Comparing Cirrus Cloud Formation and Evolution Using in Situ Aircraft Observations and a Cloud Resolving Model

Cirrus clouds, covering ~30% of the Earth, play important roles in Earth’s climate and weather. As a major uncertainty in climate models, cirrus clouds’ radiative forcing (cooling or warming) is influenced by both the microphysical properties (such as ice crystal concentration and size) and the larger scale structure (such as horizontal and vertical extent).

Recent studies (Diao et al. 2013; Diao et al. 2014), based on in situ observations with ~200 m horizontal resolution, showed that the initial conditions of cirrus formation - ice supersaturated regions (ISSRs, where ISS is spatially continuous) - occur mostly at horizontal scales around 1 km, in contrast to the ~100 km scales by previous observations (Gierens et al. 2000). Yet it is still unknown whether current cloud resolving models can capture these small-scale ISSR features.

In this work, we compare the observed characteristics of the ice supersaturation (ISS) with an idealized, cloud-resolving simulation of a squall line (Bryan and Morrison, 2012). The model (CM1) was run with 250 m grid spacing using a double-moment microphysics scheme (Morrison et al. 2005). Our comparisons show that the CM1 model has captured the majority of the small-scale ISSRs (~1 km). In addition, the simulated ISSRs are dominated by water vapor horizontal heterogeneities (~90%) as opposed to temperature heterogeneities (~10%). This result is comparable to the observed values of ~88% and ~9%, respectively. However, when comparing the evolution phases of cirrus clouds (clear-sky ISS, nucleation/freezing, growth and sedimentation/sublimation; Diao et al. 2013), the CM1 simulation does not have sufficient amount of ISS in clear-sky and nucleation phases. This disagreement indicates a shortcoming of the idealized model setup. Overall, the observations show more ISS at higher magnitude (up to ~150% of RHi) than CM1 (~up to 130% of RHi). Also the largest ISSRs in the observations are up to ~100 km, compared with those in CM1 of up to ~10 km. These results suggest possible changes for future modeling studies.