Camino Muysca hacia el ritual de agradecimiento.

In the following, the global occurrence of relative humidity with respect to ice is discussed. In order to understand the distribution of ice supersaturation in the atmosphere the global model data from the GME and GMEice is divided for the tropics, mid-latitudes and polar regions.

The relative frequencies with a further differentiation between cloud free and all model output points are plotted for the GME (Fig. 5.15a) and the GMEice (Fig. 5.15b). In Fig. 5.16 it is

additionally distinguished between the different temperature regimes case a: 220<T<243 K and case b: T _≤220 K. These different cases are then compared to a distribution law for the relative humidity in the upper troposphere/ lower stratosphere based on Gierens et al. (1999). The exponential law is derived from three years of data from the MOZAIC airborne program (Marenco et al., 1998).

First, the results of Fig. 5.15a and Fig. 5.15b for the GME and the GMEice are discussed,

respectively. The relative frequency distribution for the GME in Fig. 5.15a is shown for all model points and only the cloud free points, where the mixing ratios for cloud ice and cloud water are equal to zero. In the GME plots (Fig. 5.15) it can be seen that relative humidity w.r.t.

a) GME

b) GMEice

Figure 5.15: Frequency of occurrence of relative humidity w.r.t. ice in the tropics, mid-latitudes and polar regions distinguishing between cloud free and all points. For this the GME a) and the GMEiceb) model data for July 2011 are compared to MOZAIC data.

ice frequencies denoted by a reference line at RHi=100 % only has a peak when considering all points. The mentioned peak will be removed, once the cloud clearing is performed. This shows that for the operational ice microphysics scheme a lot of the in-cloud relative humidity is equal to ice saturation. The ice supersaturation distribution exponentially decreases with increasing humidities. When looking at the different geographic regions, the distribution for all points in Fig. 5.15a shows that most of the ice supersaturation occurs in the polar regions and the least in the tropics. This looks similar when looking at the cloud free points for the polar and mid-latitudes, which are slightly reduced. However, the frequency of ice supersaturation in the tropics is small, which indicates that most of the ice supersaturation in the tropics exist within clouds in the GME. The high relative frequencies described by the MOZAIC exponential law are not reached even though high supersaturations are present.

The RHi distribution for the GMEice Fig. 5.15b looks very different for the comparison be-

tween all points and only the cloud free points. The distribution considering all points is almost parallel to the MOZAIC distribution until reaching 120 %. After that the relative frequency drops steeply and bulging at RHi=140 %. Thus a break in the distribution can be observed for the GMEice. This might be a consequence of the two nucleation mechanism being triggered

at these humidities. This is further discussed later in combination with the temperature depen- dencies. When regarding the relative frequencies in dependency of the tropics, mid-latitudes

and polar regions in Fig. 5.15b, the differences are not that obvious. A main difference to the GME relative frequency of RHi lies in the location of the maximum, which is near a relative frequency of 120 % for the polar region and is shifted further to saturation in the tropics. When looking at the cloud free scenario in Fig. 5.15b it is quite surprising that the ice supersatura- tion values are reduced so severely. Responsible for this might be the process of sedimentation as it plays a crucial role in maintaining in-cloud ice supersaturation (Spichtinger and Gierens, 2009a). This finding of in-cloud supersaturation is also in accordance with Spichtinger and Gierens (2004) who investigated this phenomenon through statistical evaluations of MOZAIC and INCA measurements. They argue that the shape of the humidity distribution differ in de- pendency of the maturity of the cloud. The cirrus clouds warmer than₋40 °C are considered to be more mature and have a more symmetric distribution, while it is positively skewed in colder cirrus (Spichtinger and Gierens, 2004). Thus it is important to investigate the different temperature regimes in which the high relative humidities occur.

The temperature dependency of the ice supersaturation frequency is important to investigate due to the nucleation regimes and is depicted in Fig. 5.16. Heterogeneous nucleation occurs at temperatures above_∼220 K and RHi around_∼120 % while homogeneous freezing of liquid aerosols happens at colder temperatures and higher ice supersaturations. Thus differences in the

RHi distribution are expected in relation to the different nucleation regimes. The temperature

ranges that are considered for the GME (Fig. 5.16a) and the GMEice (Fig. 5.16b) are 220<

T<243 K and T_≤220 K, where heterogeneous and homogeneous nucleation are active, if sufficient ice nuclei and ice supersaturation are existent.

Looking at Fig. 5.16a for the GME model output reveals that the relative humidities over ice peak at ice saturation for both temperature intervals. Case a shows maximum relative humidi- ties with respect to ice of 110 %, whereas the colder case b includes the higher values. It is interesting to see that the ice supersaturation seems to be quite insensitive towards the temper- ature in the tropical region. In the other regions the RHi values are much higher which would be expected in colder regions.

When regarding the relative frequency plots in Fig. 5.16b the observed behaviour for case a is analogue to the findings in Lamquin et al. (2012) for relative humidities over ice for T<

243 K. For very cold temperatures below 200 K with high level cirrus there is a shift in the RHi peak. This peak at 120% for high clouds can also be seen in Fig. 4 in Lamquin et al. (2012), where the RHi field is plotted for the northern mid-latitudes MOZAIC data. This bulge in the

RHi distribution in the MOZAIC data at 120% for high cirrus is also discussed in Spichtinger

and Gierens (2004). Therein it is also argued that for thin high cirrus clouds the depletion of supersaturation through depositional growth is very slow. This behaviour is also observable in the relative frequency plot Fig. 5.16b with the new microphysical scheme including the altered depositional growth.

a) GME

b) GMEice

Figure 5.16: Frequency of occurrence of relative humidity w.r.t. ice in the tropics, mid-latitudes and polar regions as in Fig. 5.15 but now with distinguishing between the temperature regions 220<T<243 K and T_≤220 K. The GME (top) and the GMEice(bottom) model data for July 2011 are compared to MOZAIC data.