Sensitivity of convectively driven tropical tropopause cirrus to ice habit

Abstract. Cirrus clouds that form in the tropical tropopause layer (TTL) can play a key role in vertical transport through the upper troposphere and lower stratosphere, which can significantly impact the radiative energy budget and stratospheric chemistry. The main scientific objectives of this work are to (a) investigate the influence of predefined versus non-predefined categories of ice hydrometeors and ice habit on convectively driven TTL cirrus in simulations of the Asian summer monsoon, and (b) determine the optimal model configuration to adequately simulate TTL cirrus clouds. During the StratoClim field campaign in 2017, aircraft measurements from the southern slopes of the Himalayas captured high ice water content (HIWC) up to 2400 ppmv and ice particle aggregates exceeding 700 µm in size with unusually long residence times. The observed ice particles were mainly of liquid origin, with a small amount formed in situ. The corresponding profile of IWC from the ERA5 reanalysis corroborates the presence of HIWC detrained from deep convective plumes in the TTL but underestimates HIWC by an order of magnitude. We compare the aircraft measurements to a set of numerical experiments at the scales of large-eddy simulations (LES) of the same case study that employ different microphysics schemes. In the TTL, only the scheme that predicts ice habits can reproduce the observed HIWC, ice number concentration, and bimodal ice particle size distribution. The lower range of particle sizes is mostly represented by planar and columnar habits, while the upper range is dominated by aggregates. Large aggregates with sizes between 600 and 800 μm have fall speeds of less than 20 cm s-1, which explains the long residence time of the aggregates in the TTL. Planar ice particles of liquid origin contribute substantially to HIWC. The columnar and aggregates habits are in the in-situ range with lower IWC and number concentration. For all habits, the ice number concentration increases with decreasing temperature. For the planar ice habit, relative humidity is inversely correlated with fall speed. This correlation is less evident for the other two ice habits. In the lower range of supersaturation with respect to ice, columnar habit has the highest fall speed. The difference in ice number concentration across habits can be up to four orders of magnitude, and aggregates occurring in much smaller numbers. We demonstrate and quantify the linear relationship between the differential sedimentation of pristine ice crystals and the size of the aggregates that form when pristine crystals collide. The slope of this relationship depends on which pristine ice habit sediments faster. Each simulated ice habit is associated with distinct radiative and latent heating rates. This study suggests that a model configuration nested down to LES scales with a microphysical parameterization that predicts ice shape evolution is crucial to provide an accurate representation of the microphysical properties of TTL cirrus, and thus the associated (de)hydration process.