Recent studies using ~200 m in-situ observations showed that regions of ice supersaturation are mostly around 1 km in horizontal scale (Diao et al. 2013; Diao et al. 2014). In addition, ice supersaturation is found to be highly correlated with high water vapor spatial heterogeneities rather than low temperature heterogeneities. It is still unclear if such observed characteristics of ice supersaturation can be represented by model simulations at various spatial resolutions.
In this work, we compare the simulated characteristics of ice supersaturated regions (ISSRs) with the aircraft-based observations from the Deep Convective Clouds & Chemistry Experiment (DC3) campaign in May June 2012. The CM1 model (Bryan and Morrison, 2012) was run at 250 m and 4 km horizontal grid spacing for a squall line scenario with a double-moment microphysics scheme (Morrison et al. 2009). Our comparisons show that the 250 m simulation captures the majority of the small-scale ISSRs with sizes around 1 km. The length distribution of ISSRs in the 250 m run has a much faster decay (i.e., many more smaller ISSRs than larger ones) than the 4 km run. In addition, the 250 m run shows that water vapor horizontal heterogeneities are the dominant (~91-97%) contributor to the high RHi values as opposed to temperature heterogeneities (~1-2%). This result is comparable to the observed values of ~88% and ~9%, respectively. However, at coarser resolution, the 4 km-run water vapor heterogeneities have smaller contributions (~50-70%) to ISSRs while temperature heterogeneities have comparatively larger contributions (~30-50%) to the formation of ISSRs. In addition, the average RHi values inside the ISSRs are about 10% higher in the 250 m run than those in the 4 km run. When comparing the total spatial coverage of ice supersaturation, the 4 km run has ~1.5 times higher ice supersaturation coverage than the 250 m run. Future investigations are needed to address the reason behind the higher spatial coverage of ice supersaturation in the coarser resolution run. Overall, our results suggest that the simulated characteristics of ice supersaturation are very sensitive to the model resolution, which points to the need of future work on the sensitivity of ice cloud formation to model spatial resolution in weather forecasting models.