Sensitivities of Lagrangian modeling of a cirrus cloud observed above Jungfraujoch to the used trajectory data

Several studies have been performed using the microphysical box model ZOMM (Zurich Optical und Microphysical Model) to better understand cirrus cloud formation as there are still many uncertainties in cirrus cloud modeling. The box model ZOMM requires trajectories of temperature, pressure and humidity as input. The trajectories studied are calculated backwards from the observation using different weather models such as COSMO-forecast, COSMO-analysis and ECMWF-Reanalysis. Superposing small-scale temperature fluctuations was necessary to represent processes occurring on spatial and temporal scales not resolved in the underlying weather model. Only trajectories with small scale temperature fluctuations are able to reproduce the cirrus measurements. The question now arises, how large the influence of the chosen input trajectories is on the simulated cirrus clouds and if this effect may have the same order of magnitude as adding small-scale temperature fluctuations.

In this study we used offline and online trajectories from COSMO2-forecast and analysis. The offline trajectories are based on meteorological data fields available every hour, while the online trajectories are based on wind fields every model time step. Due to the much higher temporal resolution of online trajectories, we expect that the online trajectories are better able to resolve small scale and rapid air movements present in the atmosphere than offline trajectories do.

In the present study, we investigate the persistent cirrus clouds at altitude between 10 and 11.5 km asl observed by our lidar on 22th of November 2011, which were present for more than 20 hours. Focusing on a persistent cloud in stationary large-scale flow enhances the probability to hit Jungfraujoch with the online-trajectories as these trajectories cannot be calculated backwards. With the performed COSMO simulations we got a reasonable coverage of about 50 online trajectories between 9 and 12 km asl for a fixed point in time.

The offline and online trajectories shows significant difference both in wind directions as well as in humidity. Simulating cirrus cloud formation using ZOMM, we find that the model results are very sensitive to the input trajectory path and the initial humidity. Both quantities seem at least as important as superposing small-scale temperature fluctuations to the chosen trajectories for simulating the observed cirrus cloud at the correct location and time.