 aircraft, outfitted with an array of microphysical probes, ice nuclei instrumentation, and a cloud radar. When focusing upon data collected near the ascending cloud tops, the observations document a strong warm rain process below 0°C, an early and active Hallet-Mossop (rime-splintering) ice multiplication process, and occasionally, ice crystal number concentrations exceeding 100 L-1, when ice nuclei estimates were far less.

High-resolution simulations of a single tropical cumulus, meant to represent those observed during ICE-T, show the importance of the early rain formation to ice processes later. Results show that the copious amounts of supercooled cloud and raindrops produced at warmer levels initiate graupel sooner and enhance rime-splintering, which in turn can help to explain an early burst of ice during the developing stage of the cloud, up to 50 L-1. The model cannot replicate ice crystal number concentrations in excess of 100 L-1, however, without adding far more ice nuclei (than observed) active at temperatures less than -10°C. The talk will end with a discussion of some processes not presently included in the model that may help to direct future efforts in this regard.